Xabsurd commited on
Commit
08d4d84
·
verified ·
1 Parent(s): 2a96160

Release V1 with updated nodes, comparisons and workflows

Browse files

Publish the final V1 control adapter, current MOSS-LLLite nodes, English model card, concise development report, and four verified V1 examples with embedded/UI/API workflows. Preserve beta2 and beta3 weights.

This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +1 -0
  2. workflow-and-nodes.zip → ComfyUI-MOSS-LLLite.zip +2 -2
  3. ComfyUI-MOSS-LLLite/.gitignore +8 -0
  4. ComfyUI-MOSS-LLLite/LICENSE +674 -0
  5. ComfyUI-MOSS-LLLite/NOTICE +15 -0
  6. ComfyUI-MOSS-LLLite/README.md +91 -0
  7. {ComfyUI-Moss-Tile-Control → ComfyUI-MOSS-LLLite}/__init__.py +4 -3
  8. ComfyUI-MOSS-LLLite/lllite_model.py +121 -0
  9. ComfyUI-MOSS-LLLite/lllite_nodes.py +137 -0
  10. {ComfyUI-Moss-Tile-Control → ComfyUI-MOSS-LLLite}/model.py +0 -0
  11. {ComfyUI-Moss-Tile-Control → ComfyUI-MOSS-LLLite}/nodes.py +0 -0
  12. ComfyUI-MOSS-LLLite/pyproject.toml +11 -0
  13. ComfyUI-MOSS-LLLite/tests/test_lllite_nodes.py +187 -0
  14. ComfyUI-MOSS-LLLite/tests/test_nodes.py +191 -0
  15. ComfyUI-Moss-Tile-Control/README.md +0 -30
  16. README.md +204 -153
  17. RESEARCH_REPORT.md +49 -0
  18. SHA256SUMS.txt +5 -0
  19. V1/MOSS-LLLite-Tile-V1.safetensors +3 -0
  20. examples/beta3-epoch2_1.png +3 -0
  21. examples/beta3-epoch2_2.png +3 -0
  22. examples/input1.png +3 -0
  23. examples/input2.png +3 -0
  24. examples/v1/README.md +50 -0
  25. examples/v1/SHA256SUMS.txt +22 -0
  26. examples/v1/cat-photo-control.png +3 -0
  27. examples/v1/cat-photo-input.png +3 -0
  28. examples/v1/cat-photo-v1.png +3 -0
  29. examples/v1/cat-photo.api.json +205 -0
  30. examples/v1/cat-photo.workflow.json +1027 -0
  31. examples/v1/examples.json +109 -0
  32. examples/v1/flying-cat-control.png +3 -0
  33. examples/v1/flying-cat-input.png +3 -0
  34. examples/v1/flying-cat-v1.png +3 -0
  35. examples/v1/flying-cat.api.json +194 -0
  36. examples/v1/flying-cat.workflow.json +975 -0
  37. examples/v1/grass-control.png +3 -0
  38. examples/v1/grass-input.png +3 -0
  39. examples/v1/grass-v1.png +3 -0
  40. examples/v1/grass.api.json +205 -0
  41. examples/v1/grass.workflow.json +1027 -0
  42. examples/v1/jetpack-cat-control.png +3 -0
  43. examples/v1/jetpack-cat-input.png +3 -0
  44. examples/v1/jetpack-cat-v1.png +3 -0
  45. examples/v1/jetpack-cat.api.json +194 -0
  46. examples/v1/jetpack-cat.workflow.json +975 -0
  47. release-manifest.json +36 -0
  48. workflow/moss_image_tile-beta3-color-match.json +1739 -0
  49. workflow/moss_image_tile-beta3.json +1666 -0
  50. workflow/moss_image_tile-test-beta2-color-match.json +1721 -0
.gitattributes CHANGED
@@ -37,3 +37,4 @@ examples/beta2-comparison.png filter=lfs diff=lfs merge=lfs -text
37
  examples/beta2-comparison2.png filter=lfs diff=lfs merge=lfs -text
38
  examples/beta3-epoch2-comparison.png filter=lfs diff=lfs merge=lfs -text
39
  examples/beta3-epoch2-comparison2.png filter=lfs diff=lfs merge=lfs -text
 
 
37
  examples/beta2-comparison2.png filter=lfs diff=lfs merge=lfs -text
38
  examples/beta3-epoch2-comparison.png filter=lfs diff=lfs merge=lfs -text
39
  examples/beta3-epoch2-comparison2.png filter=lfs diff=lfs merge=lfs -text
40
+ *.png filter=lfs diff=lfs merge=lfs -text
workflow-and-nodes.zip → ComfyUI-MOSS-LLLite.zip RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bef6f71f5bba0c460b45833f98c848d48f4071b6d1410dfd5987a3aac1efae63
3
- size 50869
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7c2fa276e04b576ba2bb46c114ea8aef6f86fa7435fc300864d391b95d7a3999
3
+ size 30244
ComfyUI-MOSS-LLLite/.gitignore ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ __pycache__/
2
+ *.py[cod]
3
+ .pytest_cache/
4
+ .venv/
5
+ *.safetensors
6
+ *.pt
7
+ *.pth
8
+ .DS_Store
ComfyUI-MOSS-LLLite/LICENSE ADDED
@@ -0,0 +1,674 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ GNU GENERAL PUBLIC LICENSE
2
+ Version 3, 29 June 2007
3
+
4
+ Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
5
+ Everyone is permitted to copy and distribute verbatim copies
6
+ of this license document, but changing it is not allowed.
7
+
8
+ Preamble
9
+
10
+ The GNU General Public License is a free, copyleft license for
11
+ software and other kinds of works.
12
+
13
+ The licenses for most software and other practical works are designed
14
+ to take away your freedom to share and change the works. By contrast,
15
+ the GNU General Public License is intended to guarantee your freedom to
16
+ share and change all versions of a program--to make sure it remains free
17
+ software for all its users. We, the Free Software Foundation, use the
18
+ GNU General Public License for most of our software; it applies also to
19
+ any other work released this way by its authors. You can apply it to
20
+ your programs, too.
21
+
22
+ When we speak of free software, we are referring to freedom, not
23
+ price. Our General Public Licenses are designed to make sure that you
24
+ have the freedom to distribute copies of free software (and charge for
25
+ them if you wish), that you receive source code or can get it if you
26
+ want it, that you can change the software or use pieces of it in new
27
+ free programs, and that you know you can do these things.
28
+
29
+ To protect your rights, we need to prevent others from denying you
30
+ these rights or asking you to surrender the rights. Therefore, you have
31
+ certain responsibilities if you distribute copies of the software, or if
32
+ you modify it: responsibilities to respect the freedom of others.
33
+
34
+ For example, if you distribute copies of such a program, whether
35
+ gratis or for a fee, you must pass on to the recipients the same
36
+ freedoms that you received. You must make sure that they, too, receive
37
+ or can get the source code. And you must show them these terms so they
38
+ know their rights.
39
+
40
+ Developers that use the GNU GPL protect your rights with two steps:
41
+ (1) assert copyright on the software, and (2) offer you this License
42
+ giving you legal permission to copy, distribute and/or modify it.
43
+
44
+ For the developers' and authors' protection, the GPL clearly explains
45
+ that there is no warranty for this free software. For both users' and
46
+ authors' sake, the GPL requires that modified versions be marked as
47
+ changed, so that their problems will not be attributed erroneously to
48
+ authors of previous versions.
49
+
50
+ Some devices are designed to deny users access to install or run
51
+ modified versions of the software inside them, although the manufacturer
52
+ can do so. This is fundamentally incompatible with the aim of
53
+ protecting users' freedom to change the software. The systematic
54
+ pattern of such abuse occurs in the area of products for individuals to
55
+ use, which is precisely where it is most unacceptable. Therefore, we
56
+ have designed this version of the GPL to prohibit the practice for those
57
+ products. If such problems arise substantially in other domains, we
58
+ stand ready to extend this provision to those domains in future versions
59
+ of the GPL, as needed to protect the freedom of users.
60
+
61
+ Finally, every program is threatened constantly by software patents.
62
+ States should not allow patents to restrict development and use of
63
+ software on general-purpose computers, but in those that do, we wish to
64
+ avoid the special danger that patents applied to a free program could
65
+ make it effectively proprietary. To prevent this, the GPL assures that
66
+ patents cannot be used to render the program non-free.
67
+
68
+ The precise terms and conditions for copying, distribution and
69
+ modification follow.
70
+
71
+ TERMS AND CONDITIONS
72
+
73
+ 0. Definitions.
74
+
75
+ "This License" refers to version 3 of the GNU General Public License.
76
+
77
+ "Copyright" also means copyright-like laws that apply to other kinds of
78
+ works, such as semiconductor masks.
79
+
80
+ "The Program" refers to any copyrightable work licensed under this
81
+ License. Each licensee is addressed as "you". "Licensees" and
82
+ "recipients" may be individuals or organizations.
83
+
84
+ To "modify" a work means to copy from or adapt all or part of the work
85
+ in a fashion requiring copyright permission, other than the making of an
86
+ exact copy. The resulting work is called a "modified version" of the
87
+ earlier work or a work "based on" the earlier work.
88
+
89
+ A "covered work" means either the unmodified Program or a work based
90
+ on the Program.
91
+
92
+ To "propagate" a work means to do anything with it that, without
93
+ permission, would make you directly or secondarily liable for
94
+ infringement under applicable copyright law, except executing it on a
95
+ computer or modifying a private copy. Propagation includes copying,
96
+ distribution (with or without modification), making available to the
97
+ public, and in some countries other activities as well.
98
+
99
+ To "convey" a work means any kind of propagation that enables other
100
+ parties to make or receive copies. Mere interaction with a user through
101
+ a computer network, with no transfer of a copy, is not conveying.
102
+
103
+ An interactive user interface displays "Appropriate Legal Notices"
104
+ to the extent that it includes a convenient and prominently visible
105
+ feature that (1) displays an appropriate copyright notice, and (2)
106
+ tells the user that there is no warranty for the work (except to the
107
+ extent that warranties are provided), that licensees may convey the
108
+ work under this License, and how to view a copy of this License. If
109
+ the interface presents a list of user commands or options, such as a
110
+ menu, a prominent item in the list meets this criterion.
111
+
112
+ 1. Source Code.
113
+
114
+ The "source code" for a work means the preferred form of the work
115
+ for making modifications to it. "Object code" means any non-source
116
+ form of a work.
117
+
118
+ A "Standard Interface" means an interface that either is an official
119
+ standard defined by a recognized standards body, or, in the case of
120
+ interfaces specified for a particular programming language, one that
121
+ is widely used among developers working in that language.
122
+
123
+ The "System Libraries" of an executable work include anything, other
124
+ than the work as a whole, that (a) is included in the normal form of
125
+ packaging a Major Component, but which is not part of that Major
126
+ Component, and (b) serves only to enable use of the work with that
127
+ Major Component, or to implement a Standard Interface for which an
128
+ implementation is available to the public in source code form. A
129
+ "Major Component", in this context, means a major essential component
130
+ (kernel, window system, and so on) of the specific operating system
131
+ (if any) on which the executable work runs, or a compiler used to
132
+ produce the work, or an object code interpreter used to run it.
133
+
134
+ The "Corresponding Source" for a work in object code form means all
135
+ the source code needed to generate, install, and (for an executable
136
+ work) run the object code and to modify the work, including scripts to
137
+ control those activities. However, it does not include the work's
138
+ System Libraries, or general-purpose tools or generally available free
139
+ programs which are used unmodified in performing those activities but
140
+ which are not part of the work. For example, Corresponding Source
141
+ includes interface definition files associated with source files for
142
+ the work, and the source code for shared libraries and dynamically
143
+ linked subprograms that the work is specifically designed to require,
144
+ such as by intimate data communication or control flow between those
145
+ subprograms and other parts of the work.
146
+
147
+ The Corresponding Source need not include anything that users
148
+ can regenerate automatically from other parts of the Corresponding
149
+ Source.
150
+
151
+ The Corresponding Source for a work in source code form is that
152
+ same work.
153
+
154
+ 2. Basic Permissions.
155
+
156
+ All rights granted under this License are granted for the term of
157
+ copyright on the Program, and are irrevocable provided the stated
158
+ conditions are met. This License explicitly affirms your unlimited
159
+ permission to run the unmodified Program. The output from running a
160
+ covered work is covered by this License only if the output, given its
161
+ content, constitutes a covered work. This License acknowledges your
162
+ rights of fair use or other equivalent, as provided by copyright law.
163
+
164
+ You may make, run and propagate covered works that you do not
165
+ convey, without conditions so long as your license otherwise remains
166
+ in force. You may convey covered works to others for the sole purpose
167
+ of having them make modifications exclusively for you, or provide you
168
+ with facilities for running those works, provided that you comply with
169
+ the terms of this License in conveying all material for which you do
170
+ not control copyright. Those thus making or running the covered works
171
+ for you must do so exclusively on your behalf, under your direction
172
+ and control, on terms that prohibit them from making any copies of
173
+ your copyrighted material outside their relationship with you.
174
+
175
+ Conveying under any other circumstances is permitted solely under
176
+ the conditions stated below. Sublicensing is not allowed; section 10
177
+ makes it unnecessary.
178
+
179
+ 3. Protecting Users' Legal Rights From Anti-Circumvention Law.
180
+
181
+ No covered work shall be deemed part of an effective technological
182
+ measure under any applicable law fulfilling obligations under article
183
+ 11 of the WIPO copyright treaty adopted on 20 December 1996, or
184
+ similar laws prohibiting or restricting circumvention of such
185
+ measures.
186
+
187
+ When you convey a covered work, you waive any legal power to forbid
188
+ circumvention of technological measures to the extent such circumvention
189
+ is effected by exercising rights under this License with respect to
190
+ the covered work, and you disclaim any intention to limit operation or
191
+ modification of the work as a means of enforcing, against the work's
192
+ users, your or third parties' legal rights to forbid circumvention of
193
+ technological measures.
194
+
195
+ 4. Conveying Verbatim Copies.
196
+
197
+ You may convey verbatim copies of the Program's source code as you
198
+ receive it, in any medium, provided that you conspicuously and
199
+ appropriately publish on each copy an appropriate copyright notice;
200
+ keep intact all notices stating that this License and any
201
+ non-permissive terms added in accord with section 7 apply to the code;
202
+ keep intact all notices of the absence of any warranty; and give all
203
+ recipients a copy of this License along with the Program.
204
+
205
+ You may charge any price or no price for each copy that you convey,
206
+ and you may offer support or warranty protection for a fee.
207
+
208
+ 5. Conveying Modified Source Versions.
209
+
210
+ You may convey a work based on the Program, or the modifications to
211
+ produce it from the Program, in the form of source code under the
212
+ terms of section 4, provided that you also meet all of these conditions:
213
+
214
+ a) The work must carry prominent notices stating that you modified
215
+ it, and giving a relevant date.
216
+
217
+ b) The work must carry prominent notices stating that it is
218
+ released under this License and any conditions added under section
219
+ 7. This requirement modifies the requirement in section 4 to
220
+ "keep intact all notices".
221
+
222
+ c) You must license the entire work, as a whole, under this
223
+ License to anyone who comes into possession of a copy. This
224
+ License will therefore apply, along with any applicable section 7
225
+ additional terms, to the whole of the work, and all its parts,
226
+ regardless of how they are packaged. This License gives no
227
+ permission to license the work in any other way, but it does not
228
+ invalidate such permission if you have separately received it.
229
+
230
+ d) If the work has interactive user interfaces, each must display
231
+ Appropriate Legal Notices; however, if the Program has interactive
232
+ interfaces that do not display Appropriate Legal Notices, your
233
+ work need not make them do so.
234
+
235
+ A compilation of a covered work with other separate and independent
236
+ works, which are not by their nature extensions of the covered work,
237
+ and which are not combined with it such as to form a larger program,
238
+ in or on a volume of a storage or distribution medium, is called an
239
+ "aggregate" if the compilation and its resulting copyright are not
240
+ used to limit the access or legal rights of the compilation's users
241
+ beyond what the individual works permit. Inclusion of a covered work
242
+ in an aggregate does not cause this License to apply to the other
243
+ parts of the aggregate.
244
+
245
+ 6. Conveying Non-Source Forms.
246
+
247
+ You may convey a covered work in object code form under the terms
248
+ of sections 4 and 5, provided that you also convey the
249
+ machine-readable Corresponding Source under the terms of this License,
250
+ in one of these ways:
251
+
252
+ a) Convey the object code in, or embodied in, a physical product
253
+ (including a physical distribution medium), accompanied by the
254
+ Corresponding Source fixed on a durable physical medium
255
+ customarily used for software interchange.
256
+
257
+ b) Convey the object code in, or embodied in, a physical product
258
+ (including a physical distribution medium), accompanied by a
259
+ written offer, valid for at least three years and valid for as
260
+ long as you offer spare parts or customer support for that product
261
+ model, to give anyone who possesses the object code either (1) a
262
+ copy of the Corresponding Source for all the software in the
263
+ product that is covered by this License, on a durable physical
264
+ medium customarily used for software interchange, for a price no
265
+ more than your reasonable cost of physically performing this
266
+ conveying of source, or (2) access to copy the
267
+ Corresponding Source from a network server at no charge.
268
+
269
+ c) Convey individual copies of the object code with a copy of the
270
+ written offer to provide the Corresponding Source. This
271
+ alternative is allowed only occasionally and noncommercially, and
272
+ only if you received the object code with such an offer, in accord
273
+ with subsection 6b.
274
+
275
+ d) Convey the object code by offering access from a designated
276
+ place (gratis or for a charge), and offer equivalent access to the
277
+ Corresponding Source in the same way through the same place at no
278
+ further charge. You need not require recipients to copy the
279
+ Corresponding Source along with the object code. If the place to
280
+ copy the object code is a network server, the Corresponding Source
281
+ may be on a different server (operated by you or a third party)
282
+ that supports equivalent copying facilities, provided you maintain
283
+ clear directions next to the object code saying where to find the
284
+ Corresponding Source. Regardless of what server hosts the
285
+ Corresponding Source, you remain obligated to ensure that it is
286
+ available for as long as needed to satisfy these requirements.
287
+
288
+ e) Convey the object code using peer-to-peer transmission, provided
289
+ you inform other peers where the object code and Corresponding
290
+ Source of the work are being offered to the general public at no
291
+ charge under subsection 6d.
292
+
293
+ A separable portion of the object code, whose source code is excluded
294
+ from the Corresponding Source as a System Library, need not be
295
+ included in conveying the object code work.
296
+
297
+ A "User Product" is either (1) a "consumer product", which means any
298
+ tangible personal property which is normally used for personal, family,
299
+ or household purposes, or (2) anything designed or sold for incorporation
300
+ into a dwelling. In determining whether a product is a consumer product,
301
+ doubtful cases shall be resolved in favor of coverage. For a particular
302
+ product received by a particular user, "normally used" refers to a
303
+ typical or common use of that class of product, regardless of the status
304
+ of the particular user or of the way in which the particular user
305
+ actually uses, or expects or is expected to use, the product. A product
306
+ is a consumer product regardless of whether the product has substantial
307
+ commercial, industrial or non-consumer uses, unless such uses represent
308
+ the only significant mode of use of the product.
309
+
310
+ "Installation Information" for a User Product means any methods,
311
+ procedures, authorization keys, or other information required to install
312
+ and execute modified versions of a covered work in that User Product from
313
+ a modified version of its Corresponding Source. The information must
314
+ suffice to ensure that the continued functioning of the modified object
315
+ code is in no case prevented or interfered with solely because
316
+ modification has been made.
317
+
318
+ If you convey an object code work under this section in, or with, or
319
+ specifically for use in, a User Product, and the conveying occurs as
320
+ part of a transaction in which the right of possession and use of the
321
+ User Product is transferred to the recipient in perpetuity or for a
322
+ fixed term (regardless of how the transaction is characterized), the
323
+ Corresponding Source conveyed under this section must be accompanied
324
+ by the Installation Information. But this requirement does not apply
325
+ if neither you nor any third party retains the ability to install
326
+ modified object code on the User Product (for example, the work has
327
+ been installed in ROM).
328
+
329
+ The requirement to provide Installation Information does not include a
330
+ requirement to continue to provide support service, warranty, or updates
331
+ for a work that has been modified or installed by the recipient, or for
332
+ the User Product in which it has been modified or installed. Access to a
333
+ network may be denied when the modification itself materially and
334
+ adversely affects the operation of the network or violates the rules and
335
+ protocols for communication across the network.
336
+
337
+ Corresponding Source conveyed, and Installation Information provided,
338
+ in accord with this section must be in a format that is publicly
339
+ documented (and with an implementation available to the public in
340
+ source code form), and must require no special password or key for
341
+ unpacking, reading or copying.
342
+
343
+ 7. Additional Terms.
344
+
345
+ "Additional permissions" are terms that supplement the terms of this
346
+ License by making exceptions from one or more of its conditions.
347
+ Additional permissions that are applicable to the entire Program shall
348
+ be treated as though they were included in this License, to the extent
349
+ that they are valid under applicable law. If additional permissions
350
+ apply only to part of the Program, that part may be used separately
351
+ under those permissions, but the entire Program remains governed by
352
+ this License without regard to the additional permissions.
353
+
354
+ When you convey a copy of a covered work, you may at your option
355
+ remove any additional permissions from that copy, or from any part of
356
+ it. (Additional permissions may be written to require their own
357
+ removal in certain cases when you modify the work.) You may place
358
+ additional permissions on material, added by you to a covered work,
359
+ for which you have or can give appropriate copyright permission.
360
+
361
+ Notwithstanding any other provision of this License, for material you
362
+ add to a covered work, you may (if authorized by the copyright holders of
363
+ that material) supplement the terms of this License with terms:
364
+
365
+ a) Disclaiming warranty or limiting liability differently from the
366
+ terms of sections 15 and 16 of this License; or
367
+
368
+ b) Requiring preservation of specified reasonable legal notices or
369
+ author attributions in that material or in the Appropriate Legal
370
+ Notices displayed by works containing it; or
371
+
372
+ c) Prohibiting misrepresentation of the origin of that material, or
373
+ requiring that modified versions of such material be marked in
374
+ reasonable ways as different from the original version; or
375
+
376
+ d) Limiting the use for publicity purposes of names of licensors or
377
+ authors of the material; or
378
+
379
+ e) Declining to grant rights under trademark law for use of some
380
+ trade names, trademarks, or service marks; or
381
+
382
+ f) Requiring indemnification of licensors and authors of that
383
+ material by anyone who conveys the material (or modified versions of
384
+ it) with contractual assumptions of liability to the recipient, for
385
+ any liability that these contractual assumptions directly impose on
386
+ those licensors and authors.
387
+
388
+ All other non-permissive additional terms are considered "further
389
+ restrictions" within the meaning of section 10. If the Program as you
390
+ received it, or any part of it, contains a notice stating that it is
391
+ governed by this License along with a term that is a further
392
+ restriction, you may remove that term. If a license document contains
393
+ a further restriction but permits relicensing or conveying under this
394
+ License, you may add to a covered work material governed by the terms
395
+ of that license document, provided that the further restriction does
396
+ not survive such relicensing or conveying.
397
+
398
+ If you add terms to a covered work in accord with this section, you
399
+ must place, in the relevant source files, a statement of the
400
+ additional terms that apply to those files, or a notice indicating
401
+ where to find the applicable terms.
402
+
403
+ Additional terms, permissive or non-permissive, may be stated in the
404
+ form of a separately written license, or stated as exceptions;
405
+ the above requirements apply either way.
406
+
407
+ 8. Termination.
408
+
409
+ You may not propagate or modify a covered work except as expressly
410
+ provided under this License. Any attempt otherwise to propagate or
411
+ modify it is void, and will automatically terminate your rights under
412
+ this License (including any patent licenses granted under the third
413
+ paragraph of section 11).
414
+
415
+ However, if you cease all violation of this License, then your
416
+ license from a particular copyright holder is reinstated (a)
417
+ provisionally, unless and until the copyright holder explicitly and
418
+ finally terminates your license, and (b) permanently, if the copyright
419
+ holder fails to notify you of the violation by some reasonable means
420
+ prior to 60 days after the cessation.
421
+
422
+ Moreover, your license from a particular copyright holder is
423
+ reinstated permanently if the copyright holder notifies you of the
424
+ violation by some reasonable means, this is the first time you have
425
+ received notice of violation of this License (for any work) from that
426
+ copyright holder, and you cure the violation prior to 30 days after
427
+ your receipt of the notice.
428
+
429
+ Termination of your rights under this section does not terminate the
430
+ licenses of parties who have received copies or rights from you under
431
+ this License. If your rights have been terminated and not permanently
432
+ reinstated, you do not qualify to receive new licenses for the same
433
+ material under section 10.
434
+
435
+ 9. Acceptance Not Required for Having Copies.
436
+
437
+ You are not required to accept this License in order to receive or
438
+ run a copy of the Program. Ancillary propagation of a covered work
439
+ occurring solely as a consequence of using peer-to-peer transmission
440
+ to receive a copy likewise does not require acceptance. However,
441
+ nothing other than this License grants you permission to propagate or
442
+ modify any covered work. These actions infringe copyright if you do
443
+ not accept this License. Therefore, by modifying or propagating a
444
+ covered work, you indicate your acceptance of this License to do so.
445
+
446
+ 10. Automatic Licensing of Downstream Recipients.
447
+
448
+ Each time you convey a covered work, the recipient automatically
449
+ receives a license from the original licensors, to run, modify and
450
+ propagate that work, subject to this License. You are not responsible
451
+ for enforcing compliance by third parties with this License.
452
+
453
+ An "entity transaction" is a transaction transferring control of an
454
+ organization, or substantially all assets of one, or subdividing an
455
+ organization, or merging organizations. If propagation of a covered
456
+ work results from an entity transaction, each party to that
457
+ transaction who receives a copy of the work also receives whatever
458
+ licenses to the work the party's predecessor in interest had or could
459
+ give under the previous paragraph, plus a right to possession of the
460
+ Corresponding Source of the work from the predecessor in interest, if
461
+ the predecessor has it or can get it with reasonable efforts.
462
+
463
+ You may not impose any further restrictions on the exercise of the
464
+ rights granted or affirmed under this License. For example, you may
465
+ not impose a license fee, royalty, or other charge for exercise of
466
+ rights granted under this License, and you may not initiate litigation
467
+ (including a cross-claim or counterclaim in a lawsuit) alleging that
468
+ any patent claim is infringed by making, using, selling, offering for
469
+ sale, or importing the Program or any portion of it.
470
+
471
+ 11. Patents.
472
+
473
+ A "contributor" is a copyright holder who authorizes use under this
474
+ License of the Program or a work on which the Program is based. The
475
+ work thus licensed is called the contributor's "contributor version".
476
+
477
+ A contributor's "essential patent claims" are all patent claims
478
+ owned or controlled by the contributor, whether already acquired or
479
+ hereafter acquired, that would be infringed by some manner, permitted
480
+ by this License, of making, using, or selling its contributor version,
481
+ but do not include claims that would be infringed only as a
482
+ consequence of further modification of the contributor version. For
483
+ purposes of this definition, "control" includes the right to grant
484
+ patent sublicenses in a manner consistent with the requirements of
485
+ this License.
486
+
487
+ Each contributor grants you a non-exclusive, worldwide, royalty-free
488
+ patent license under the contributor's essential patent claims, to
489
+ make, use, sell, offer for sale, import and otherwise run, modify and
490
+ propagate the contents of its contributor version.
491
+
492
+ In the following three paragraphs, a "patent license" is any express
493
+ agreement or commitment, however denominated, not to enforce a patent
494
+ (such as an express permission to practice a patent or covenant not to
495
+ sue for patent infringement). To "grant" such a patent license to a
496
+ party means to make such an agreement or commitment not to enforce a
497
+ patent against the party.
498
+
499
+ If you convey a covered work, knowingly relying on a patent license,
500
+ and the Corresponding Source of the work is not available for anyone
501
+ to copy, free of charge and under the terms of this License, through a
502
+ publicly available network server or other readily accessible means,
503
+ then you must either (1) cause the Corresponding Source to be so
504
+ available, or (2) arrange to deprive yourself of the benefit of the
505
+ patent license for this particular work, or (3) arrange, in a manner
506
+ consistent with the requirements of this License, to extend the patent
507
+ license to downstream recipients. "Knowingly relying" means you have
508
+ actual knowledge that, but for the patent license, your conveying the
509
+ covered work in a country, or your recipient's use of the covered work
510
+ in a country, would infringe one or more identifiable patents in that
511
+ country that you have reason to believe are valid.
512
+
513
+ If, pursuant to or in connection with a single transaction or
514
+ arrangement, you convey, or propagate by procuring conveyance of, a
515
+ covered work, and grant a patent license to some of the parties
516
+ receiving the covered work authorizing them to use, propagate, modify
517
+ or convey a specific copy of the covered work, then the patent license
518
+ you grant is automatically extended to all recipients of the covered
519
+ work and works based on it.
520
+
521
+ A patent license is "discriminatory" if it does not include within
522
+ the scope of its coverage, prohibits the exercise of, or is
523
+ conditioned on the non-exercise of one or more of the rights that are
524
+ specifically granted under this License. You may not convey a covered
525
+ work if you are a party to an arrangement with a third party that is
526
+ in the business of distributing software, under which you make payment
527
+ to the third party based on the extent of your activity of conveying
528
+ the work, and under which the third party grants, to any of the
529
+ parties who would receive the covered work from you, a discriminatory
530
+ patent license (a) in connection with copies of the covered work
531
+ conveyed by you (or copies made from those copies), or (b) primarily
532
+ for and in connection with specific products or compilations that
533
+ contain the covered work, unless you entered into that arrangement,
534
+ or that patent license was granted, prior to 28 March 2007.
535
+
536
+ Nothing in this License shall be construed as excluding or limiting
537
+ any implied license or other defenses to infringement that may
538
+ otherwise be available to you under applicable patent law.
539
+
540
+ 12. No Surrender of Others' Freedom.
541
+
542
+ If conditions are imposed on you (whether by court order, agreement or
543
+ otherwise) that contradict the conditions of this License, they do not
544
+ excuse you from the conditions of this License. If you cannot convey a
545
+ covered work so as to satisfy simultaneously your obligations under this
546
+ License and any other pertinent obligations, then as a consequence you may
547
+ not convey it at all. For example, if you agree to terms that obligate you
548
+ to collect a royalty for further conveying from those to whom you convey
549
+ the Program, the only way you could satisfy both those terms and this
550
+ License would be to refrain entirely from conveying the Program.
551
+
552
+ 13. Use with the GNU Affero General Public License.
553
+
554
+ Notwithstanding any other provision of this License, you have
555
+ permission to link or combine any covered work with a work licensed
556
+ under version 3 of the GNU Affero General Public License into a single
557
+ combined work, and to convey the resulting work. The terms of this
558
+ License will continue to apply to the part which is the covered work,
559
+ but the special requirements of the GNU Affero General Public License,
560
+ section 13, concerning interaction through a network will apply to the
561
+ combination as such.
562
+
563
+ 14. Revised Versions of this License.
564
+
565
+ The Free Software Foundation may publish revised and/or new versions of
566
+ the GNU General Public License from time to time. Such new versions will
567
+ be similar in spirit to the present version, but may differ in detail to
568
+ address new problems or concerns.
569
+
570
+ Each version is given a distinguishing version number. If the
571
+ Program specifies that a certain numbered version of the GNU General
572
+ Public License "or any later version" applies to it, you have the
573
+ option of following the terms and conditions either of that numbered
574
+ version or of any later version published by the Free Software
575
+ Foundation. If the Program does not specify a version number of the
576
+ GNU General Public License, you may choose any version ever published
577
+ by the Free Software Foundation.
578
+
579
+ If the Program specifies that a proxy can decide which future
580
+ versions of the GNU General Public License can be used, that proxy's
581
+ public statement of acceptance of a version permanently authorizes you
582
+ to choose that version for the Program.
583
+
584
+ Later license versions may give you additional or different
585
+ permissions. However, no additional obligations are imposed on any
586
+ author or copyright holder as a result of your choosing to follow a
587
+ later version.
588
+
589
+ 15. Disclaimer of Warranty.
590
+
591
+ THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
592
+ APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
593
+ HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
594
+ OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
595
+ THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
596
+ PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
597
+ IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
598
+ ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
599
+
600
+ 16. Limitation of Liability.
601
+
602
+ IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
603
+ WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
604
+ THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
605
+ GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
606
+ USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
607
+ DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
608
+ PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
609
+ EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
610
+ SUCH DAMAGES.
611
+
612
+ 17. Interpretation of Sections 15 and 16.
613
+
614
+ If the disclaimer of warranty and limitation of liability provided
615
+ above cannot be given local legal effect according to their terms,
616
+ reviewing courts shall apply local law that most closely approximates
617
+ an absolute waiver of all civil liability in connection with the
618
+ Program, unless a warranty or assumption of liability accompanies a
619
+ copy of the Program in return for a fee.
620
+
621
+ END OF TERMS AND CONDITIONS
622
+
623
+ How to Apply These Terms to Your New Programs
624
+
625
+ If you develop a new program, and you want it to be of the greatest
626
+ possible use to the public, the best way to achieve this is to make it
627
+ free software which everyone can redistribute and change under these terms.
628
+
629
+ To do so, attach the following notices to the program. It is safest
630
+ to attach them to the start of each source file to most effectively
631
+ state the exclusion of warranty; and each file should have at least
632
+ the "copyright" line and a pointer to where the full notice is found.
633
+
634
+ <one line to give the program's name and a brief idea of what it does.>
635
+ Copyright (C) <year> <name of author>
636
+
637
+ This program is free software: you can redistribute it and/or modify
638
+ it under the terms of the GNU General Public License as published by
639
+ the Free Software Foundation, either version 3 of the License, or
640
+ (at your option) any later version.
641
+
642
+ This program is distributed in the hope that it will be useful,
643
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
644
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
645
+ GNU General Public License for more details.
646
+
647
+ You should have received a copy of the GNU General Public License
648
+ along with this program. If not, see <https://www.gnu.org/licenses/>.
649
+
650
+ Also add information on how to contact you by electronic and paper mail.
651
+
652
+ If the program does terminal interaction, make it output a short
653
+ notice like this when it starts in an interactive mode:
654
+
655
+ <program> Copyright (C) <year> <name of author>
656
+ This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
657
+ This is free software, and you are welcome to redistribute it
658
+ under certain conditions; type `show c' for details.
659
+
660
+ The hypothetical commands `show w' and `show c' should show the appropriate
661
+ parts of the General Public License. Of course, your program's commands
662
+ might be different; for a GUI interface, you would use an "about box".
663
+
664
+ You should also get your employer (if you work as a programmer) or school,
665
+ if any, to sign a "copyright disclaimer" for the program, if necessary.
666
+ For more information on this, and how to apply and follow the GNU GPL, see
667
+ <https://www.gnu.org/licenses/>.
668
+
669
+ The GNU General Public License does not permit incorporating your program
670
+ into proprietary programs. If your program is a subroutine library, you
671
+ may consider it more useful to permit linking proprietary applications with
672
+ the library. If this is what you want to do, use the GNU Lesser General
673
+ Public License instead of this License. But first, please read
674
+ <https://www.gnu.org/licenses/why-not-lgpl.html>.
ComfyUI-MOSS-LLLite/NOTICE ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ MOSS-LLLite
2
+
3
+ This project adapts Anima LLLite conditioning to ComfyUI's MOSS-Image model.
4
+ It imports AnimaLLLiteConditioning and AnimaLLLiteModule from ComfyUI and adapts
5
+ the native MOSS image-generation block for attention-input control.
6
+
7
+ Upstream project: ComfyUI
8
+ https://github.com/Comfy-Org/ComfyUI
9
+ Anima LLLite: comfy/ldm/anima/lllite.py
10
+ MOSS-Image: comfy/ldm/moss_image/model.py
11
+ Upstream license: GNU General Public License, version 3.
12
+ Upstream copyright notices and licenses remain applicable.
13
+
14
+ This repository contains inference code only. Model weights are distributed
15
+ separately and are subject to their own applicable licenses.
ComfyUI-MOSS-LLLite/README.md ADDED
@@ -0,0 +1,91 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # MOSS-LLLite
2
+
3
+ LLLite control for MOSS-Image in ComfyUI, adapting Anima LLLite's RGB conditioning
4
+ and attention input adapters to MOSS. Control strength is adjustable throughout
5
+ sampling. At strength **0**, the apply node returns the original input model
6
+ without encoding the control image or adding patches.
7
+
8
+ ## Installation
9
+
10
+ Use a recent [ComfyUI](https://github.com/Comfy-Org/ComfyUI) with native MOSS-Image,
11
+ Anima LLLite, and the V3 node API. No additional Python packages are required.
12
+
13
+ From your ComfyUI directory:
14
+
15
+ ```bash
16
+ git clone https://github.com/X-absurd/ComfyUI-MOSS-LLLite.git custom_nodes/ComfyUI-MOSS-LLLite
17
+ ```
18
+
19
+ Restart ComfyUI. If you previously installed `ComfyUI-Moss-Tile-Control`, remove
20
+ that directory before installing this package to avoid duplicate node IDs.
21
+ Existing workflow node IDs are preserved.
22
+
23
+ ## Usage
24
+
25
+ For Tile restoration, the recommended final release is **V1**, the final design
26
+ developed through four iterations. Use **Load MOSS-LLLite** and **Apply MOSS-LLLite** for these
27
+ weights. V1 refines details while preserving subject features, with good color
28
+ retention in the author's visual tests; exact color matching is not guaranteed.
29
+
30
+ 1. Put a **MOSS-LLLite** `.safetensors` checkpoint in
31
+ `ComfyUI/models/model_patches/`. Model weights are provided separately.
32
+ 2. Add **Load MOSS-LLLite** and **Apply MOSS-LLLite**, under the
33
+ **MOSS-LLLite** category.
34
+ 3. Connect the MOSS `MODEL`, loaded controller, and control `IMAGE` to the apply
35
+ node. Connect its output `MODEL` to your sampler.
36
+ 4. Keep your normal text conditioning and VAE connections. The apply node takes
37
+ RGB pixels directly and has **no VAE input**.
38
+
39
+ The control task depends on how the checkpoint was trained. Tile checkpoints
40
+ support image restoration; other tasks require matching MOSS-LLLite checkpoints
41
+ and control images. Renaming the nodes does not add new tasks to existing weights.
42
+
43
+ Start with strength `0.5`–`1.0`. Higher values usually preserve more of the control
44
+ image's structure. `start_percent` and `end_percent` select the portion of
45
+ sampling where control is active; the default is the full interval, `0` to `1`.
46
+ Other LoRAs can remain in the model chain, but their effects still apply when
47
+ this controller's strength is zero.
48
+
49
+ The controller does not merge weights into the backbone. It uses ComfyUI's
50
+ ModelPatcher, additional-model loading, and block replacement interfaces.
51
+
52
+ ## Compatibility and limitations
53
+
54
+ - Supports ordinary MOSS-Image generation. MOSS-Edit reference images and cached
55
+ reference KV inputs are currently unsupported.
56
+ - Use a controller compatible with your MOSS Base/SFT backbone.
57
+ - MOSS-LLLite checkpoints are distinct from Anima LLLite, beta2 concat Tile
58
+ LoRAs, and beta3 residual Tile checkpoints. They are not interchangeable.
59
+ - Remove beta2 concat Tile LoRAs before applying MOSS-LLLite.
60
+ - Fused BF16 and INT8/ConvRot QKV weights use only the rows needed by each
61
+ controlled projection, while retaining ComfyUI weight casting and patches.
62
+ Other quantization formats retain the original full-projection path.
63
+ - Color preservation is experimental. Sampling can still change colors; an
64
+ optional Color Match step can help correct the result.
65
+
66
+ ## Existing beta3 workflows
67
+
68
+ **Load MOSS Tile Control** and **Apply MOSS Tile Control** are retained for
69
+ existing beta3 residual checkpoints. These nodes use the Flux2 VAE input.
70
+ Use the MOSS-LLLite nodes above for new LLLite checkpoints.
71
+
72
+ ## Tests
73
+
74
+ From the ComfyUI directory, using its Python environment:
75
+
76
+ ```bash
77
+ python -m unittest discover -s custom_nodes/ComfyUI-MOSS-LLLite/tests -v
78
+ ```
79
+
80
+ The tests cover strength-zero bypass, control intervals, checkpoint loading,
81
+ native block agreement, fused/unfused projections, and existing beta3 behavior.
82
+ They use small CPU models and do not download model weights.
83
+
84
+ ## Credits and license
85
+
86
+ The conditioning encoder and LLLite adapters reuse the
87
+ [Anima LLLite implementation in ComfyUI](https://github.com/Comfy-Org/ComfyUI/blob/master/comfy/ldm/anima/lllite.py).
88
+ MOSS integration uses ComfyUI's native
89
+ [MOSS-Image implementation](https://github.com/Comfy-Org/ComfyUI/blob/master/comfy/ldm/moss_image/model.py).
90
+ See [NOTICE](NOTICE) for attribution and [LICENSE](LICENSE) for GPL-3.0 terms.
91
+ Model weights have their own applicable licenses.
{ComfyUI-Moss-Tile-Control → ComfyUI-MOSS-LLLite}/__init__.py RENAMED
@@ -1,12 +1,13 @@
1
  from comfy_api.latest import ComfyExtension
2
 
3
  from .nodes import MossTileControlLoader, MossTileControlApply
 
4
 
5
 
6
- class MossTileExtension(ComfyExtension):
7
  async def get_node_list(self):
8
- return [MossTileControlLoader, MossTileControlApply]
9
 
10
 
11
  async def comfy_entrypoint():
12
- return MossTileExtension()
 
1
  from comfy_api.latest import ComfyExtension
2
 
3
  from .nodes import MossTileControlLoader, MossTileControlApply
4
+ from .lllite_nodes import MossLLLiteLoader, MossLLLiteApply
5
 
6
 
7
+ class MossLLLiteExtension(ComfyExtension):
8
  async def get_node_list(self):
9
+ return [MossTileControlLoader, MossTileControlApply, MossLLLiteLoader, MossLLLiteApply]
10
 
11
 
12
  async def comfy_entrypoint():
13
+ return MossLLLiteExtension()
ComfyUI-MOSS-LLLite/lllite_model.py ADDED
@@ -0,0 +1,121 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """Anima LLLite's RGB encoder and Q/K/V-input adapters, adapted to MOSS tokens.
2
+
3
+ Reuses the upstream ComfyUI Anima LLLite modules without changing their math.
4
+ https://github.com/Comfy-Org/ComfyUI/blob/master/comfy/ldm/anima/lllite.py
5
+ """
6
+
7
+ from contextlib import nullcontext
8
+ from dataclasses import replace
9
+
10
+ import torch
11
+ from torch import nn
12
+ import torch.nn.functional as F
13
+ import comfy.ops
14
+ import comfy.ldm.moss_image.model as moss
15
+ from comfy.quant_ops import QuantizedTensor, TensorWiseINT8Layout
16
+
17
+ from comfy.ldm.anima.lllite import AnimaLLLiteConditioning, AnimaLLLiteModule
18
+
19
+ FORMAT = 'moss_lllite'
20
+ VERSION = '1'
21
+ PREFIX = 'moss_lllite.'
22
+ TARGETS = ('q', 'k', 'v')
23
+
24
+
25
+ class MossLLLiteControl(nn.Module):
26
+ def __init__(self, model_dim, num_blocks, rank=64, condition_dim=64,
27
+ condition_embedding_dim=32, condition_resblocks=3,
28
+ aspp_dilations=(1, 2, 4, 8), *, operations=nn, device=None, dtype=None):
29
+ super().__init__()
30
+ self.config = dict(model_dim=model_dim, num_blocks=num_blocks, rank=rank,
31
+ condition_dim=condition_dim, condition_embedding_dim=condition_embedding_dim,
32
+ condition_resblocks=condition_resblocks, aspp_dilations=list(aspp_dilations))
33
+ self.conditioner = AnimaLLLiteConditioning(
34
+ 3, condition_dim, condition_embedding_dim, condition_resblocks, aspp_dilations,
35
+ device=device, dtype=dtype, operations=operations,
36
+ )
37
+ self.blocks = nn.ModuleList([
38
+ nn.ModuleDict({target: AnimaLLLiteModule(
39
+ model_dim, condition_embedding_dim, rank,
40
+ device=device, dtype=dtype, operations=operations,
41
+ ) for target in TARGETS}) for _ in range(num_blocks)
42
+ ])
43
+
44
+ def metadata(self):
45
+ return {'format': FORMAT, 'moss_lllite.version': VERSION,
46
+ **{f'moss_lllite.{key}': ','.join(map(str, value)) if isinstance(value, list) else str(value)
47
+ for key, value in self.config.items()}}
48
+
49
+
50
+ def config_from_metadata(metadata):
51
+ if not metadata or metadata.get('format') != FORMAT or metadata.get('moss_lllite.version') != VERSION:
52
+ raise ValueError('Expected a MOSS LLLite checkpoint; beta2, beta3 and Anima weights are incompatible')
53
+ config = {key: int(metadata[f'moss_lllite.{key}']) for key in (
54
+ 'model_dim', 'num_blocks', 'rank', 'condition_dim', 'condition_embedding_dim', 'condition_resblocks')}
55
+ config['aspp_dilations'] = [int(value) for value in metadata['moss_lllite.aspp_dilations'].split(',') if value]
56
+ return config
57
+
58
+
59
+ def apply_image_control(module, tokens, condition, strength):
60
+ count = condition.shape[1]
61
+ image = module(tokens[:, :count], condition, strength)
62
+ return torch.cat((image, tokens[:, count:]), dim=1)
63
+
64
+
65
+ def project_fused_qkv(projection, inputs):
66
+ if isinstance(projection, comfy.ops.MixedPrecisionOp):
67
+ if (projection.quant_format not in (None, 'int8_tensorwise') or projection._full_precision_mm
68
+ or getattr(projection, 'pre_quant_scale', None) is not None):
69
+ return NotImplemented
70
+ elif type(projection) not in (nn.Linear, comfy.ops.disable_weight_init.Linear, comfy.ops.manual_cast.Linear):
71
+ return NotImplemented
72
+ if isinstance(projection, comfy.ops.CastWeightBiasOp) and projection.comfy_force_forward:
73
+ return NotImplemented
74
+
75
+ comfy.ops.run_every_op()
76
+ context = (comfy.ops.CastBiasWeightContext(projection, inputs[0], offloadable=True,
77
+ compute_dtype=inputs[0].dtype, want_requant=True)
78
+ if isinstance(projection, comfy.ops.CastWeightBiasOp)
79
+ else nullcontext((projection.weight, projection.bias)))
80
+ with context as (weight, bias):
81
+ width = weight.shape[0] // 3
82
+ outputs = []
83
+ for index, value in enumerate(inputs):
84
+ start, end = index * width, (index + 1) * width
85
+ if isinstance(weight, QuantizedTensor):
86
+ params = weight.params
87
+ if not isinstance(params, TensorWiseINT8Layout.Params) or params.transposed:
88
+ return NotImplemented
89
+ # Like ComfyUI's MoE bank rows: preserve per-output scales and ConvRot metadata.
90
+ scale = params.scale if params.scale.numel() == 1 else params.scale[start:end]
91
+ rows = QuantizedTensor(weight._qdata[start:end], weight._layout_cls,
92
+ replace(params, orig_shape=(width, weight.shape[1]), scale=scale))
93
+ else:
94
+ rows = weight[start:end]
95
+ outputs.append(F.linear(value, rows, None if bias is None else bias[start:end]))
96
+ return tuple(outputs)
97
+
98
+
99
+ def project_qkv(projections, fused, tokens, modules, condition, strength=1.0):
100
+ # Same pre-projection location as Anima/Cosmos compute_qkv's attn1_patch.
101
+ inputs = [apply_image_control(modules[target], tokens, condition, strength) for target in TARGETS]
102
+ if fused:
103
+ output = project_fused_qkv(projections['qkv'], inputs)
104
+ if output is not NotImplemented:
105
+ return output
106
+ # Other quantization formats retain their module-owned input processing.
107
+ return tuple(projections['qkv'](value).chunk(3, dim=-1)[index].clone()
108
+ for index, value in enumerate(inputs))
109
+ return tuple(projections[target](value) for target, value in zip(TARGETS, inputs))
110
+
111
+
112
+ def controlled_moss_block(block, tokens, mod, pe, transformer_options, modules, condition, strength, mask=None):
113
+ """MOSS's native image-generation block with LLLite before Q/K/V projections."""
114
+ y = moss.modulated_rms_norm(tokens, block.norm1, mod[1], mod[0])
115
+ q, k, v = project_qkv(block.proj, block.fused_qkv, y, modules, condition, strength)
116
+ out = moss.attention(q, k, v, block.q_norm, block.k_norm, pe, block.num_heads,
117
+ block.comfy_attention, transformer_options, mask=mask)
118
+ tokens = moss.gated_residual(tokens, block.proj['o'](out), mod[2])
119
+ y = moss.modulated_rms_norm(tokens, block.norm2, mod[4], mod[3])
120
+ y = block.proj['ffn_up_gate'](y)
121
+ return moss.gated_residual(tokens, comfy.ops.linear_input_act(block.proj['ffn_down'], y, 'situ'), mod[5])
ComfyUI-MOSS-LLLite/lllite_nodes.py ADDED
@@ -0,0 +1,137 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """Standalone MOSS LLLite nodes using ModelPatcher-owned block replacements."""
2
+
3
+ import torch
4
+ import torch.nn.functional as F
5
+
6
+ import folder_paths
7
+ import comfy.ldm.moss_image.model
8
+ import comfy.lora
9
+ import comfy.model_management
10
+ import comfy.model_patcher
11
+ import comfy.model_prefetch
12
+ import comfy.ops
13
+ import comfy.patcher_extension
14
+ import comfy.utils
15
+ from comfy_api.latest import io
16
+
17
+ from .lllite_model import MossLLLiteControl, PREFIX, config_from_metadata, controlled_moss_block
18
+
19
+ LLLiteType = io.Custom('MOSS_LLLITE_CONTROL')
20
+ CONTROL_KEY = 'moss_lllite'
21
+
22
+
23
+ class MossLLLiteLoader(io.ComfyNode):
24
+ @classmethod
25
+ def define_schema(cls):
26
+ return io.Schema(
27
+ node_id='MossLLLiteLoader', display_name='Load MOSS-LLLite', category='MOSS-LLLite',
28
+ inputs=[io.Combo.Input('name', options=folder_paths.get_filename_list('model_patches'))],
29
+ outputs=[LLLiteType.Output()],
30
+ )
31
+
32
+ @classmethod
33
+ def execute(cls, name):
34
+ path = folder_paths.get_full_path_or_raise('model_patches', name)
35
+ state, metadata = comfy.utils.load_torch_file(path, safe_load=True, return_metadata=True)
36
+ control = MossLLLiteControl(**config_from_metadata(metadata), operations=comfy.ops.manual_cast,
37
+ device=comfy.model_management.unet_offload_device(),
38
+ dtype=comfy.utils.weight_dtype(state))
39
+ if any(not key.startswith(PREFIX) for key in state):
40
+ raise ValueError('MOSS LLLite checkpoint contains non-control weights')
41
+ control.load_state_dict({key[len(PREFIX):]: value for key, value in state.items()}, strict=True)
42
+ patcher = comfy.model_patcher.ModelPatcher(control, load_device=comfy.model_management.get_torch_device(),
43
+ offload_device=comfy.model_management.unet_offload_device())
44
+ return io.NodeOutput(patcher)
45
+
46
+
47
+ class MossLLLitePatch:
48
+ def __init__(self, control, image, strength, sigma_start, sigma_end):
49
+ self.control, self.image, self.strength = control, image, strength
50
+ self.sigma_start, self.sigma_end = sigma_start, sigma_end
51
+
52
+ def __call__(self, executor, x, timestep, context, transformer_options={}, **kwargs):
53
+ sigmas = transformer_options.get('sigmas')
54
+ if sigmas is not None and not self.sigma_end <= float(sigmas.max()) <= self.sigma_start:
55
+ return executor(x, timestep, context, transformer_options, **kwargs)
56
+ if x.shape[1] != 128:
57
+ raise ValueError('Remove legacy concat Tile LoRAs before applying MOSS LLLite')
58
+ if len(kwargs.get('ref_latents', [])) or kwargs.get('ref_kv') is not None:
59
+ raise ValueError('MOSS LLLite currently supports image generation, without MOSS-Edit reference tokens')
60
+ image = comfy.utils.common_upscale(self.image.movedim(-1, 1), x.shape[-1] * 16, x.shape[-2] * 16,
61
+ 'bicubic', 'center').clamp(0, 1)
62
+ # Preserve encoder batch shape: BF16 convolution rounding depends on it.
63
+ image = comfy.utils.repeat_to_batch_size(image.to(device=x.device, dtype=x.dtype), x.shape[0])
64
+ condition = self.control.model.conditioner(image * 2 - 1)
65
+ mask = kwargs.get('context_attention_bias')
66
+ if mask is not None:
67
+ mask = F.pad(mask, (x.shape[-2] * x.shape[-1], 0))[:, None, None]
68
+ options = transformer_options.copy()
69
+ options[CONTROL_KEY] = {**options.get(CONTROL_KEY, {}), self: (condition, mask)}
70
+ return executor(x, timestep, context, options, **kwargs)
71
+
72
+
73
+ class MossLLLiteBlockPatch:
74
+ def __init__(self, patch, index, block, previous):
75
+ self.patch, self.index, self.block, self.previous = patch, index, block, previous
76
+
77
+ def __call__(self, args, extra):
78
+ data = args['transformer_options'].get(CONTROL_KEY, {}).get(self.patch)
79
+ if data is None:
80
+ return self.previous(args, extra) if self.previous is not None else extra['original_block'](args)
81
+ condition, mask = data
82
+
83
+ def controlled(arguments):
84
+ # Control tensors must not be owned by the backbone memory compiler's allocation scope.
85
+ with comfy.model_prefetch.pause_malloc_graph():
86
+ output = controlled_moss_block(
87
+ self.block, arguments['img'], arguments['mod'], arguments['pe'],
88
+ arguments['transformer_options'], self.patch.control.model.blocks[self.index],
89
+ condition, self.patch.strength, mask,
90
+ )
91
+ return {'img': output}
92
+
93
+ if self.previous is not None:
94
+ return self.previous(args, {**extra, 'original_block': controlled})
95
+ return controlled(args)
96
+
97
+ def to(self, device_or_dtype):
98
+ return self
99
+
100
+
101
+ class MossLLLiteApply(io.ComfyNode):
102
+ @classmethod
103
+ def define_schema(cls):
104
+ return io.Schema(
105
+ node_id='MossLLLiteApply', display_name='Apply MOSS-LLLite', category='MOSS-LLLite',
106
+ description='Anima-style RGB LLLite adapted to MOSS. Strength 0 returns the input model unchanged.',
107
+ inputs=[io.Model.Input('model'), LLLiteType.Input('control'), io.Image.Input('image'),
108
+ io.Float.Input('strength', default=1.0, min=0.0, max=10.0, step=0.01),
109
+ io.Float.Input('start_percent', default=0.0, min=0.0, max=1.0, step=0.001),
110
+ io.Float.Input('end_percent', default=1.0, min=0.0, max=1.0, step=0.001)],
111
+ outputs=[io.Model.Output()],
112
+ )
113
+
114
+ @classmethod
115
+ def execute(cls, model, control, image, strength=1.0, start_percent=0.0, end_percent=1.0):
116
+ if strength == 0 or start_percent >= end_percent:
117
+ return io.NodeOutput(model)
118
+ backbone = model.get_model_object('diffusion_model')
119
+ if not isinstance(backbone, comfy.ldm.moss_image.model.MOSSImage):
120
+ raise ValueError('MOSS LLLite requires a MOSS-Image backbone')
121
+ key = 'diffusion_model.video_embedder.weight'
122
+ weights = model.get_key_patches(key)[key]
123
+ shape = comfy.lora.calculate_shape(weights[1:], weights[0][0], key)
124
+ config = control.model.config
125
+ if shape != (config['model_dim'], 128) or len(backbone.layers) != config['num_blocks']:
126
+ raise ValueError('MOSS LLLite architecture does not match this backbone; remove concat Tile LoRAs')
127
+ sampling = model.get_model_object('model_sampling')
128
+ patch = MossLLLitePatch(control, image[..., :3], strength, float(sampling.percent_to_sigma(start_percent)),
129
+ float(sampling.percent_to_sigma(end_percent)))
130
+ result = model.clone()
131
+ result.add_wrapper(comfy.patcher_extension.WrappersMP.DIFFUSION_MODEL, patch)
132
+ existing = result.model_options['transformer_options'].get('patches_replace', {}).get('dit', {})
133
+ for index, block in enumerate(backbone.layers):
134
+ result.set_model_patch_replace(MossLLLiteBlockPatch(patch, index, block, existing.get(('single_block', index))),
135
+ 'dit', 'single_block', index)
136
+ result.set_additional_models(CONTROL_KEY, result.get_additional_models_with_key(CONTROL_KEY) + [control])
137
+ return io.NodeOutput(result)
{ComfyUI-Moss-Tile-Control → ComfyUI-MOSS-LLLite}/model.py RENAMED
File without changes
{ComfyUI-Moss-Tile-Control → ComfyUI-MOSS-LLLite}/nodes.py RENAMED
File without changes
ComfyUI-MOSS-LLLite/pyproject.toml ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [project]
2
+ name = "comfyui-moss-lllite"
3
+ version = "0.1.2"
4
+ description = "Anima-style LLLite control for MOSS-Image in ComfyUI"
5
+ readme = "README.md"
6
+ license = { file = "LICENSE" }
7
+ dependencies = []
8
+
9
+ [project.urls]
10
+ Repository = "https://github.com/X-absurd/ComfyUI-MOSS-LLLite"
11
+ Issues = "https://github.com/X-absurd/ComfyUI-MOSS-LLLite/issues"
ComfyUI-MOSS-LLLite/tests/test_lllite_nodes.py ADDED
@@ -0,0 +1,187 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import importlib.util
2
+ from pathlib import Path
3
+ import sys
4
+ import tempfile
5
+ import unittest
6
+ from unittest.mock import patch
7
+
8
+ ROOT = Path(__file__).resolve().parents[3]
9
+ sys.path.insert(0, str(ROOT))
10
+ import comfy.cli_args
11
+ comfy.cli_args.args.cpu = True
12
+ import comfy.model_management
13
+ import comfy.model_patcher
14
+ import comfy.model_sampling
15
+ import comfy.ldm.moss_image.model as moss
16
+ import comfy.ops
17
+ import safetensors.torch
18
+ import torch
19
+ from torch import nn
20
+
21
+ PACKAGE = Path(__file__).resolve().parents[1]
22
+ spec = importlib.util.spec_from_file_location('moss_lllite_test', PACKAGE / '__init__.py', submodule_search_locations=[str(PACKAGE)])
23
+ package = importlib.util.module_from_spec(spec)
24
+ sys.modules[spec.name] = package
25
+ spec.loader.exec_module(package)
26
+ nodes = sys.modules[spec.name + '.lllite_nodes']
27
+ architecture = sys.modules[spec.name + '.lllite_model']
28
+
29
+
30
+ def fixture(fused=False):
31
+ backbone = moss.MOSSImage(dim=32, ffn_dim=64, num_heads=2, num_layers=2, text_dim=16,
32
+ fused_qkv=fused, operations=nn, dtype=torch.float32, device='cpu')
33
+ base = nn.Module()
34
+ base.diffusion_model = backbone
35
+ base.model_sampling = comfy.model_sampling.ModelSamplingDiscreteFlow()
36
+ model = comfy.model_patcher.ModelPatcher(base, load_device=torch.device('cpu'), offload_device=torch.device('cpu'))
37
+ control = architecture.MossLLLiteControl(32, 2, rank=4, condition_dim=16, condition_embedding_dim=8)
38
+ for block in control.blocks:
39
+ for module in block.values():
40
+ nn.init.normal_(module.up.weight, std=0.1)
41
+ nn.init.zeros_(module.up.bias)
42
+ nn.init.zeros_(module.depth_embed)
43
+ holder = comfy.model_patcher.ModelPatcher(control, load_device=torch.device('cpu'), offload_device=torch.device('cpu'))
44
+ return model, holder, torch.rand(1, 64, 64, 3)
45
+
46
+
47
+ class NodeTests(unittest.TestCase):
48
+ def setUp(self):
49
+ torch.manual_seed(21)
50
+ torch.set_num_threads(1)
51
+
52
+ def test_zero_and_empty_interval_return_original_model(self):
53
+ model, control, image = fixture()
54
+ self.assertIs(nodes.MossLLLiteApply.execute(model, control, image, 0).result[0], model)
55
+ self.assertIs(nodes.MossLLLiteApply.execute(model, control, image, 1, .8, .2).result[0], model)
56
+
57
+ def test_real_block_matches_original_at_zero_and_changes_at_nonzero(self):
58
+ # Native CPU attention and differentiable norm path, without kitchen CUDA kernels.
59
+ with patch.object(comfy.model_management, 'in_training', True):
60
+ for fused in (False, True):
61
+ model, control, image = fixture(fused)
62
+ backbone = model.model.diffusion_model
63
+ x = torch.randn(2, 19, 32)
64
+ condition = control.model.conditioner(image.movedim(-1, 1) * 2 - 1).repeat(2, 1, 1)
65
+ mod = tuple(torch.randn(2, 1, 32) * 0.1 for _ in range(6))
66
+ pe = backbone.rope(4, 4, 3, torch.device('cpu'))
67
+ block = backbone.layers[0]
68
+ baseline = block(x.clone(), mod, pe)
69
+ zero = architecture.controlled_moss_block(block, x.clone(), mod, pe, {}, control.model.blocks[0], condition, 0)
70
+ active = architecture.controlled_moss_block(block, x.clone(), mod, pe, {}, control.model.blocks[0], condition, 1)
71
+ torch.testing.assert_close(zero, baseline, atol=1e-6, rtol=1e-5)
72
+ self.assertFalse(torch.equal(active, baseline))
73
+ self.assertTrue(torch.isfinite(active).all())
74
+
75
+ def test_wrapper_cfg_repeat_schedule_and_prompt_weights(self):
76
+ model, control, image = fixture()
77
+ patcher = nodes.MossLLLiteApply.execute(model, control, image).result[0]
78
+ wrapper = patcher.get_all_wrappers(comfy.patcher_extension.WrappersMP.DIFFUSION_MODEL)[0]
79
+ observed = {}
80
+ def executor(x, timestep, context, options, **kwargs):
81
+ observed.update(options); return x
82
+ x, context = torch.randn(2, 128, 4, 4), torch.randn(2, 3, 16)
83
+ with patch.object(control.model.conditioner, 'forward', wraps=control.model.conditioner.forward) as encode:
84
+ wrapper(executor, x, torch.tensor([.5]), context, {}, context_attention_bias=torch.zeros(2, 3))
85
+ self.assertEqual(encode.call_args.args[0].shape[0], 2)
86
+ condition, mask = observed[nodes.CONTROL_KEY][wrapper]
87
+ self.assertEqual(condition.shape, (2, 16, 8))
88
+ self.assertTrue(torch.equal(condition[0], condition[1]))
89
+ self.assertEqual(mask.shape, (2, 1, 1, 19))
90
+ observed.clear()
91
+ wrapper.sigma_start, wrapper.sigma_end = .3, .1
92
+ wrapper(executor, x, torch.tensor([.8]), context, {'sigmas': torch.tensor([.8])})
93
+ self.assertNotIn(nodes.CONTROL_KEY, observed)
94
+
95
+ def test_fused_projection_rows_bias_and_gradients(self):
96
+ projection = nn.Linear(32, 96)
97
+ inputs = [torch.randn(2, 19, 32, requires_grad=True) for _ in range(3)]
98
+ expected = tuple(projection(value).chunk(3, dim=-1)[i] for i, value in enumerate(inputs))
99
+ actual = architecture.project_fused_qkv(projection, inputs)
100
+ for result, reference in zip(actual, expected):
101
+ torch.testing.assert_close(result, reference)
102
+ parameters = [*inputs, projection.weight, projection.bias]
103
+ reference_grads = torch.autograd.grad(sum(value.square().sum() for value in expected), parameters)
104
+ actual_grads = torch.autograd.grad(sum(value.square().sum() for value in actual), parameters)
105
+ for result, reference in zip(actual_grads, reference_grads):
106
+ torch.testing.assert_close(result, reference)
107
+
108
+ def test_fused_projection_preserves_comfy_weight_and_bias_patches(self):
109
+ projection = comfy.ops.manual_cast.Linear(32, 96)
110
+ projection.load_state_dict(nn.Linear(32, 96).state_dict())
111
+ delta = torch.randn_like(projection.weight) * .01
112
+ projection.weight_function = [lambda weight: weight + delta]
113
+ projection.bias_function = [lambda bias: bias + .1]
114
+ inputs = [torch.randn(2, 19, 32) for _ in range(3)]
115
+ expected = tuple(projection(value).chunk(3, dim=-1)[i] for i, value in enumerate(inputs))
116
+ actual = architecture.project_fused_qkv(projection, inputs)
117
+ for result, reference in zip(actual, expected):
118
+ torch.testing.assert_close(result, reference)
119
+
120
+ def test_custom_projection_retains_its_forward(self):
121
+ class CustomLinear(nn.Linear):
122
+ def forward(self, value):
123
+ return super().forward(value) * 2
124
+
125
+ projection = CustomLinear(32, 96)
126
+ inputs = [torch.randn(1, 19, 32) for _ in range(3)]
127
+ self.assertIs(architecture.project_fused_qkv(projection, inputs), NotImplemented)
128
+
129
+ @unittest.skipUnless(torch.cuda.is_available(), 'CUDA required for native INT8 ConvRot')
130
+ def test_fused_int8_convrot_rows_match_native_projection(self):
131
+ projection = comfy.ops.mixed_precision_ops(compute_dtype=torch.bfloat16).Linear(
132
+ 256, 768, bias=False, device='cuda')
133
+ weight = architecture.QuantizedTensor.from_float(
134
+ torch.randn(768, 256, device='cuda', dtype=torch.bfloat16),
135
+ 'TensorWiseINT8Layout', per_channel=True, convrot=True, convrot_groupsize=256)
136
+ projection.weight = nn.Parameter(weight, requires_grad=False)
137
+ projection.quant_format = 'int8_tensorwise'
138
+ projection.layout_type = 'TensorWiseINT8Layout'
139
+ inputs = [torch.randn(2, 32, 256, device='cuda', dtype=torch.bfloat16) for _ in range(3)]
140
+ expected = tuple(projection(value).chunk(3, dim=-1)[i] for i, value in enumerate(inputs))
141
+ actual = architecture.project_fused_qkv(projection, inputs)
142
+ for result, reference in zip(actual, expected):
143
+ torch.testing.assert_close(result, reference, atol=0, rtol=0)
144
+ delta = torch.randn(768, 256, device='cuda', dtype=torch.bfloat16) * .01
145
+ projection.weight_function = [lambda value: value + delta]
146
+ expected = tuple(projection(value).chunk(3, dim=-1)[i] for i, value in enumerate(inputs))
147
+ actual = architecture.project_fused_qkv(projection, inputs)
148
+ for result, reference in zip(actual, expected):
149
+ torch.testing.assert_close(result, reference, atol=0, rtol=0)
150
+
151
+ def test_cfg_condition_batch_order(self):
152
+ model, control, image = fixture()
153
+ image = image.repeat(2, 1, 1, 1)
154
+ image[1] = 1 - image[1]
155
+ wrapper = nodes.MossLLLitePatch(control, image, 1, float('inf'), 0)
156
+ observed = {}
157
+ def executor(x, timestep, context, options, **kwargs):
158
+ observed.update(options); return x
159
+ wrapper(executor, torch.randn(4, 128, 4, 4), torch.tensor([.5]), torch.randn(4, 3, 16), {})
160
+ actual = observed[nodes.CONTROL_KEY][wrapper][0]
161
+ pixels = image.movedim(-1, 1).repeat(2, 1, 1, 1) * 2 - 1
162
+ reference = control.model.conditioner(pixels)
163
+ torch.testing.assert_close(actual, reference)
164
+
165
+ def test_loader_roundtrip(self):
166
+ model, control, image = fixture()
167
+ state = {architecture.PREFIX + k: v for k, v in control.model.state_dict().items()}
168
+ with tempfile.TemporaryDirectory() as directory:
169
+ path = Path(directory) / 'control.safetensors'
170
+ safetensors.torch.save_file(state, path, metadata=control.model.metadata())
171
+ with patch.object(nodes.folder_paths, 'get_full_path_or_raise', return_value=str(path)):
172
+ loaded = nodes.MossLLLiteLoader.execute('control.safetensors').result[0]
173
+ for name, value in control.model.state_dict().items():
174
+ self.assertTrue(torch.equal(value, loaded.model.state_dict()[name]))
175
+
176
+ def test_reject_concat_projection_and_reference_tokens(self):
177
+ model, control, image = fixture()
178
+ model.add_patches({'diffusion_model.video_embedder.weight': ('set', (torch.randn(32, 384),))})
179
+ with self.assertRaises(ValueError): nodes.MossLLLiteApply.execute(model, control, image)
180
+ wrapper = nodes.MossLLLitePatch(control, image, 1, float('inf'), 0)
181
+ with self.assertRaises(ValueError):
182
+ wrapper(lambda *a: None, torch.zeros(1, 128, 4, 4), torch.zeros(1), torch.zeros(1, 3, 16),
183
+ ref_latents=[torch.zeros(1, 128, 4, 4)])
184
+
185
+
186
+ if __name__ == '__main__':
187
+ unittest.main()
ComfyUI-MOSS-LLLite/tests/test_nodes.py ADDED
@@ -0,0 +1,191 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """Run from the ComfyUI root: python custom_nodes/ComfyUI-MOSS-LLLite/tests/test_nodes.py."""
2
+
3
+ import importlib.util
4
+ from contextlib import contextmanager
5
+ from pathlib import Path
6
+ import sys
7
+ import tempfile
8
+ import unittest
9
+ from unittest.mock import Mock, patch
10
+
11
+ ROOT = Path(__file__).resolve().parents[3]
12
+ sys.path.insert(0, str(ROOT))
13
+ import comfy.cli_args
14
+ comfy.cli_args.args.cpu = True
15
+ import comfy.model_management
16
+ import comfy.model_patcher
17
+ import comfy.model_sampling
18
+ import comfy.patcher_extension
19
+ import comfy.ldm.moss_image.model
20
+ import safetensors.torch
21
+ import torch
22
+ from torch import nn
23
+
24
+ PACKAGE = Path(__file__).resolve().parents[1]
25
+ spec = importlib.util.spec_from_file_location("moss_tile_node_test", PACKAGE / "__init__.py", submodule_search_locations=[str(PACKAGE)])
26
+ package = importlib.util.module_from_spec(spec)
27
+ sys.modules[spec.name] = package
28
+ spec.loader.exec_module(package)
29
+ nodes = sys.modules[spec.name + ".nodes"]
30
+ architecture = sys.modules[spec.name + ".model"]
31
+
32
+
33
+ def fixture():
34
+ backbone = comfy.ldm.moss_image.model.MOSSImage(
35
+ dim=32, ffn_dim=64, num_heads=2, num_layers=2, text_dim=16,
36
+ dtype=torch.float32, device="cpu", operations=nn,
37
+ )
38
+ base = nn.Module()
39
+ base.diffusion_model = backbone
40
+ base.model_sampling = comfy.model_sampling.ModelSamplingDiscreteFlow()
41
+ model = comfy.model_patcher.ModelPatcher(base, load_device=torch.device("cpu"), offload_device=torch.device("cpu"))
42
+ control = architecture.MossTileControl(32, 2, 8, condition_dim=8)
43
+ holder = comfy.model_patcher.ModelPatcher(control, load_device=torch.device("cpu"), offload_device=torch.device("cpu"))
44
+ vae = Mock()
45
+ vae.encode.return_value = torch.randn(1, 128, 2, 3)
46
+ image = torch.rand(1, 32, 48, 3)
47
+ return model, holder, vae, image
48
+
49
+
50
+ class NodeTests(unittest.TestCase):
51
+ def setUp(self):
52
+ torch.manual_seed(11)
53
+ torch.set_num_threads(1)
54
+
55
+ def test_zero_returns_exact_model_without_encoding(self):
56
+ model, control, vae, image = fixture()
57
+ self.assertIs(nodes.MossTileControlApply.execute(model, control, vae, image, 0).result[0], model)
58
+ vae.encode.assert_not_called()
59
+ self.assertFalse(model.get_all_wrappers(comfy.patcher_extension.WrappersMP.DIFFUSION_MODEL))
60
+
61
+ def test_reject_pending_legacy_projection_before_encoding(self):
62
+ model, control, vae, image = fixture()
63
+ model.add_patches({"diffusion_model.video_embedder.weight": ("set", (torch.randn(32, 384),))})
64
+ self.assertEqual(model.model.diffusion_model.video_embedder.weight.shape[1], 128)
65
+ with self.assertRaisesRegex(ValueError, "legacy concat tile LoRA"):
66
+ nodes.MossTileControlApply.execute(model, control, vae, image)
67
+ vae.encode.assert_not_called()
68
+
69
+ def test_apply_after_legacy_clone_has_resized_shared_projection(self):
70
+ model, control, vae, image = fixture()
71
+ key = "diffusion_model.video_embedder.weight"
72
+ legacy = model.clone()
73
+ legacy.add_patches({key: ("set", (torch.randn(32, 384),))})
74
+ legacy.patch_weight_to_device(key, device_to=torch.device("cpu"))
75
+ self.assertEqual(model.model.diffusion_model.video_embedder.weight.shape, (32, 384))
76
+ self.assertEqual(model.backup[key].weight.shape, (32, 128))
77
+
78
+ result = nodes.MossTileControlApply.execute(model, control, vae, image).result[0]
79
+
80
+ vae.encode.assert_called_once()
81
+ self.assertIsNot(result, model)
82
+ self.assertEqual(model.model.diffusion_model.video_embedder.weight.shape, (32, 384))
83
+ with self.assertRaisesRegex(ValueError, "legacy concat tile LoRA"):
84
+ nodes.MossTileControlApply.execute(legacy, control, vae, image)
85
+
86
+ def test_residual_allocations_are_outside_backbone_graph(self):
87
+ _, control, _, _ = fixture()
88
+ controller = nodes.MossTilePatch(control, torch.randn(1, 128, 2, 3), 1, 1, 0)
89
+ block = nodes.MossTileBlockPatch(controller, 0, None)
90
+ x = torch.randn(1, 8, 32)
91
+ condition = torch.randn(1, 6, 8)
92
+ paused = False
93
+ @contextmanager
94
+ def pause():
95
+ nonlocal paused
96
+ paused = True
97
+ try:
98
+ yield
99
+ finally:
100
+ paused = False
101
+ def base(args):
102
+ self.assertFalse(paused)
103
+ return {"img": args["img"]}
104
+ forward = control.model.blocks[0].forward
105
+ def residual(*args):
106
+ self.assertTrue(paused)
107
+ return forward(*args)
108
+ with patch.object(nodes.comfy.model_prefetch, "pause_malloc_graph", pause), patch.object(control.model.blocks[0], "forward", residual):
109
+ out = block({"img": x, "transformer_options": {nodes.CONTROL_KEY: {controller: condition}}}, {"original_block": base})
110
+ self.assertFalse(paused)
111
+ self.assertTrue(torch.equal(out["img"], forward(x, condition, 1)))
112
+
113
+ def test_schedule_batch_and_patch_composition(self):
114
+ model, control, vae, image = fixture()
115
+ before = {k: v.clone() for k, v in model.model.diffusion_model.state_dict().items()}
116
+ def previous(args, extra):
117
+ out = extra["original_block"](args)
118
+ return {"img": out["img"] + 0.25}
119
+ model.set_model_patch_replace(previous, "dit", "single_block", 0)
120
+ result = nodes.MossTileControlApply.execute(model, control, vae, image, 0.5, 0.2, 0.8).result[0]
121
+ self.assertIsNot(result, model)
122
+ self.assertEqual(result.get_additional_models_with_key(nodes.CONTROL_KEY), [control])
123
+ self.assertIs(model.model_options["transformer_options"]["patches_replace"]["dit"][("single_block", 0)], previous)
124
+ wrapper = result.get_all_wrappers(comfy.patcher_extension.WrappersMP.DIFFUSION_MODEL)[0]
125
+ block = result.model_options["transformer_options"]["patches_replace"]["dit"][("single_block", 0)]
126
+ x = torch.randn(2, 8, 32)
127
+ options = {"sigmas": torch.tensor([0.5])}
128
+ observed = {}
129
+ def execute(latent, time, context, transformer_options, **kwargs):
130
+ observed.update(transformer_options)
131
+ return block({"img": x, "transformer_options": transformer_options},
132
+ {"original_block": lambda args: {"img": args["img"]}})["img"]
133
+ out = wrapper(execute, torch.randn(2, 128, 2, 3), torch.ones(2), None, options)
134
+ condition = observed[nodes.CONTROL_KEY][wrapper]
135
+ self.assertEqual(condition.shape, (2, 6, 8))
136
+ self.assertTrue(torch.equal(out, control.model.blocks[0](x + 0.25, condition, 0.5)))
137
+ self.assertNotIn(nodes.CONTROL_KEY, options)
138
+ options["sigmas"] = torch.tensor([0.99])
139
+ self.assertTrue(torch.equal(wrapper(execute, torch.randn(2, 128, 2, 3), torch.ones(2), None, options), x + 0.25))
140
+ self.assertFalse(hasattr(wrapper, "condition"))
141
+ for key, value in model.model.diffusion_model.state_dict().items():
142
+ self.assertTrue(torch.equal(value, before[key]))
143
+
144
+ def test_loader_roundtrip_and_reject_legacy(self):
145
+ _, control, _, _ = fixture()
146
+ with tempfile.TemporaryDirectory() as tmp:
147
+ path = Path(tmp) / "tile.safetensors"
148
+ state = {architecture.PREFIX + k: v for k, v in control.model.state_dict().items()}
149
+ safetensors.torch.save_file(state, path, metadata=control.model.metadata())
150
+ with patch.object(nodes.folder_paths, "get_full_path_or_raise", return_value=str(path)):
151
+ loaded = nodes.MossTileControlLoader.execute("tile.safetensors").result[0]
152
+ for key, value in control.model.state_dict().items():
153
+ self.assertTrue(torch.equal(value, loaded.model.state_dict()[key]))
154
+ safetensors.torch.save_file({"diffusion_model.video_embedder.set_weight": torch.ones(32, 384)}, path, metadata={"format": "pt"})
155
+ with patch.object(nodes.folder_paths, "get_full_path_or_raise", return_value=str(path)):
156
+ with self.assertRaises(ValueError):
157
+ nodes.MossTileControlLoader.execute("tile.safetensors")
158
+
159
+ def test_cfg_batch_preserves_control_image_order(self):
160
+ _, control, _, _ = fixture()
161
+ latent = torch.randn(2, 128, 2, 3)
162
+ wrapper = nodes.MossTilePatch(control, latent, 1, 1, 0)
163
+ def execute(x, t, context, options):
164
+ return options[nodes.CONTROL_KEY][wrapper]
165
+ condition = wrapper(execute, torch.zeros(4, 128, 2, 3), None, None)
166
+ expected = control.model.conditioner(latent)
167
+ self.assertTrue(torch.equal(condition[:2], expected))
168
+ self.assertTrue(torch.equal(condition[2:], expected))
169
+
170
+ def test_real_moss_forward_uses_patches_and_zero_schedule(self):
171
+ model, control, vae, image = fixture()
172
+ result = nodes.MossTileControlApply.execute(model, control, vae, image, 0.5, 0.2, 0.8).result[0]
173
+ backbone = model.model.diffusion_model
174
+ x, t, context = torch.randn(1, 128, 2, 3), torch.tensor([0.5]), torch.randn(1, 4, 16)
175
+ def forward(patcher, sigma):
176
+ options = comfy.patcher_extension.copy_nested_dicts(patcher.model_options["transformer_options"])
177
+ options["sigmas"] = torch.tensor([sigma])
178
+ options["wrappers"] = patcher.wrappers
179
+ return backbone(x, t, context, transformer_options=options)
180
+ # Exercise the differentiable CPU attention path; CUDA inference is checked separately.
181
+ with patch.object(comfy.model_management, "in_training", True):
182
+ baseline = forward(model, 0.99)
183
+ outside = forward(result, 0.99)
184
+ active = forward(result, 0.5)
185
+ self.assertTrue(torch.equal(baseline, outside))
186
+ self.assertFalse(torch.equal(baseline, active))
187
+ self.assertEqual(active.shape, x.shape)
188
+
189
+
190
+ if __name__ == "__main__":
191
+ unittest.main()
ComfyUI-Moss-Tile-Control/README.md DELETED
@@ -1,30 +0,0 @@
1
- # MOSS Tile Control
2
-
3
- Standalone residual tile control for MOSS-Image, trained in diffusion-pipe with
4
- `control_adapter = 'tile_residual'` and adapter `type = 'moss_tile'`.
5
- No additional Python packages are required beyond ComfyUI's dependencies.
6
-
7
- 1. Restart ComfyUI after installing this directory under `custom_nodes`.
8
- 2. Put a trained residual adapter in `models/model_patches`.
9
- 3. Use **Load MOSS Tile Control** → **Apply MOSS Tile Control**.
10
- 4. Connect MOSS MODEL, Flux2 VAE and the tile image; send the output MODEL to the sampler.
11
-
12
- Strength scales every control residual. At 0 the input model is returned unchanged.
13
- Start/end percentages restrict the control to part of sampling. Text conditioning
14
- connects directly to the sampler. Other LoRAs may be loaded before or after this node.
15
-
16
- Remove legacy concat tile LoRAs from the input MODEL chain before applying this
17
- adapter; pending input-projection replacements are checked before VAE encoding.
18
- Compatibility checks use the current ModelPatcher's original weights and patches,
19
- so switching from a previous legacy tile run does not reject a compatible model.
20
- Control residual allocations run outside the backbone's aimdo allocation scope,
21
- so dynamic VRAM and the Comfy memory compiler can remain enabled.
22
-
23
- Legacy MOSS concat tile LoRAs and Anima LLLite weights cannot be loaded here. These
24
- weights require a fresh training run. Training and inference should use the same
25
- Base/SFT family. Establish quality with ordinary sampling before testing accelerated
26
- LoRAs or quantized backbones.
27
-
28
- The architecture in `model.py` matches diffusion-pipe's `models/moss_tile_control.py`.
29
- The node uses ModelPatcher wrappers, block patches and additional-model loading;
30
- it does not replace backbone modules or merge control weights into the backbone.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
README.md CHANGED
@@ -1,153 +1,204 @@
1
- ---
2
- language:
3
- - en
4
- tags:
5
- - moss-image
6
- - tile
7
- - controlnet
8
- - image-to-image
9
- - image-restoration
10
- - comfyui
11
- - adapter
12
- base_model:
13
- - OpenMOSS-Team/MOSS-Image-Base
14
- - OpenMOSS-Team/MOSS-Image-SFT
15
- base_model_relation: adapter
16
- pipeline_tag: image-to-image
17
- library_name: diffusion-single-file
18
- ---
19
- A little support goes a long way! If you’d like to help me keep creating, you can do so at https://ko-fi.com/xabsurd
20
- # MOSS-Image Tile Control
21
-
22
- Tile control for MOSS-Image, designed to preserve composition, subject outlines, and local structure while allowing the model to regenerate textures and details.
23
-
24
- Use it for detail enhancement, refinement after upscaling, and image variations guided by an existing image. Outputs are generated rather than copied, so exact pixel reconstruction and recovery of authentic original details are not guaranteed.
25
-
26
- Two beta versions are available: **beta 2** for a simpler setup and **beta 3** for finer control over how much the output changes. The author has manually tested a variety of images with positive overall results.
27
-
28
- ## Before and after
29
-
30
- Each comparison shows the blurred input on the left and the Tile Control output on the right.
31
-
32
- | Beta 3 · epoch 2 | Beta 2 |
33
- | --- | --- |
34
- | ![Flying cat before and after with beta 3](examples/beta3-epoch2-comparison.png) | ![Flying cat before and after with beta 2](examples/beta2-comparison.png) |
35
- | ![Jetpack Cat before and after with beta 3](examples/beta3-epoch2-comparison2.png) | ![Jetpack Cat before and after with beta 2](examples/beta2-comparison2.png) |
36
-
37
- ## Downloads
38
-
39
- - **Recommended** · [beta 3 epoch 2](https://huggingface.co/Xabsurd/MOSS-Image-Tile-Control/resolve/main/beta3/epoch2/adapter_model.safetensors) · MOSS-Image-SFT.
40
- - [beta 3 epoch 1](https://huggingface.co/Xabsurd/MOSS-Image-Tile-Control/resolve/main/beta3/epoch1/adapter_model.safetensors) · Alternate checkpoint for MOSS-Image-SFT.
41
- - [beta 2 epoch 1](https://huggingface.co/Xabsurd/MOSS-Image-Tile-Control/resolve/main/beta2/epoch1/adapter_model.safetensors) · LoRA for MOSS-Image-Base.
42
- - [ComfyUI workflows and custom nodes](https://huggingface.co/Xabsurd/MOSS-Image-Tile-Control/resolve/main/workflow-and-nodes.zip).
43
-
44
- The checkpoints share a filename. Keep their version and epoch subdirectories when installing more than one checkpoint. The beta 3 sample workflows expect the weight at `ComfyUI/models/model_patches/moss-tile-control/beta3/epoch2/adapter_model.safetensors`.
45
-
46
- The archive contains test workflows. After importing a workflow, select your local model, encoder, VAE, adapter, and input image. The test workflows are saved with SFT INT8 and an active third-party Coffmeteor 15-step acceleration LoRA. For beta 2, select MOSS-Image-Base as described below. The color-match variants also require the `ColorMatchV2` node from [ComfyUI-KJNodes](https://github.com/kijai/ComfyUI-KJNodes).
47
-
48
- ## Which version should I use?
49
-
50
- | Feature | beta 2 | beta 3 |
51
- | --- | --- | --- |
52
- | Main advantage | No additional custom nodes required | Finer control over structure preservation and detail regeneration |
53
- | Control method | Concatenated image conditioning with a backbone LoRA | Independent lightweight residual control branch |
54
- | ComfyUI support | Built-in nodes in a ComfyUI version with MOSS-Image control support | Requires the accompanying `ComfyUI-Moss-Tile-Control` custom nodes |
55
- | Weight location | `ComfyUI/models/loras/` | `ComfyUI/models/model_patches/` |
56
- | Strength adjustment | Control-image strength and LoRA strength affect the result separately | One strength parameter scales the control residuals across layers |
57
- | At control strength 0 | The Tile LoRA can still affect the output | Tile control is bypassed, preserving the input model and its existing settings |
58
- | Base model family | MOSS-Image-Base | MOSS-Image-SFT |
59
-
60
- **Choose beta 2 for convenience. Start with beta 3 if you want finer adjustment of the modification strength.**
61
-
62
- The weight formats and loading methods are different. Files are not interchangeable, and equal strength values do not imply equal control across the two versions.
63
-
64
- ## beta 2: Built-in nodes, no extra installation
65
-
66
- beta 2 uses ComfyUI's existing MOSS-Image control interface. No additional custom nodes are required for this adapter.
67
-
68
- ### Setup
69
-
70
- 1. Use a ComfyUI version that supports MOSS-Image and includes `MOSS-Image Color/Depth Control`.
71
- 2. Place the beta 2 weights in `ComfyUI/models/loras/`.
72
- 3. Load them with `Load LoRA (Model Only)` and connect its MODEL output to your sampling chain.
73
- 4. Add the built-in `MOSS-Image Color/Depth Control` node.
74
- 5. Connect your reference image to `color_map`, leave `depth_map` empty, and connect the Flux2 VAE used by MOSS-Image.
75
- 6. Connect the node's positive and negative conditioning outputs to the sampler.
76
-
77
- For this adapter, `color_map` accepts the actual reference image. **You do not need to convert it into a color-region map or a depth map.** Resize the reference image to your intended output dimensions first.
78
-
79
- ### Strength behavior
80
-
81
- The control node's `strength` scales image conditioning. It does not disable the Tile LoRA already applied to the model. Setting control strength to 0 can therefore produce a different result from running without beta 2.
82
-
83
- For a comparison with beta 2 fully disabled, bypass both its LoRA loader and its control-conditioning node. The LoRA strength slider should not be treated as equivalent to beta 3's residual strength control.
84
-
85
- beta 2 targets the Base model family. Results on SFT should be evaluated separately.
86
-
87
- ## beta 3: Finer strength adjustment
88
-
89
- beta 3 uses an independent residual control branch. Its `strength` parameter scales the control residuals across layers, letting you adjust how closely generation follows the reference image.
90
-
91
- It does not merge Tile weights into the backbone as a LoRA or replace the model's input projection.
92
-
93
- ### Setup
94
-
95
- 1. Install the accompanying `ComfyUI-Moss-Tile-Control` directory under `ComfyUI/custom_nodes/`, then restart ComfyUI.
96
- 2. Place beta 3 weights in `ComfyUI/models/model_patches/`. Version and epoch subfolders can help keep checkpoints organized.
97
- 3. Select the weights with `Load MOSS Tile Control`.
98
- 4. Add `Apply MOSS Tile Control` and connect the MOSS-Image MODEL, control weights, Flux2 VAE, and reference image.
99
- 5. Connect its MODEL output to the sampling chain. Connect positive and negative text conditioning as in a normal MOSS-Image workflow.
100
-
101
- Load beta 3 with its dedicated nodes. It cannot be applied through a standard LoRA loader or the built-in `MOSS-Image Color/Depth Control` node. Remove any older beta 2 Tile LoRA from the model chain when using beta 3.
102
-
103
- ### Parameters
104
-
105
- | Parameter | Description |
106
- | --- | --- |
107
- | `strength` | Multiplier for the control residuals. Start at `0.5` and adjust for your desired balance of structure preservation and regeneration |
108
- | `start_percent` | Sampling progress at which control begins. Default: `0.0` |
109
- | `end_percent` | Sampling progress at which control ends. Default: `1.0` |
110
-
111
- Try `0 / 0.25 / 0.5 / 0.75 / 1.0` with a fixed seed. Higher strength generally increases the influence of the reference image, while the final result also depends on the prompt, base model, sampler settings, and other LoRAs.
112
-
113
- **Strength is continuously adjustable, but visual changes are not guaranteed to be linear or strictly monotonic.** A strength of 1 applies the control residuals at their nominal scale; it does not lock the output to the input image.
114
-
115
- At `strength = 0`, the node returns the input model directly and skips control-image encoding. In a paired real-workflow test with other settings held constant, the zero-strength output was pixel-identical to bypassing the Tile node. Existing style or acceleration LoRAs on the input model remain in effect.
116
-
117
- ### Epoch 1 or epoch 2?
118
-
119
- Both checkpoints provide effective structural control. In a fixed-seed portrait comparison, their results were similar at strength `0.5`. At strength `1.0`, epoch 2 preserved some structures and local features more closely.
120
-
121
- **Start with epoch 2**, and keep epoch 1 as an alternative. These observations do not establish a universal ranking across all images.
122
-
123
- ## Testing and feedback
124
-
125
- Beyond the initial portrait comparisons, the author completed another round of manual testing across a variety of images and was satisfied with the overall results.
126
-
127
- Evaluation currently relies on generated examples and subjective visual assessment. No quantitative benchmark on a standardized test set is provided. The epoch comparison above refers specifically to the controlled portrait test.
128
-
129
- Feedback from other images and workflows is welcome. Please include the version and epoch, base model, control strength, prompt, seed, sampling settings, and any additional LoRAs so results can be reproduced.
130
-
131
- ## Compatibility and sampling
132
-
133
- - For beta 3, MOSS-Image-SFT is the recommended base model family.
134
- - Actual generations have been tested locally with SFT INT8 and a 15-step acceleration LoRA. This does not establish compatibility with every quantization or acceleration method.
135
- - When investigating color, texture, or structural issues, compare against a BF16 base model with additional style and acceleration LoRAs disabled, using the base model's recommended sampling settings.
136
- - Different checkpoints, base models, and LoRA combinations may need different control strengths.
137
-
138
- ## Known limitations
139
-
140
- - **Color and tonal shifts can occur.** Portrait tests showed warmer skin tones, deeper shadows, and contrast changes. Evaluate the output carefully if preserving the original colors is important.
141
- - **Fine details are regenerated.** Hair, skin, fine lines, and other textures may change. Generated details should not be treated as accurate recovery of missing source information.
142
- - **Results depend on the image and workflow.** Manual testing across varied images has been positive, but systematic benchmark results are not available.
143
- - **Higher strength does not always mean better quality.** Choose a value that balances structural fidelity, natural detail, and the amount of change you want.
144
-
145
- ## Base models and acknowledgments
146
-
147
- - [MOSS-Image-Base](https://huggingface.co/OpenMOSS-Team/MOSS-Image-Base)
148
- - [MOSS-Image-SFT](https://huggingface.co/OpenMOSS-Team/MOSS-Image-SFT)
149
- - [ComfyUI](https://github.com/Comfy-Org/ComfyUI)
150
-
151
- Thanks to the OpenMOSS team and the open-source projects that make this work possible. This is a community-developed Tile control adapter.
152
-
153
- The official MOSS-Image base models use the [Apache-2.0 license](https://huggingface.co/OpenMOSS-Team/MOSS-Image-SFT/blob/main/LICENSE).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ tags:
5
+ - moss-image
6
+ - tile
7
+ - controlnet
8
+ - lllite
9
+ - image-to-image
10
+ - image-restoration
11
+ - comfyui
12
+ - adapter
13
+ base_model:
14
+ - OpenMOSS-Team/MOSS-Image-Base
15
+ - OpenMOSS-Team/MOSS-Image-SFT
16
+ base_model_relation: adapter
17
+ library_name: diffusion-single-file
18
+ ---
19
+
20
+ # MOSS-Image Tile Control — V1
21
+
22
+ **V1 is the recommended final release**, developed through four design iterations. It uses
23
+ MOSS-LLLite to refine image details while preserving subject features and overall
24
+ structure, with good color retention in the author's visual tests. **V1 provides
25
+ the best clarity and detail enhancement among these releases in the author's
26
+ comparisons, making it the author's personal choice for overall quality.**
27
+
28
+ Use it for refinement after upscaling and image restoration guided by an existing
29
+ image. Generated details are plausible reconstructions, not guaranteed recovery
30
+ of the original information.
31
+
32
+ If you would like to support this project: [Ko-fi](https://ko-fi.com/xabsurd).
33
+
34
+ ## Which version should I use?
35
+
36
+ | Version | Best use | Observed behavior | Relative sampling speed | Setup |
37
+ | --- | --- | --- | --- | --- |
38
+ | **V1** | **Recommended for detail refinement and restoration** | Best clarity and detail enhancement in the author's comparisons, with subject-feature preservation and good color retention | Between beta2 and beta3 in the author's workflows | MOSS-LLLite custom nodes; MOSS-Image-SFT |
39
+ | **beta2 — LoRA** | High control strength and minimal changes | Strongest preservation of original features; changes the image least. Preserves existing detail rather than repairing it | Slowest | Built-in control node and LoRA loader; MOSS-Image-Base |
40
+ | **beta3 — residual adapter** | Speed first | Strong structural control, but tends to smooth away high-frequency detail; currently not generally recommended | Fastest | Legacy residual nodes included in this package; MOSS-Image-SFT |
41
+
42
+ These comparisons reflect the author's visual evaluation and workflows, not a
43
+ standardized benchmark. The versions use different control paths and Base/SFT
44
+ families; the same numeric strength does not imply the same control intensity.
45
+
46
+ ## Downloads
47
+
48
+ | File | Load with |
49
+ | --- | --- |
50
+ | [V1/MOSS-LLLite-Tile-V1.safetensors](V1/MOSS-LLLite-Tile-V1.safetensors) | **Load MOSS-LLLite** |
51
+ | [beta2/epoch1/adapter_model.safetensors](beta2/epoch1/adapter_model.safetensors) | **Load LoRA (Model Only)** |
52
+ | [beta3/epoch1/adapter_model.safetensors](beta3/epoch1/adapter_model.safetensors) | **Load MOSS Tile Control** |
53
+ | [beta3/epoch2/adapter_model.safetensors](beta3/epoch2/adapter_model.safetensors) | **Load MOSS Tile Control** |
54
+ | [ComfyUI-MOSS-LLLite.zip](ComfyUI-MOSS-LLLite.zip) | Custom node installation archive |
55
+
56
+ The beta2 and beta3 files are retained unchanged for existing users. V1 is the
57
+ final design established through four iterations of the control approach.
58
+
59
+ ## V1: installation and usage
60
+
61
+ Use a recent ComfyUI with native MOSS-Image, Anima LLLite, and the V3 node API.
62
+ No extra Python packages are required by these custom nodes.
63
+
64
+ 1. Extract [ComfyUI-MOSS-LLLite.zip](ComfyUI-MOSS-LLLite.zip) into
65
+ `ComfyUI/custom_nodes/`, producing `custom_nodes/ComfyUI-MOSS-LLLite/`.
66
+ Alternatively, copy the included [ComfyUI-MOSS-LLLite](ComfyUI-MOSS-LLLite/)
67
+ directory there, or install from [GitHub](https://github.com/X-absurd/ComfyUI-MOSS-LLLite).
68
+ 2. If the old `ComfyUI-Moss-Tile-Control` custom node directory is installed,
69
+ remove it before installing this package to avoid duplicate node IDs. The
70
+ updated package includes the beta3 nodes and preserves existing workflow IDs.
71
+ 3. Place `MOSS-LLLite-Tile-V1.safetensors` in `ComfyUI/models/model_patches/`.
72
+ 4. Restart ComfyUI. Add **Load MOSS-LLLite** and **Apply MOSS-LLLite** from the
73
+ **MOSS-LLLite** category.
74
+ 5. Select V1 in the loader. Connect the MOSS-Image-SFT `MODEL`, controller, and
75
+ reference `IMAGE` to the apply node. Connect its output `MODEL` to the sampler.
76
+ 6. Keep normal text conditioning and VAE connections. The apply node accepts
77
+ RGB images directly and has **no VAE input**. Resize the reference image to the
78
+ intended output dimensions before applying control.
79
+
80
+ | Parameter | Suggested starting point |
81
+ | --- | --- |
82
+ | `strength` | `0.5`–`1.0`; adjust for the desired preservation/refinement balance |
83
+ | `start_percent` | `0.0` |
84
+ | `end_percent` | `1.0` |
85
+
86
+ Strength is continuously adjustable. Higher values generally increase the image's
87
+ influence, but visual changes need not be linear or strictly monotonic. At
88
+ `strength = 0`, the apply node returns the input model directly, without encoding
89
+ the control image or adding control patches. Other LoRAs already applied to the
90
+ input model remain active.
91
+
92
+ **Load MOSS-LLLite** and **Apply MOSS-LLLite** were previously displayed with a
93
+ `Tile` suffix. Their internal IDs are unchanged. The nodes are not task-locked to
94
+ Tile, but other control tasks require appropriately trained MOSS-LLLite weights;
95
+ V1 itself is a Tile checkpoint.
96
+
97
+ ## beta2: maximum preservation
98
+
99
+ Choose beta2 when high control strength and minimal modification matter more than
100
+ detail repair. In the author's tests, it preserves original features most strongly
101
+ and changes the image least, but does not repair fine details. Its behavior is
102
+ closer to a conservative upscaler: strong input preservation leaves less room for
103
+ generative detail refinement. It is also the slowest option in the tested workflows.
104
+
105
+ 1. Put beta2 weights in `ComfyUI/models/loras/` and use **Load LoRA (Model Only)**.
106
+ 2. Use the built-in **MOSS-Image Color/Depth Control** node. Connect the actual
107
+ reference image to `color_map`, leave `depth_map` empty, and use the Flux2 VAE.
108
+ No color-region or depth preprocessing is needed for this Tile adapter.
109
+ 3. Route both positive and negative conditioning through that node to the sampler.
110
+ Use MOSS-Image-Base and match the reference and output dimensions.
111
+
112
+ No additional custom nodes are needed. Control-image strength and LoRA strength
113
+ are separate: setting control strength to zero does **not** remove the LoRA's
114
+ effect on the backbone. Bypass both for a fully disabled beta2 comparison.
115
+
116
+ ## beta3: speed first, legacy option
117
+
118
+ beta3 is the fastest option in the author's workflows and provides strong
119
+ structural control. It tends to smooth high-frequency textures and fine details,
120
+ so V1 is recommended for general restoration.
121
+
122
+ Put beta3 weights in `ComfyUI/models/model_patches/`. Use **Load MOSS Tile Control**
123
+ and **Apply MOSS Tile Control**, included in the current package. Connect the
124
+ MOSS-Image-SFT `MODEL`, controller, Flux2 `VAE`, and reference `IMAGE`. At strength
125
+ zero, control is bypassed. beta3 weights cannot be loaded with the LLLite nodes.
126
+
127
+ ## V1 examples and comparisons
128
+
129
+ These four examples use the released V1 weights. The left column is the
130
+ workflow's upscaled control image, aligned to the output dimensions; the right
131
+ column is the actual V1 output. **Color Match and sharpening were not applied.**
132
+ Images below link to their full-resolution files.
133
+
134
+ | Example | Input control | V1 output | Workflow |
135
+ | --- | --- | --- | --- |
136
+ | Flying cat | ![Flying cat input](examples/v1/flying-cat-control.png) | ![Flying cat V1](examples/v1/flying-cat-v1.png) | [JSON](examples/v1/flying-cat.workflow.json) |
137
+ | Grass photograph | ![Grass photograph input](examples/v1/grass-control.png) | ![Grass photograph V1](examples/v1/grass-v1.png) | [JSON](examples/v1/grass.workflow.json) |
138
+ | Jetpack cat | ![Jetpack cat input](examples/v1/jetpack-cat-control.png) | ![Jetpack cat V1](examples/v1/jetpack-cat-v1.png) | [JSON](examples/v1/jetpack-cat.workflow.json) |
139
+ | Cat photograph | ![Cat photograph input](examples/v1/cat-photo-control.png) | ![Cat photograph V1](examples/v1/cat-photo-v1.png) | [JSON](examples/v1/cat-photo.workflow.json) |
140
+
141
+ Each output PNG also embeds its cleaned ComfyUI workflow: drag it into ComfyUI
142
+ to load the graph. Copy the matching `*-input.png` from [examples/v1](examples/v1/)
143
+ into `ComfyUI/input/`, install V1, then select your installed backbone, text
144
+ encoder, VAE, and acceleration LoRA in the loaders. The separate `*.workflow.json`
145
+ files are UI workflows; `*.api.json` files are API prompts.
146
+
147
+ The recorded setup uses **MOSS-Image-SFT INT8/ConvRot**, the **Qwen3.5-9B BF16**
148
+ text encoder, **Flux2 VAE**, and **Coffmeteor-sft-15step-cfg3** acceleration LoRA
149
+ at strength `1.0`. Sampling: **15 steps, CFG 5, Euler / SGM uniform, denoise 1,
150
+ shift 3, seed 994529790995974**. V1 strength is `1.0`, active from `0` to `1`.
151
+ The acceleration LoRA is separate from V1 and is not included in this repository.
152
+
153
+ The flying-cat and jetpack-cat inputs are enlarged 4× with Lanczos. The two
154
+ photographs are deliberately reduced to 10% and then enlarged 5× to test
155
+ restoration; their original high-resolution sources are included for reproducing
156
+ that preprocessing. Small border adjustments follow the model's 16-pixel grid.
157
+ These are prompt-guided restoration examples, not a standardized benchmark or
158
+ proof that all generated details match the original scene.
159
+
160
+ See [example instructions](examples/v1/README.md) and
161
+ [exact settings](examples/v1/examples.json) for the files and preprocessing.
162
+
163
+ ## Legacy beta2/beta3 examples and workflows
164
+
165
+ The following examples are **legacy beta2/beta3 comparisons**, not V1 results.
166
+ Each comparison shows the input on the left and output on the right.
167
+
168
+ | beta3 · epoch2 | beta2 |
169
+ | --- | --- |
170
+ | ![beta3 comparison](examples/beta3-epoch2-comparison.png) | ![beta2 comparison](examples/beta2-comparison.png) |
171
+ | ![beta3 comparison 2](examples/beta3-epoch2-comparison2.png) | ![beta2 comparison 2](examples/beta2-comparison2.png) |
172
+
173
+ Existing [workflows](workflow/) are retained for beta2 and beta3. Follow the V1
174
+ instructions above or load one of the V1 example workflows; the old graphs use
175
+ different load/apply nodes.
176
+
177
+ ## Compatibility and limitations
178
+
179
+ - V1 uses MOSS-Image-SFT. MOSS-Edit reference-image and cached-reference-KV paths
180
+ are currently unsupported by the LLLite nodes.
181
+ - V1, beta3, beta2, and Anima LLLite weight files are not interchangeable. Remove
182
+ old beta2 concat Tile LoRAs when switching to V1 or beta3.
183
+ - V1 retains colors well in the author's comparisons, but color, contrast, or
184
+ saturation shifts can still occur. **Color Match** can optionally correct the
185
+ output using the original image as the color reference.
186
+ - Fine detail may be invented or changed. Evaluate results when exact identity,
187
+ text, patterns, or faithful document reconstruction matters.
188
+ - The prompt, base model, sampling settings, quantization, and other LoRAs affect
189
+ results. Compare versions with the same image and controlled settings where
190
+ possible; higher strength is not automatically better.
191
+
192
+ ## Development notes and credits
193
+
194
+ See [RESEARCH_REPORT.md](RESEARCH_REPORT.md) for a concise development report in
195
+ Chinese, including the architectural choices, color investigation, and checkpoint
196
+ selection. [SHA256SUMS.txt](SHA256SUMS.txt) records hashes of the distributed weights
197
+ and node ZIP.
198
+
199
+ Built on [MOSS-Image](https://huggingface.co/OpenMOSS-Team/MOSS-Image-SFT),
200
+ [ComfyUI](https://github.com/Comfy-Org/ComfyUI), and ComfyUI's native Anima LLLite
201
+ modules. This is a community-developed control adapter. Inference code attribution
202
+ and GPL-3.0 terms are included in
203
+ [NOTICE](ComfyUI-MOSS-LLLite/NOTICE) and [LICENSE](ComfyUI-MOSS-LLLite/LICENSE);
204
+ model weights have their own applicable terms.
RESEARCH_REPORT.md ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # MOSS-Image Tile Control:开发与实验简报
2
+
3
+ 本项目探索如何为 MOSS-Image 增加连续可调的图像控制,用于放大后的细化和高清化,同时兼顾原始特征、颜色与推理速度。经过四次方案迭代,最终确定并发布 **V1**。beta2、beta3 保留作为不同取舍的历史方案。
4
+
5
+ ## 1. 问题从哪里开始
6
+
7
+ 最初采用图像条件拼接与主干 LoRA 的方案。它可以强力约束画面,但控制图强度归零时,已经加载的 LoRA 仍会改变主干输出。因此,“关掉图像条件”和“完全撤去控制模块”并不等价。
8
+
9
+ 目标随之明确:控制强度应作用于独立的控制路径,强度为零时直接返回输入模型,并跳过控制图编码。这一要求也成为后续实现的验证重点。
10
+
11
+ ## 2. 发布版本与实际取舍
12
+
13
+ | 方案 | 结构 | 实际观察与定位 |
14
+ | --- | --- | --- |
15
+ | beta2 | 图像条件拼接 + 主干 LoRA | 适合高强度控制,原始特征保持最强,画面修改最小;不会修复细节,采样最慢。控制图归零仍不能消除 LoRA 的影响 |
16
+ | beta3 | VAE 潜空间条件 + 独立低秩残差分支 | 解决控制归零时的主干污染问题,控制力度较强、速度最快;但倾向于抹平高频细节,目前不作为通用修复首选 |
17
+ | V1 | RGB 条件编码器 + 图像 token 的 Q/K/V 输入 LLLite 适配器 | 作者人工比较中清晰度与细节强化最好,同时保留主体特征,颜色保持也较好;综合观感为作者个人首选 |
18
+
19
+ 项目共经历四次方案迭代,上表列出保留发布的三个版本及作者在实际工作流中的观察。不同路线还涉及 Base/SFT 底模差异,速度和画质取舍不能全部归因于控制结构;目前没有统一基准测试支持跨模型的严格排名。
20
+
21
+ 作者选择 V1 的主要理由是清晰度与细节强化,而不是单纯追求与输入最接近。beta2 对原始特征约束很强,观感更接近保守的放大模型,生成细节的变化空间较小。beta2 的 LoRA 会改变主干扩散层权重,这是实现上的事实;这种影响可能与细节强化较弱有关,但尚未通过消融实验确立因果关系,底模差异、训练与控制强度也可能影响结果。
22
+
23
+ ## 3. 偏色排查与保色实验
24
+
25
+ 早期实验出现颜色、饱和度和对比度变化。用户在原工作流中确认:VAE 编码再解码没有明显变色,关闭加速 LoRA 也没有明显改善,而关闭控制分支后重新采样仍会改变颜色。这表明底模的重新生成行为本身也是颜色变化的来源,不能把所有偏色都归因于控制分支。
26
+
27
+ 项目尝试过从已有残差检查点出发的保色微调和更高学习率实验。改进没有稳定出现在所有检查点上,说明继续训练或加大颜色约束并不必然改善整体观感。最终仍以实际比较结果选模型;Color Match 作为可选后处理保留,而不是把保色问题视为已经彻底解决。
28
+
29
+ ## 4. LLLite 适配、训练与模型选择
30
+
31
+ LLLite 路线复用了 ComfyUI 原生 Anima LLLite 的条件编码器和适配模块,并适配到 MOSS 的 token 排布、联合注意力与层数。控制作用于图像 token 的 Q/K/V 投影输入。这里对齐的是模块计算与注入位置,并不意味着两个底模或训练结果相同。
32
+
33
+ 同权重模块数值核对中,条件编码器输出逐位一致,Q/K/V 输入控制路径的最大绝对误差约为 `1.2e-6`,处于浮点运算误差范围。这项核对不代表端到端生成质量相同。
34
+
35
+ LLLite 路线的早期结果更清晰,但颜色、风格和细节表现一度不及预期。继续训练后,实际观感明显改善;结合多轮人工比较,最终确定 V1。后期结果改善支持“早期训练尚不充分”的解释,但还不足以证明普遍欠拟合,也不意味着训练步数越高必然越好。
36
+
37
+ ## 5. 推理工程与验证
38
+
39
+ 节点通过 ComfyUI 的 ModelPatcher、附加模型加载和 block replacement 接口挂载控制,保留原有显存管理机制。强度零直接绕过控制,不向底模合并控制权重。节点显示名统一为 `Load MOSS-LLLite` / `Apply MOSS-LLLite`,内部 ID 保留,兼容旧工作流。
40
+
41
+ 推理优化将融合 QKV 的三次完整投影改为各自只计算所需的输出行,并复用原生权重转换和量化路径。在一次 1024 分辨率、2 步、CFG 3、INT8/ConvRot 的预热后短采样测试中,约从 `1.94 s` 降至 `1.52 s`,最终 latent 逐位一致。该结果仅说明这项局部优化在所测设置中有效,不代表所有工作流的固定加速比。
42
+
43
+ 另一个减少 RGB 编码批次计算的尝试因改变 BF16 数值结果而未采用。性能优化以保持生成结果为约束,而不是只比较运行时间。
44
+
45
+ ## 6. 结论与边界
46
+
47
+ 最终形成三个明确取舍:**beta2 重保留,beta3 重速度,V1 重细化与综合观感**。V1 是经过四次方案迭代及多轮人工测试确定的最终方案,也是作者基于清晰度、细节强化、特征与颜��保持的个人综合首选。
48
+
49
+ 目前的证据包括模块数值核对、节点行为测试、局部性能测试及人工出图比较。尚缺少统一测试集上的感知质量、颜色误差、特征一致性与多随机种子统计;也未通过完整文献比较确立方法新颖性。本报告是一份工程探索记录,不把主观评价包装成严格学术结论。
SHA256SUMS.txt ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ 21856898240e6f21e2aee8e856726948cedfde35a8d899c11b05c2105d4af4b9 beta2/epoch1/adapter_model.safetensors
2
+ 45d81afa8b5afbac9b8813106d195d9778013c7b10f8c535db80e7da96eacd0c beta3/epoch1/adapter_model.safetensors
3
+ e2d4de5d18cba07cd94d999c0d60ad6849676991a996f69f7a9b3b2b192164d8 beta3/epoch2/adapter_model.safetensors
4
+ 1dc931f3f5533fd96768c64886d87537e7bd19f4f16c1893fb40a459758ae601 V1/MOSS-LLLite-Tile-V1.safetensors
5
+ 7c2fa276e04b576ba2bb46c114ea8aef6f86fa7435fc300864d391b95d7a3999 ComfyUI-MOSS-LLLite.zip
V1/MOSS-LLLite-Tile-V1.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1dc931f3f5533fd96768c64886d87537e7bd19f4f16c1893fb40a459758ae601
3
+ size 103462080
examples/beta3-epoch2_1.png ADDED

Git LFS Details

  • SHA256: 4a3e730ea45a320cc186dc977c36774368e561a5a935c46f71096fdd1d3848f7
  • Pointer size: 132 Bytes
  • Size of remote file: 3.66 MB
examples/beta3-epoch2_2.png ADDED

Git LFS Details

  • SHA256: e3822f45a3660de28f561ddae17ccbd77a23079aa482d42dc9ac899d46ff08eb
  • Pointer size: 132 Bytes
  • Size of remote file: 1.22 MB
examples/input1.png ADDED

Git LFS Details

  • SHA256: bdc2426c382d0105dfb87b395ad085e2e1f8b0113a50ead4a242761c51a50e57
  • Pointer size: 132 Bytes
  • Size of remote file: 1.79 MB
examples/input2.png ADDED

Git LFS Details

  • SHA256: 9a3eaaed600972d1c9a28554aab92f52377b710bb05b4ded51ee9050e92408ed
  • Pointer size: 132 Bytes
  • Size of remote file: 1.01 MB
examples/v1/README.md ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # V1 examples and workflows
2
+
3
+ All four examples use the released **V1** weights. Outputs retain the original
4
+ generated pixels. Their PNG metadata embeds a cleaned ComfyUI workflow using the
5
+ public V1 filename. No Color Match or sharpening was applied.
6
+
7
+ | Example | Original input | Upscaled control preview | V1 output | UI workflow | API prompt |
8
+ | --- | --- | --- | --- | --- | --- |
9
+ | Flying cat | [PNG](flying-cat-input.png) | [PNG](flying-cat-control.png) | [PNG](flying-cat-v1.png) | [JSON](flying-cat.workflow.json) | [JSON](flying-cat.api.json) |
10
+ | Grass photograph | [PNG](grass-input.png) | [PNG](grass-control.png) | [PNG](grass-v1.png) | [JSON](grass.workflow.json) | [JSON](grass.api.json) |
11
+ | Jetpack cat | [PNG](jetpack-cat-input.png) | [PNG](jetpack-cat-control.png) | [PNG](jetpack-cat-v1.png) | [JSON](jetpack-cat.workflow.json) | [JSON](jetpack-cat.api.json) |
12
+ | Cat photograph | [PNG](cat-photo-input.png) | [PNG](cat-photo-control.png) | [PNG](cat-photo-v1.png) | [JSON](cat-photo.workflow.json) | [JSON](cat-photo.api.json) |
13
+
14
+ ## Run an example
15
+
16
+ 1. Install the accompanying `ComfyUI-MOSS-LLLite` nodes and restart ComfyUI.
17
+ 2. Copy `MOSS-LLLite-Tile-V1.safetensors` to `ComfyUI/models/model_patches/`.
18
+ 3. Copy the chosen `*-input.png` into `ComfyUI/input/`. Do not use the control
19
+ preview as the source: the workflow already performs the recorded resizing.
20
+ 4. Drag the corresponding `*-v1.png` or `*.workflow.json` onto the ComfyUI canvas.
21
+ 5. Select the installed files in the model loaders. These examples use:
22
+ - `moss_image_sft_int8_convrot.safetensors` — diffusion model;
23
+ - `qwen3.5_9b_bf16.safetensors` — text encoder, type `moss_image`;
24
+ - `flux2-vae.safetensors` — VAE;
25
+ - `Coffmeteor-sft-15step-cfg3.safetensors` — acceleration LoRA, strength `1`;
26
+ - `MOSS-LLLite-Tile-V1.safetensors` — controller.
27
+ 6. Queue the workflow. The additional models above are not bundled with this
28
+ repository. Replacing them or changing prompts/settings can change the result.
29
+
30
+ Only core ComfyUI nodes and MOSS-LLLite nodes are used. Inactive captioning,
31
+ comparison, color-matching, and unrelated control nodes have been removed from
32
+ the shared graphs. `*.api.json` files are for the ComfyUI API, not canvas import.
33
+
34
+ ## Recorded settings
35
+
36
+ - Euler sampler, SGM uniform schedule, 15 steps, CFG `5`, denoise `1`, shift `3`.
37
+ - Seed `994529790995974`; V1 strength `1`, start `0`, end `1`.
38
+ - Flying cat and jetpack cat: Lanczos enlargement by `4`.
39
+ - Grass and cat photographs: Lanczos reduction to `0.1`, followed by enlargement
40
+ by `5`. These are deliberately degraded restoration tests, not enlargement of
41
+ the full-resolution photograph. Original input pixels are included as PNGs.
42
+ - VAE dimensions align to a 16-pixel grid; control previews are aligned to the
43
+ resulting output size using the node's resize method.
44
+
45
+ The control preview reconstructs preprocessing for side-by-side viewing; it is
46
+ not an additional model output. The V1 PNGs contain the actual generated pixels.
47
+ Source-image EXIF and unrelated workflow metadata are omitted.
48
+
49
+ See [examples.json](examples.json) for exact dimensions, parameters, and output
50
+ pixel hashes. Results are visual examples rather than a standardized benchmark.
examples/v1/SHA256SUMS.txt ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ d1775d8a265f20432851088aa3791655f7705fcd5017d70f03677eb13ba80d37 README.md
2
+ 3c2dd83485dfd0a1f7cb19150ae05c22f631fa2dc4c3be5abe8bb2aa8516c140 cat-photo-control.png
3
+ 587b6792707324db5773b4ae98b4567a68341736298b2f10025973c650d439f0 cat-photo-input.png
4
+ eb80199b8ecbc80682dca2a6146d57cf2c6a02d050f2b9ecf15e535ffe9a9110 cat-photo-v1.png
5
+ bfda1091bd612418ec6cfd7bc2a27537885af72dd0fee9d418d81a38947910ac cat-photo.api.json
6
+ ebb06f9e8092ec168ddfa1a8ad4594baaad9b3b6fd78e29e4f2f75929520c720 cat-photo.workflow.json
7
+ 8d5ecb67ac54e875e887a020cb2f4c2011327db554828e3c3c2a9ac0953b420f examples.json
8
+ 65bd1da0d39890fadd7d26776613c8c3dd9beee0023bdbfdb7f1ab5ae2201ed2 flying-cat-control.png
9
+ bf4b113061bb642c978e84fdd064f6ff14995d0c8ff5698e490d34b6fa632626 flying-cat-input.png
10
+ 845067155f577bc080566ebfe29786467a1225f61066a3ed488b2b77bd2eb7a2 flying-cat-v1.png
11
+ ee354deec134cf855003c5a8b9126392f4a91d7e1019b2bd367adbab5824e0f6 flying-cat.api.json
12
+ 259be58ec68f4ba88894505c29469277c63dff2b322098f96cde2a759927f133 flying-cat.workflow.json
13
+ 14d3a2c02e67720949d41539d6db42e59c7595eb19a102da6fa4af8ba775ff76 grass-control.png
14
+ ac406ef7c8a3410bd199901366e83430a7aa6c49a82318e4e1b8d95b1b0b6585 grass-input.png
15
+ c967a858169b2a8310990e708d03bb4c372475272af666d62d5aff3a276bfc8d grass-v1.png
16
+ 55aea9aa3b2ffc5f04408fd6ed414c8d8ca4fb9fc7ee0f2ec62d46d3569bfb13 grass.api.json
17
+ 2464b8d8be2ed2f0176e945e49b1543b1f859f92effcb99e23669690427d69a1 grass.workflow.json
18
+ 0ad6ce18316831a1908a66afc4f82a0b273e714b40c32505547dd70ce489d70a jetpack-cat-control.png
19
+ 870075c52ec12b031794e0018a43aa0df74fe2ce4000b121ab0fc7faf2db1a93 jetpack-cat-input.png
20
+ 3045fed4dbe7d1f4256bc6cb21f1beff387990b636bba0cb2694cca82979f090 jetpack-cat-v1.png
21
+ 98d461f700d09861ea6ef43896f49447a4bda4bf4042b27e03d703ed1f153338 jetpack-cat.api.json
22
+ 169a8ac635b8a574419f4c8231a2b03a95cbdc001d1635c6d5584eee3f5f64df jetpack-cat.workflow.json
examples/v1/cat-photo-control.png ADDED

Git LFS Details

  • SHA256: 3c2dd83485dfd0a1f7cb19150ae05c22f631fa2dc4c3be5abe8bb2aa8516c140
  • Pointer size: 132 Bytes
  • Size of remote file: 1.5 MB
examples/v1/cat-photo-input.png ADDED

Git LFS Details

  • SHA256: 587b6792707324db5773b4ae98b4567a68341736298b2f10025973c650d439f0
  • Pointer size: 132 Bytes
  • Size of remote file: 8.61 MB
examples/v1/cat-photo-v1.png ADDED

Git LFS Details

  • SHA256: eb80199b8ecbc80682dca2a6146d57cf2c6a02d050f2b9ecf15e535ffe9a9110
  • Pointer size: 132 Bytes
  • Size of remote file: 4.07 MB
examples/v1/cat-photo.api.json ADDED
@@ -0,0 +1,205 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "1": {
3
+ "class_type": "UNETLoader",
4
+ "inputs": {
5
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
6
+ "weight_dtype": "default"
7
+ }
8
+ },
9
+ "3": {
10
+ "class_type": "CLIPLoader",
11
+ "inputs": {
12
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
13
+ "type": "moss_image",
14
+ "device": "default"
15
+ }
16
+ },
17
+ "4": {
18
+ "class_type": "VAELoader",
19
+ "inputs": {
20
+ "vae_name": "flux2-vae.safetensors"
21
+ }
22
+ },
23
+ "5": {
24
+ "class_type": "CLIPTextEncode",
25
+ "inputs": {
26
+ "text": "一只猫咪趴在纸壳子上",
27
+ "clip": [
28
+ "3",
29
+ 0
30
+ ]
31
+ }
32
+ },
33
+ "6": {
34
+ "class_type": "CLIPTextEncode",
35
+ "inputs": {
36
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact",
37
+ "clip": [
38
+ "3",
39
+ 0
40
+ ]
41
+ }
42
+ },
43
+ "9": {
44
+ "class_type": "VAEDecode",
45
+ "inputs": {
46
+ "samples": [
47
+ "20",
48
+ 0
49
+ ],
50
+ "vae": [
51
+ "4",
52
+ 0
53
+ ]
54
+ }
55
+ },
56
+ "20": {
57
+ "class_type": "SamplerCustom",
58
+ "inputs": {
59
+ "add_noise": true,
60
+ "noise_seed": 994529790995974,
61
+ "cfg": 5.0,
62
+ "model": [
63
+ "130",
64
+ 0
65
+ ],
66
+ "positive": [
67
+ "5",
68
+ 0
69
+ ],
70
+ "negative": [
71
+ "6",
72
+ 0
73
+ ],
74
+ "sampler": [
75
+ "86",
76
+ 0
77
+ ],
78
+ "sigmas": [
79
+ "22",
80
+ 0
81
+ ],
82
+ "latent_image": [
83
+ "124",
84
+ 0
85
+ ]
86
+ }
87
+ },
88
+ "22": {
89
+ "class_type": "BasicScheduler",
90
+ "inputs": {
91
+ "scheduler": "sgm_uniform",
92
+ "steps": 15,
93
+ "denoise": 1.0,
94
+ "model": [
95
+ "130",
96
+ 0
97
+ ]
98
+ }
99
+ },
100
+ "36": {
101
+ "class_type": "ModelSamplingAuraFlow",
102
+ "inputs": {
103
+ "shift": 3.0,
104
+ "sampling": "flow",
105
+ "model": [
106
+ "88",
107
+ 0
108
+ ]
109
+ }
110
+ },
111
+ "55": {
112
+ "class_type": "LoadImage",
113
+ "inputs": {
114
+ "image": "cat-photo-input.png"
115
+ }
116
+ },
117
+ "85": {
118
+ "class_type": "SaveImage",
119
+ "inputs": {
120
+ "filename_prefix": "MOSS-LLLite-V1/cat-photo",
121
+ "images": [
122
+ "9",
123
+ 0
124
+ ]
125
+ }
126
+ },
127
+ "86": {
128
+ "class_type": "KSamplerSelect",
129
+ "inputs": {
130
+ "sampler_name": "euler"
131
+ }
132
+ },
133
+ "87": {
134
+ "class_type": "ImageScaleBy",
135
+ "inputs": {
136
+ "upscale_method": "lanczos",
137
+ "scale_by": 0.1,
138
+ "image": [
139
+ "55",
140
+ 0
141
+ ]
142
+ }
143
+ },
144
+ "88": {
145
+ "class_type": "LoraLoaderModelOnly",
146
+ "inputs": {
147
+ "lora_name": "Coffmeteor-sft-15step-cfg3.safetensors",
148
+ "strength_model": 1.0,
149
+ "model": [
150
+ "1",
151
+ 0
152
+ ]
153
+ }
154
+ },
155
+ "90": {
156
+ "class_type": "ImageScaleBy",
157
+ "inputs": {
158
+ "upscale_method": "lanczos",
159
+ "scale_by": 5.0,
160
+ "image": [
161
+ "87",
162
+ 0
163
+ ]
164
+ }
165
+ },
166
+ "124": {
167
+ "class_type": "VAEEncode",
168
+ "inputs": {
169
+ "pixels": [
170
+ "90",
171
+ 0
172
+ ],
173
+ "vae": [
174
+ "4",
175
+ 0
176
+ ]
177
+ }
178
+ },
179
+ "130": {
180
+ "class_type": "MossLLLiteApply",
181
+ "inputs": {
182
+ "strength": 1.0,
183
+ "start_percent": 0.0,
184
+ "end_percent": 1.0,
185
+ "model": [
186
+ "36",
187
+ 0
188
+ ],
189
+ "control": [
190
+ "131",
191
+ 0
192
+ ],
193
+ "image": [
194
+ "90",
195
+ 0
196
+ ]
197
+ }
198
+ },
199
+ "131": {
200
+ "class_type": "MossLLLiteLoader",
201
+ "inputs": {
202
+ "name": "MOSS-LLLite-Tile-V1.safetensors"
203
+ }
204
+ }
205
+ }
examples/v1/cat-photo.workflow.json ADDED
@@ -0,0 +1,1027 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "id": "b22f9ce3-c288-4bd8-a11f-829121f23835",
3
+ "revision": 0,
4
+ "last_node_id": 131,
5
+ "last_link_id": 21,
6
+ "nodes": [
7
+ {
8
+ "id": 1,
9
+ "type": "UNETLoader",
10
+ "pos": [
11
+ -423.26391629976365,
12
+ -689.4382881990866
13
+ ],
14
+ "size": [
15
+ 500,
16
+ 90
17
+ ],
18
+ "flags": {},
19
+ "order": 0,
20
+ "mode": 0,
21
+ "inputs": [],
22
+ "outputs": [
23
+ {
24
+ "name": "MODEL",
25
+ "type": "MODEL",
26
+ "slot_index": 0,
27
+ "links": [
28
+ 15
29
+ ]
30
+ }
31
+ ],
32
+ "properties": {
33
+ "cnr_id": "comfy-core",
34
+ "ver": "0.44.0",
35
+ "Node name for S&R": "UNETLoader",
36
+ "models": [
37
+ {
38
+ "name": "moss_image_sft_int8_convrot.safetensors",
39
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors",
40
+ "directory": "diffusion_models"
41
+ }
42
+ ]
43
+ },
44
+ "widgets_values": [
45
+ "moss-image/moss_image_sft_int8_convrot.safetensors",
46
+ "default"
47
+ ],
48
+ "widgets_values_named": {
49
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
50
+ "weight_dtype": "default"
51
+ }
52
+ },
53
+ {
54
+ "id": 3,
55
+ "type": "CLIPLoader",
56
+ "pos": [
57
+ -440,
58
+ -50
59
+ ],
60
+ "size": [
61
+ 500,
62
+ 110
63
+ ],
64
+ "flags": {},
65
+ "order": 1,
66
+ "mode": 0,
67
+ "inputs": [],
68
+ "outputs": [
69
+ {
70
+ "name": "CLIP",
71
+ "type": "CLIP",
72
+ "slot_index": 0,
73
+ "links": [
74
+ 1,
75
+ 2
76
+ ]
77
+ }
78
+ ],
79
+ "properties": {
80
+ "cnr_id": "comfy-core",
81
+ "ver": "0.44.0",
82
+ "Node name for S&R": "CLIPLoader",
83
+ "models": [
84
+ {
85
+ "name": "qwen3.5_9b_int8_convrot.safetensors",
86
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors",
87
+ "directory": "text_encoders"
88
+ }
89
+ ]
90
+ },
91
+ "widgets_values": [
92
+ "qwen3.5_9b_bf16.safetensors",
93
+ "moss_image",
94
+ "default"
95
+ ],
96
+ "widgets_values_named": {
97
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
98
+ "type": "moss_image",
99
+ "device": "default"
100
+ }
101
+ },
102
+ {
103
+ "id": 4,
104
+ "type": "VAELoader",
105
+ "pos": [
106
+ -440,
107
+ 130
108
+ ],
109
+ "size": [
110
+ 500,
111
+ 60
112
+ ],
113
+ "flags": {},
114
+ "order": 2,
115
+ "mode": 0,
116
+ "inputs": [],
117
+ "outputs": [
118
+ {
119
+ "name": "VAE",
120
+ "type": "VAE",
121
+ "slot_index": 0,
122
+ "links": [
123
+ 4,
124
+ 18
125
+ ]
126
+ }
127
+ ],
128
+ "properties": {
129
+ "cnr_id": "comfy-core",
130
+ "ver": "0.44.0",
131
+ "Node name for S&R": "VAELoader",
132
+ "models": [
133
+ {
134
+ "name": "flux2-vae_small.safetensors",
135
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors",
136
+ "directory": "vae"
137
+ }
138
+ ]
139
+ },
140
+ "widgets_values": [
141
+ "flux2-vae.safetensors"
142
+ ],
143
+ "widgets_values_named": {
144
+ "vae_name": "flux2-vae.safetensors"
145
+ }
146
+ },
147
+ {
148
+ "id": 5,
149
+ "type": "CLIPTextEncode",
150
+ "pos": [
151
+ 185.95627952212953,
152
+ -36.734098034606404
153
+ ],
154
+ "size": [
155
+ 470,
156
+ 240
157
+ ],
158
+ "flags": {},
159
+ "order": 3,
160
+ "mode": 0,
161
+ "inputs": [
162
+ {
163
+ "name": "clip",
164
+ "type": "CLIP",
165
+ "link": 1
166
+ }
167
+ ],
168
+ "outputs": [
169
+ {
170
+ "name": "CONDITIONING",
171
+ "type": "CONDITIONING",
172
+ "slot_index": 0,
173
+ "links": [
174
+ 6
175
+ ]
176
+ }
177
+ ],
178
+ "properties": {
179
+ "cnr_id": "comfy-core",
180
+ "ver": "0.44.0",
181
+ "Node name for S&R": "CLIPTextEncode"
182
+ },
183
+ "widgets_values": [
184
+ "一只猫咪趴在纸壳子上"
185
+ ],
186
+ "widgets_values_named": {
187
+ "text": "一只猫咪趴在纸壳子上"
188
+ },
189
+ "color": "#232",
190
+ "bgcolor": "#353"
191
+ },
192
+ {
193
+ "id": 6,
194
+ "type": "CLIPTextEncode",
195
+ "pos": [
196
+ 150,
197
+ 250
198
+ ],
199
+ "size": [
200
+ 480,
201
+ 110
202
+ ],
203
+ "flags": {},
204
+ "order": 4,
205
+ "mode": 0,
206
+ "inputs": [
207
+ {
208
+ "name": "clip",
209
+ "type": "CLIP",
210
+ "link": 2
211
+ }
212
+ ],
213
+ "outputs": [
214
+ {
215
+ "name": "CONDITIONING",
216
+ "type": "CONDITIONING",
217
+ "slot_index": 0,
218
+ "links": [
219
+ 7
220
+ ]
221
+ }
222
+ ],
223
+ "properties": {
224
+ "cnr_id": "comfy-core",
225
+ "ver": "0.44.0",
226
+ "Node name for S&R": "CLIPTextEncode"
227
+ },
228
+ "widgets_values": [
229
+ "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
230
+ ],
231
+ "widgets_values_named": {
232
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
233
+ },
234
+ "color": "#223",
235
+ "bgcolor": "#335"
236
+ },
237
+ {
238
+ "id": 9,
239
+ "type": "VAEDecode",
240
+ "pos": [
241
+ 1910,
242
+ -520
243
+ ],
244
+ "size": [
245
+ 210,
246
+ 50
247
+ ],
248
+ "flags": {},
249
+ "order": 5,
250
+ "mode": 0,
251
+ "inputs": [
252
+ {
253
+ "name": "samples",
254
+ "type": "LATENT",
255
+ "link": 3
256
+ },
257
+ {
258
+ "name": "vae",
259
+ "type": "VAE",
260
+ "link": 4
261
+ }
262
+ ],
263
+ "outputs": [
264
+ {
265
+ "name": "IMAGE",
266
+ "type": "IMAGE",
267
+ "slot_index": 0,
268
+ "links": [
269
+ 13
270
+ ]
271
+ }
272
+ ],
273
+ "properties": {
274
+ "cnr_id": "comfy-core",
275
+ "ver": "0.44.0",
276
+ "Node name for S&R": "VAEDecode"
277
+ },
278
+ "color": "#322",
279
+ "bgcolor": "#533"
280
+ },
281
+ {
282
+ "id": 20,
283
+ "type": "SamplerCustom",
284
+ "pos": [
285
+ 1518.3539918681,
286
+ -304.93798438717295
287
+ ],
288
+ "size": [
289
+ 280,
290
+ 500
291
+ ],
292
+ "flags": {},
293
+ "order": 6,
294
+ "mode": 0,
295
+ "inputs": [
296
+ {
297
+ "name": "model",
298
+ "type": "MODEL",
299
+ "link": 5
300
+ },
301
+ {
302
+ "name": "positive",
303
+ "type": "CONDITIONING",
304
+ "link": 6
305
+ },
306
+ {
307
+ "name": "negative",
308
+ "type": "CONDITIONING",
309
+ "link": 7
310
+ },
311
+ {
312
+ "name": "sampler",
313
+ "type": "SAMPLER",
314
+ "link": 8
315
+ },
316
+ {
317
+ "name": "sigmas",
318
+ "type": "SIGMAS",
319
+ "link": 9
320
+ },
321
+ {
322
+ "name": "latent_image",
323
+ "type": "LATENT",
324
+ "link": 10
325
+ }
326
+ ],
327
+ "outputs": [
328
+ {
329
+ "name": "output",
330
+ "type": "LATENT",
331
+ "links": [
332
+ 3
333
+ ]
334
+ },
335
+ {
336
+ "name": "denoised_output",
337
+ "type": "LATENT",
338
+ "links": null
339
+ }
340
+ ],
341
+ "properties": {
342
+ "cnr_id": "comfy-core",
343
+ "ver": "0.44.0",
344
+ "Node name for S&R": "SamplerCustom"
345
+ },
346
+ "widgets_values": [
347
+ true,
348
+ 994529790995974,
349
+ "fixed",
350
+ 5
351
+ ],
352
+ "widgets_values_named": {
353
+ "add_noise": true,
354
+ "noise_seed": 994529790995974,
355
+ "control_after_generate": "fixed",
356
+ "cfg": 5.0
357
+ }
358
+ },
359
+ {
360
+ "id": 22,
361
+ "type": "BasicScheduler",
362
+ "pos": [
363
+ 1169.6140490851362,
364
+ 1.7699193319720672
365
+ ],
366
+ "size": [
367
+ 270,
368
+ 110
369
+ ],
370
+ "flags": {},
371
+ "order": 7,
372
+ "mode": 0,
373
+ "inputs": [
374
+ {
375
+ "name": "model",
376
+ "type": "MODEL",
377
+ "link": 11
378
+ }
379
+ ],
380
+ "outputs": [
381
+ {
382
+ "name": "SIGMAS",
383
+ "type": "SIGMAS",
384
+ "links": [
385
+ 9
386
+ ]
387
+ }
388
+ ],
389
+ "properties": {
390
+ "cnr_id": "comfy-core",
391
+ "ver": "0.44.0",
392
+ "Node name for S&R": "BasicScheduler"
393
+ },
394
+ "widgets_values": [
395
+ "sgm_uniform",
396
+ 15,
397
+ 1
398
+ ],
399
+ "widgets_values_named": {
400
+ "scheduler": "sgm_uniform",
401
+ "steps": 15,
402
+ "denoise": 1.0
403
+ }
404
+ },
405
+ {
406
+ "id": 36,
407
+ "type": "ModelSamplingAuraFlow",
408
+ "pos": [
409
+ 216.72263354772872,
410
+ -300
411
+ ],
412
+ "size": [
413
+ 400,
414
+ 90
415
+ ],
416
+ "flags": {},
417
+ "order": 8,
418
+ "mode": 0,
419
+ "inputs": [
420
+ {
421
+ "name": "model",
422
+ "type": "MODEL",
423
+ "link": 12
424
+ }
425
+ ],
426
+ "outputs": [
427
+ {
428
+ "name": "MODEL",
429
+ "type": "MODEL",
430
+ "links": [
431
+ 19
432
+ ]
433
+ }
434
+ ],
435
+ "properties": {
436
+ "cnr_id": "comfy-core",
437
+ "ver": "0.44.0",
438
+ "Node name for S&R": "ModelSamplingAuraFlow"
439
+ },
440
+ "widgets_values": [
441
+ 3,
442
+ "flow"
443
+ ],
444
+ "widgets_values_named": {
445
+ "shift": 3.0,
446
+ "sampling": "flow"
447
+ }
448
+ },
449
+ {
450
+ "id": 55,
451
+ "type": "LoadImage",
452
+ "pos": [
453
+ -400,
454
+ 630
455
+ ],
456
+ "size": [
457
+ 290,
458
+ 110
459
+ ],
460
+ "flags": {},
461
+ "order": 9,
462
+ "mode": 0,
463
+ "inputs": [],
464
+ "outputs": [
465
+ {
466
+ "name": "IMAGE",
467
+ "type": "IMAGE",
468
+ "links": [
469
+ 14
470
+ ]
471
+ },
472
+ {
473
+ "name": "MASK",
474
+ "type": "MASK",
475
+ "links": null
476
+ }
477
+ ],
478
+ "properties": {
479
+ "cnr_id": "comfy-core",
480
+ "ver": "0.44.0",
481
+ "Node name for S&R": "LoadImage"
482
+ },
483
+ "widgets_values": [
484
+ "cat-photo-input.png",
485
+ "image"
486
+ ],
487
+ "widgets_values_named": {
488
+ "image": "cat-photo-input.png",
489
+ "upload": "image"
490
+ },
491
+ "color": "#322",
492
+ "bgcolor": "#533"
493
+ },
494
+ {
495
+ "id": 85,
496
+ "type": "SaveImage",
497
+ "pos": [
498
+ 3307.317013871774,
499
+ 597.8636595391883
500
+ ],
501
+ "size": [
502
+ 420,
503
+ 470
504
+ ],
505
+ "flags": {},
506
+ "order": 10,
507
+ "mode": 0,
508
+ "inputs": [
509
+ {
510
+ "name": "images",
511
+ "type": "IMAGE",
512
+ "link": 13
513
+ }
514
+ ],
515
+ "outputs": [
516
+ {
517
+ "name": "images",
518
+ "type": "IMAGE",
519
+ "links": null
520
+ }
521
+ ],
522
+ "properties": {
523
+ "cnr_id": "comfy-core",
524
+ "ver": "0.44.0",
525
+ "Node name for S&R": "SaveImage"
526
+ },
527
+ "widgets_values": [
528
+ "MOSS-LLLite-V1/cat-photo"
529
+ ],
530
+ "widgets_values_named": {
531
+ "filename_prefix": "MOSS-LLLite-V1/cat-photo"
532
+ }
533
+ },
534
+ {
535
+ "id": 86,
536
+ "type": "KSamplerSelect",
537
+ "pos": [
538
+ 1170.4423572560272,
539
+ -260.582916386391
540
+ ],
541
+ "size": [
542
+ 270,
543
+ 58
544
+ ],
545
+ "flags": {},
546
+ "order": 11,
547
+ "mode": 0,
548
+ "inputs": [],
549
+ "outputs": [
550
+ {
551
+ "name": "SAMPLER",
552
+ "type": "SAMPLER",
553
+ "links": [
554
+ 8
555
+ ]
556
+ }
557
+ ],
558
+ "properties": {
559
+ "cnr_id": "comfy-core",
560
+ "ver": "0.44.0",
561
+ "Node name for S&R": "KSamplerSelect"
562
+ },
563
+ "widgets_values": [
564
+ "euler"
565
+ ],
566
+ "widgets_values_named": {
567
+ "sampler_name": "euler"
568
+ }
569
+ },
570
+ {
571
+ "id": 87,
572
+ "type": "ImageScaleBy",
573
+ "pos": [
574
+ -7.178545924871845,
575
+ 681.7385897176745
576
+ ],
577
+ "size": [
578
+ 270,
579
+ 82
580
+ ],
581
+ "flags": {},
582
+ "order": 12,
583
+ "mode": 0,
584
+ "inputs": [
585
+ {
586
+ "name": "image",
587
+ "type": "IMAGE",
588
+ "link": 14
589
+ }
590
+ ],
591
+ "outputs": [
592
+ {
593
+ "name": "IMAGE",
594
+ "type": "IMAGE",
595
+ "links": [
596
+ 16
597
+ ]
598
+ }
599
+ ],
600
+ "properties": {
601
+ "cnr_id": "comfy-core",
602
+ "ver": "0.44.0",
603
+ "Node name for S&R": "ImageScaleBy"
604
+ },
605
+ "widgets_values": [
606
+ "lanczos",
607
+ 0.1
608
+ ],
609
+ "widgets_values_named": {
610
+ "upscale_method": "lanczos",
611
+ "scale_by": 0.1
612
+ }
613
+ },
614
+ {
615
+ "id": 88,
616
+ "type": "LoraLoaderModelOnly",
617
+ "pos": [
618
+ -447.13559989481814,
619
+ -263.5408892596166
620
+ ],
621
+ "size": [
622
+ 500,
623
+ 120
624
+ ],
625
+ "flags": {},
626
+ "order": 13,
627
+ "mode": 0,
628
+ "inputs": [
629
+ {
630
+ "name": "model",
631
+ "type": "MODEL",
632
+ "link": 15
633
+ }
634
+ ],
635
+ "outputs": [
636
+ {
637
+ "name": "MODEL",
638
+ "type": "MODEL",
639
+ "links": [
640
+ 12
641
+ ]
642
+ }
643
+ ],
644
+ "properties": {
645
+ "cnr_id": "comfy-core",
646
+ "ver": "0.44.0",
647
+ "Node name for S&R": "LoraLoaderModelOnly",
648
+ "models": [
649
+ {
650
+ "name": "moss_image_color_depth_control.safetensors",
651
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
652
+ "directory": "loras"
653
+ }
654
+ ]
655
+ },
656
+ "widgets_values": [
657
+ "Coffmeteor-sft-15step-cfg3.safetensors",
658
+ 1
659
+ ],
660
+ "widgets_values_named": {
661
+ "lora_name": "Coffmeteor-sft-15step-cfg3.safetensors",
662
+ "strength_model": 1.0
663
+ }
664
+ },
665
+ {
666
+ "id": 90,
667
+ "type": "ImageScaleBy",
668
+ "pos": [
669
+ 321.18555329294736,
670
+ 683.4828563588695
671
+ ],
672
+ "size": [
673
+ 270,
674
+ 82
675
+ ],
676
+ "flags": {},
677
+ "order": 14,
678
+ "mode": 0,
679
+ "inputs": [
680
+ {
681
+ "name": "image",
682
+ "type": "IMAGE",
683
+ "link": 16
684
+ }
685
+ ],
686
+ "outputs": [
687
+ {
688
+ "name": "IMAGE",
689
+ "type": "IMAGE",
690
+ "links": [
691
+ 17,
692
+ 21
693
+ ]
694
+ }
695
+ ],
696
+ "properties": {
697
+ "cnr_id": "comfy-core",
698
+ "ver": "0.44.0",
699
+ "Node name for S&R": "ImageScaleBy"
700
+ },
701
+ "widgets_values": [
702
+ "lanczos",
703
+ 5
704
+ ],
705
+ "widgets_values_named": {
706
+ "upscale_method": "lanczos",
707
+ "scale_by": 5.0
708
+ }
709
+ },
710
+ {
711
+ "id": 124,
712
+ "type": "VAEEncode",
713
+ "pos": [
714
+ 1149.2279511667382,
715
+ 242.2625195138657
716
+ ],
717
+ "size": [
718
+ 140,
719
+ 46
720
+ ],
721
+ "flags": {},
722
+ "order": 15,
723
+ "mode": 0,
724
+ "inputs": [
725
+ {
726
+ "name": "pixels",
727
+ "type": "IMAGE",
728
+ "link": 17
729
+ },
730
+ {
731
+ "name": "vae",
732
+ "type": "VAE",
733
+ "link": 18
734
+ }
735
+ ],
736
+ "outputs": [
737
+ {
738
+ "name": "LATENT",
739
+ "type": "LATENT",
740
+ "links": [
741
+ 10
742
+ ]
743
+ }
744
+ ],
745
+ "properties": {
746
+ "cnr_id": "comfy-core",
747
+ "ver": "0.44.0",
748
+ "Node name for S&R": "VAEEncode"
749
+ }
750
+ },
751
+ {
752
+ "id": 130,
753
+ "type": "MossLLLiteApply",
754
+ "pos": [
755
+ 754.4053011472657,
756
+ 147.5866285701938
757
+ ],
758
+ "size": [
759
+ 270,
760
+ 146
761
+ ],
762
+ "flags": {},
763
+ "order": 16,
764
+ "mode": 0,
765
+ "inputs": [
766
+ {
767
+ "name": "model",
768
+ "type": "MODEL",
769
+ "link": 19
770
+ },
771
+ {
772
+ "name": "control",
773
+ "type": "MOSS_LLLITE_CONTROL",
774
+ "link": 20
775
+ },
776
+ {
777
+ "name": "image",
778
+ "type": "IMAGE",
779
+ "link": 21
780
+ }
781
+ ],
782
+ "outputs": [
783
+ {
784
+ "name": "MODEL",
785
+ "type": "MODEL",
786
+ "links": [
787
+ 5,
788
+ 11
789
+ ]
790
+ }
791
+ ],
792
+ "properties": {
793
+ "aux_id": "X-absurd/ComfyUI-MOSS-LLLite",
794
+ "ver": "20842b2b9d08511abd32404a0be24a51c9352109",
795
+ "Node name for S&R": "MossLLLiteApply"
796
+ },
797
+ "widgets_values": [
798
+ 1,
799
+ 0,
800
+ 1
801
+ ],
802
+ "widgets_values_named": {
803
+ "strength": 1.0,
804
+ "start_percent": 0.0,
805
+ "end_percent": 1.0
806
+ }
807
+ },
808
+ {
809
+ "id": 131,
810
+ "type": "MossLLLiteLoader",
811
+ "pos": [
812
+ 757.1064855291617,
813
+ -12.82076859408437
814
+ ],
815
+ "size": [
816
+ 270,
817
+ 58
818
+ ],
819
+ "flags": {},
820
+ "order": 17,
821
+ "mode": 0,
822
+ "inputs": [],
823
+ "outputs": [
824
+ {
825
+ "name": "MOSS_LLLITE_CONTROL",
826
+ "type": "MOSS_LLLITE_CONTROL",
827
+ "links": [
828
+ 20
829
+ ]
830
+ }
831
+ ],
832
+ "properties": {
833
+ "aux_id": "X-absurd/ComfyUI-MOSS-LLLite",
834
+ "ver": "20842b2b9d08511abd32404a0be24a51c9352109",
835
+ "Node name for S&R": "MossLLLiteLoader"
836
+ },
837
+ "widgets_values": [
838
+ "MOSS-LLLite-Tile-V1.safetensors"
839
+ ],
840
+ "widgets_values_named": {
841
+ "name": "MOSS-LLLite-Tile-V1.safetensors"
842
+ }
843
+ }
844
+ ],
845
+ "links": [
846
+ [
847
+ 1,
848
+ 3,
849
+ 0,
850
+ 5,
851
+ 0,
852
+ "CLIP"
853
+ ],
854
+ [
855
+ 2,
856
+ 3,
857
+ 0,
858
+ 6,
859
+ 0,
860
+ "CLIP"
861
+ ],
862
+ [
863
+ 3,
864
+ 20,
865
+ 0,
866
+ 9,
867
+ 0,
868
+ "LATENT"
869
+ ],
870
+ [
871
+ 4,
872
+ 4,
873
+ 0,
874
+ 9,
875
+ 1,
876
+ "VAE"
877
+ ],
878
+ [
879
+ 5,
880
+ 130,
881
+ 0,
882
+ 20,
883
+ 0,
884
+ "MODEL"
885
+ ],
886
+ [
887
+ 6,
888
+ 5,
889
+ 0,
890
+ 20,
891
+ 1,
892
+ "CONDITIONING"
893
+ ],
894
+ [
895
+ 7,
896
+ 6,
897
+ 0,
898
+ 20,
899
+ 2,
900
+ "CONDITIONING"
901
+ ],
902
+ [
903
+ 8,
904
+ 86,
905
+ 0,
906
+ 20,
907
+ 3,
908
+ "SAMPLER"
909
+ ],
910
+ [
911
+ 9,
912
+ 22,
913
+ 0,
914
+ 20,
915
+ 4,
916
+ "SIGMAS"
917
+ ],
918
+ [
919
+ 10,
920
+ 124,
921
+ 0,
922
+ 20,
923
+ 5,
924
+ "LATENT"
925
+ ],
926
+ [
927
+ 11,
928
+ 130,
929
+ 0,
930
+ 22,
931
+ 0,
932
+ "MODEL"
933
+ ],
934
+ [
935
+ 12,
936
+ 88,
937
+ 0,
938
+ 36,
939
+ 0,
940
+ "MODEL"
941
+ ],
942
+ [
943
+ 13,
944
+ 9,
945
+ 0,
946
+ 85,
947
+ 0,
948
+ "IMAGE"
949
+ ],
950
+ [
951
+ 14,
952
+ 55,
953
+ 0,
954
+ 87,
955
+ 0,
956
+ "IMAGE"
957
+ ],
958
+ [
959
+ 15,
960
+ 1,
961
+ 0,
962
+ 88,
963
+ 0,
964
+ "MODEL"
965
+ ],
966
+ [
967
+ 16,
968
+ 87,
969
+ 0,
970
+ 90,
971
+ 0,
972
+ "IMAGE"
973
+ ],
974
+ [
975
+ 17,
976
+ 90,
977
+ 0,
978
+ 124,
979
+ 0,
980
+ "IMAGE"
981
+ ],
982
+ [
983
+ 18,
984
+ 4,
985
+ 0,
986
+ 124,
987
+ 1,
988
+ "VAE"
989
+ ],
990
+ [
991
+ 19,
992
+ 36,
993
+ 0,
994
+ 130,
995
+ 0,
996
+ "MODEL"
997
+ ],
998
+ [
999
+ 20,
1000
+ 131,
1001
+ 0,
1002
+ 130,
1003
+ 1,
1004
+ "MOSS_LLLITE_CONTROL"
1005
+ ],
1006
+ [
1007
+ 21,
1008
+ 90,
1009
+ 0,
1010
+ 130,
1011
+ 2,
1012
+ "IMAGE"
1013
+ ]
1014
+ ],
1015
+ "groups": [],
1016
+ "config": {},
1017
+ "extra": {
1018
+ "ds": {
1019
+ "scale": 0.55,
1020
+ "offset": [
1021
+ 550,
1022
+ 150
1023
+ ]
1024
+ }
1025
+ },
1026
+ "version": 0.4
1027
+ }
examples/v1/examples.json ADDED
@@ -0,0 +1,109 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "release": "V1",
3
+ "release_weight_sha256": "1dc931f3f5533fd96768c64886d87537e7bd19f4f16c1893fb40a459758ae601",
4
+ "sampling": {
5
+ "steps": 15,
6
+ "cfg": 5,
7
+ "seed": 994529790995974,
8
+ "sampler": "euler",
9
+ "scheduler": "sgm_uniform",
10
+ "denoise": 1.0,
11
+ "shift": 3.0
12
+ },
13
+ "examples": [
14
+ {
15
+ "title": "Flying cat",
16
+ "input": "flying-cat-input.png",
17
+ "control_preview": "flying-cat-control.png",
18
+ "output": "flying-cat-v1.png",
19
+ "workflow": "flying-cat.workflow.json",
20
+ "api": "flying-cat.api.json",
21
+ "output_size": [
22
+ 2192,
23
+ 1456
24
+ ],
25
+ "input_size": [
26
+ 550,
27
+ 366
28
+ ],
29
+ "lanczos_scales": [
30
+ 4.0
31
+ ],
32
+ "control_strength": 1.0,
33
+ "color_match": false,
34
+ "sharpening": false,
35
+ "output_pixel_sha256": "c6137784375070634daf054a9864e8d514396249ae9ad88b3683ad11d4a92ddd"
36
+ },
37
+ {
38
+ "title": "Grass photograph",
39
+ "input": "grass-input.png",
40
+ "control_preview": "grass-control.png",
41
+ "output": "grass-v1.png",
42
+ "workflow": "grass.workflow.json",
43
+ "api": "grass.api.json",
44
+ "output_size": [
45
+ 2000,
46
+ 1488
47
+ ],
48
+ "input_size": [
49
+ 4000,
50
+ 3000
51
+ ],
52
+ "lanczos_scales": [
53
+ 0.1,
54
+ 5.0
55
+ ],
56
+ "control_strength": 1.0,
57
+ "color_match": false,
58
+ "sharpening": false,
59
+ "output_pixel_sha256": "508673b0a5d8af5ddfc042e593099ba76fa98e7b60b64956a64f17eb42a11bc2"
60
+ },
61
+ {
62
+ "title": "Jetpack cat",
63
+ "input": "jetpack-cat-input.png",
64
+ "control_preview": "jetpack-cat-control.png",
65
+ "output": "jetpack-cat-v1.png",
66
+ "workflow": "jetpack-cat.workflow.json",
67
+ "api": "jetpack-cat.api.json",
68
+ "output_size": [
69
+ 2400,
70
+ 1840
71
+ ],
72
+ "input_size": [
73
+ 600,
74
+ 463
75
+ ],
76
+ "lanczos_scales": [
77
+ 4.0
78
+ ],
79
+ "control_strength": 1.0,
80
+ "color_match": false,
81
+ "sharpening": false,
82
+ "output_pixel_sha256": "3cd57da2d3bbf9b85b958cd2fccd0a5280b9c6064c0ac0a6d9246600217443de"
83
+ },
84
+ {
85
+ "title": "Cat photograph",
86
+ "input": "cat-photo-input.png",
87
+ "control_preview": "cat-photo-control.png",
88
+ "output": "cat-photo-v1.png",
89
+ "workflow": "cat-photo.workflow.json",
90
+ "api": "cat-photo.api.json",
91
+ "output_size": [
92
+ 1488,
93
+ 2000
94
+ ],
95
+ "input_size": [
96
+ 3000,
97
+ 4000
98
+ ],
99
+ "lanczos_scales": [
100
+ 0.1,
101
+ 5.0
102
+ ],
103
+ "control_strength": 1.0,
104
+ "color_match": false,
105
+ "sharpening": false,
106
+ "output_pixel_sha256": "4863badb419809814793274fe1c14d6fe208b34d5cf786e898274dfcfa51d4c2"
107
+ }
108
+ ]
109
+ }
examples/v1/flying-cat-control.png ADDED

Git LFS Details

  • SHA256: 65bd1da0d39890fadd7d26776613c8c3dd9beee0023bdbfdb7f1ab5ae2201ed2
  • Pointer size: 132 Bytes
  • Size of remote file: 2.02 MB
examples/v1/flying-cat-input.png ADDED

Git LFS Details

  • SHA256: bf4b113061bb642c978e84fdd064f6ff14995d0c8ff5698e490d34b6fa632626
  • Pointer size: 131 Bytes
  • Size of remote file: 192 kB
examples/v1/flying-cat-v1.png ADDED

Git LFS Details

  • SHA256: 845067155f577bc080566ebfe29786467a1225f61066a3ed488b2b77bd2eb7a2
  • Pointer size: 132 Bytes
  • Size of remote file: 4.14 MB
examples/v1/flying-cat.api.json ADDED
@@ -0,0 +1,194 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "1": {
3
+ "class_type": "UNETLoader",
4
+ "inputs": {
5
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
6
+ "weight_dtype": "default"
7
+ }
8
+ },
9
+ "3": {
10
+ "class_type": "CLIPLoader",
11
+ "inputs": {
12
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
13
+ "type": "moss_image",
14
+ "device": "default"
15
+ }
16
+ },
17
+ "4": {
18
+ "class_type": "VAELoader",
19
+ "inputs": {
20
+ "vae_name": "flux2-vae.safetensors"
21
+ }
22
+ },
23
+ "5": {
24
+ "class_type": "CLIPTextEncode",
25
+ "inputs": {
26
+ "text": "一只毛茸茸的猫,四肢顶端有着无人机的螺旋桨,腹部有着无人机的起落架,真实照片,可以看到根根分明的猫毛",
27
+ "clip": [
28
+ "3",
29
+ 0
30
+ ]
31
+ }
32
+ },
33
+ "6": {
34
+ "class_type": "CLIPTextEncode",
35
+ "inputs": {
36
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact",
37
+ "clip": [
38
+ "3",
39
+ 0
40
+ ]
41
+ }
42
+ },
43
+ "9": {
44
+ "class_type": "VAEDecode",
45
+ "inputs": {
46
+ "samples": [
47
+ "20",
48
+ 0
49
+ ],
50
+ "vae": [
51
+ "4",
52
+ 0
53
+ ]
54
+ }
55
+ },
56
+ "20": {
57
+ "class_type": "SamplerCustom",
58
+ "inputs": {
59
+ "add_noise": true,
60
+ "noise_seed": 994529790995974,
61
+ "cfg": 5.0,
62
+ "model": [
63
+ "130",
64
+ 0
65
+ ],
66
+ "positive": [
67
+ "5",
68
+ 0
69
+ ],
70
+ "negative": [
71
+ "6",
72
+ 0
73
+ ],
74
+ "sampler": [
75
+ "86",
76
+ 0
77
+ ],
78
+ "sigmas": [
79
+ "22",
80
+ 0
81
+ ],
82
+ "latent_image": [
83
+ "124",
84
+ 0
85
+ ]
86
+ }
87
+ },
88
+ "22": {
89
+ "class_type": "BasicScheduler",
90
+ "inputs": {
91
+ "scheduler": "sgm_uniform",
92
+ "steps": 15,
93
+ "denoise": 1.0,
94
+ "model": [
95
+ "130",
96
+ 0
97
+ ]
98
+ }
99
+ },
100
+ "36": {
101
+ "class_type": "ModelSamplingAuraFlow",
102
+ "inputs": {
103
+ "shift": 3.0,
104
+ "sampling": "flow",
105
+ "model": [
106
+ "88",
107
+ 0
108
+ ]
109
+ }
110
+ },
111
+ "55": {
112
+ "class_type": "LoadImage",
113
+ "inputs": {
114
+ "image": "flying-cat-input.png"
115
+ }
116
+ },
117
+ "85": {
118
+ "class_type": "SaveImage",
119
+ "inputs": {
120
+ "filename_prefix": "MOSS-LLLite-V1/flying-cat",
121
+ "images": [
122
+ "9",
123
+ 0
124
+ ]
125
+ }
126
+ },
127
+ "86": {
128
+ "class_type": "KSamplerSelect",
129
+ "inputs": {
130
+ "sampler_name": "euler"
131
+ }
132
+ },
133
+ "88": {
134
+ "class_type": "LoraLoaderModelOnly",
135
+ "inputs": {
136
+ "lora_name": "Coffmeteor-sft-15step-cfg3.safetensors",
137
+ "strength_model": 1.0,
138
+ "model": [
139
+ "1",
140
+ 0
141
+ ]
142
+ }
143
+ },
144
+ "90": {
145
+ "class_type": "ImageScaleBy",
146
+ "inputs": {
147
+ "upscale_method": "lanczos",
148
+ "scale_by": 4.0,
149
+ "image": [
150
+ "55",
151
+ 0
152
+ ]
153
+ }
154
+ },
155
+ "124": {
156
+ "class_type": "VAEEncode",
157
+ "inputs": {
158
+ "pixels": [
159
+ "90",
160
+ 0
161
+ ],
162
+ "vae": [
163
+ "4",
164
+ 0
165
+ ]
166
+ }
167
+ },
168
+ "130": {
169
+ "class_type": "MossLLLiteApply",
170
+ "inputs": {
171
+ "strength": 1.0,
172
+ "start_percent": 0.0,
173
+ "end_percent": 1.0,
174
+ "model": [
175
+ "36",
176
+ 0
177
+ ],
178
+ "control": [
179
+ "131",
180
+ 0
181
+ ],
182
+ "image": [
183
+ "90",
184
+ 0
185
+ ]
186
+ }
187
+ },
188
+ "131": {
189
+ "class_type": "MossLLLiteLoader",
190
+ "inputs": {
191
+ "name": "MOSS-LLLite-Tile-V1.safetensors"
192
+ }
193
+ }
194
+ }
examples/v1/flying-cat.workflow.json ADDED
@@ -0,0 +1,975 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "id": "69a91a2e-466c-4991-9edc-095217115594",
3
+ "revision": 0,
4
+ "last_node_id": 131,
5
+ "last_link_id": 20,
6
+ "nodes": [
7
+ {
8
+ "id": 1,
9
+ "type": "UNETLoader",
10
+ "pos": [
11
+ -423.26391629976365,
12
+ -689.4382881990866
13
+ ],
14
+ "size": [
15
+ 500,
16
+ 90
17
+ ],
18
+ "flags": {},
19
+ "order": 0,
20
+ "mode": 0,
21
+ "inputs": [],
22
+ "outputs": [
23
+ {
24
+ "name": "MODEL",
25
+ "type": "MODEL",
26
+ "slot_index": 0,
27
+ "links": [
28
+ 14
29
+ ]
30
+ }
31
+ ],
32
+ "properties": {
33
+ "cnr_id": "comfy-core",
34
+ "ver": "0.44.0",
35
+ "Node name for S&R": "UNETLoader",
36
+ "models": [
37
+ {
38
+ "name": "moss_image_sft_int8_convrot.safetensors",
39
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors",
40
+ "directory": "diffusion_models"
41
+ }
42
+ ]
43
+ },
44
+ "widgets_values": [
45
+ "moss-image/moss_image_sft_int8_convrot.safetensors",
46
+ "default"
47
+ ],
48
+ "widgets_values_named": {
49
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
50
+ "weight_dtype": "default"
51
+ }
52
+ },
53
+ {
54
+ "id": 3,
55
+ "type": "CLIPLoader",
56
+ "pos": [
57
+ -440,
58
+ -50
59
+ ],
60
+ "size": [
61
+ 500,
62
+ 110
63
+ ],
64
+ "flags": {},
65
+ "order": 1,
66
+ "mode": 0,
67
+ "inputs": [],
68
+ "outputs": [
69
+ {
70
+ "name": "CLIP",
71
+ "type": "CLIP",
72
+ "slot_index": 0,
73
+ "links": [
74
+ 1,
75
+ 2
76
+ ]
77
+ }
78
+ ],
79
+ "properties": {
80
+ "cnr_id": "comfy-core",
81
+ "ver": "0.44.0",
82
+ "Node name for S&R": "CLIPLoader",
83
+ "models": [
84
+ {
85
+ "name": "qwen3.5_9b_int8_convrot.safetensors",
86
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors",
87
+ "directory": "text_encoders"
88
+ }
89
+ ]
90
+ },
91
+ "widgets_values": [
92
+ "qwen3.5_9b_bf16.safetensors",
93
+ "moss_image",
94
+ "default"
95
+ ],
96
+ "widgets_values_named": {
97
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
98
+ "type": "moss_image",
99
+ "device": "default"
100
+ }
101
+ },
102
+ {
103
+ "id": 4,
104
+ "type": "VAELoader",
105
+ "pos": [
106
+ -440,
107
+ 130
108
+ ],
109
+ "size": [
110
+ 500,
111
+ 60
112
+ ],
113
+ "flags": {},
114
+ "order": 2,
115
+ "mode": 0,
116
+ "inputs": [],
117
+ "outputs": [
118
+ {
119
+ "name": "VAE",
120
+ "type": "VAE",
121
+ "slot_index": 0,
122
+ "links": [
123
+ 4,
124
+ 17
125
+ ]
126
+ }
127
+ ],
128
+ "properties": {
129
+ "cnr_id": "comfy-core",
130
+ "ver": "0.44.0",
131
+ "Node name for S&R": "VAELoader",
132
+ "models": [
133
+ {
134
+ "name": "flux2-vae_small.safetensors",
135
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors",
136
+ "directory": "vae"
137
+ }
138
+ ]
139
+ },
140
+ "widgets_values": [
141
+ "flux2-vae.safetensors"
142
+ ],
143
+ "widgets_values_named": {
144
+ "vae_name": "flux2-vae.safetensors"
145
+ }
146
+ },
147
+ {
148
+ "id": 5,
149
+ "type": "CLIPTextEncode",
150
+ "pos": [
151
+ 185.95627952212953,
152
+ -36.734098034606404
153
+ ],
154
+ "size": [
155
+ 470,
156
+ 240
157
+ ],
158
+ "flags": {},
159
+ "order": 3,
160
+ "mode": 0,
161
+ "inputs": [
162
+ {
163
+ "name": "clip",
164
+ "type": "CLIP",
165
+ "link": 1
166
+ }
167
+ ],
168
+ "outputs": [
169
+ {
170
+ "name": "CONDITIONING",
171
+ "type": "CONDITIONING",
172
+ "slot_index": 0,
173
+ "links": [
174
+ 6
175
+ ]
176
+ }
177
+ ],
178
+ "properties": {
179
+ "cnr_id": "comfy-core",
180
+ "ver": "0.44.0",
181
+ "Node name for S&R": "CLIPTextEncode"
182
+ },
183
+ "widgets_values": [
184
+ "一只毛茸茸的猫,四肢顶端有着无人机的螺旋桨,腹部有着无人机的起落架,真实照片,可以看到根根分明的猫毛"
185
+ ],
186
+ "widgets_values_named": {
187
+ "text": "一只毛茸茸的猫,四肢顶端有着无人机的螺旋桨,腹部有着无人机的起落架,真实照片,可以看到根根分明的猫毛"
188
+ },
189
+ "color": "#232",
190
+ "bgcolor": "#353"
191
+ },
192
+ {
193
+ "id": 6,
194
+ "type": "CLIPTextEncode",
195
+ "pos": [
196
+ 150,
197
+ 250
198
+ ],
199
+ "size": [
200
+ 480,
201
+ 110
202
+ ],
203
+ "flags": {},
204
+ "order": 4,
205
+ "mode": 0,
206
+ "inputs": [
207
+ {
208
+ "name": "clip",
209
+ "type": "CLIP",
210
+ "link": 2
211
+ }
212
+ ],
213
+ "outputs": [
214
+ {
215
+ "name": "CONDITIONING",
216
+ "type": "CONDITIONING",
217
+ "slot_index": 0,
218
+ "links": [
219
+ 7
220
+ ]
221
+ }
222
+ ],
223
+ "properties": {
224
+ "cnr_id": "comfy-core",
225
+ "ver": "0.44.0",
226
+ "Node name for S&R": "CLIPTextEncode"
227
+ },
228
+ "widgets_values": [
229
+ "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
230
+ ],
231
+ "widgets_values_named": {
232
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
233
+ },
234
+ "color": "#223",
235
+ "bgcolor": "#335"
236
+ },
237
+ {
238
+ "id": 9,
239
+ "type": "VAEDecode",
240
+ "pos": [
241
+ 1910,
242
+ -520
243
+ ],
244
+ "size": [
245
+ 210,
246
+ 50
247
+ ],
248
+ "flags": {},
249
+ "order": 5,
250
+ "mode": 0,
251
+ "inputs": [
252
+ {
253
+ "name": "samples",
254
+ "type": "LATENT",
255
+ "link": 3
256
+ },
257
+ {
258
+ "name": "vae",
259
+ "type": "VAE",
260
+ "link": 4
261
+ }
262
+ ],
263
+ "outputs": [
264
+ {
265
+ "name": "IMAGE",
266
+ "type": "IMAGE",
267
+ "slot_index": 0,
268
+ "links": [
269
+ 13
270
+ ]
271
+ }
272
+ ],
273
+ "properties": {
274
+ "cnr_id": "comfy-core",
275
+ "ver": "0.44.0",
276
+ "Node name for S&R": "VAEDecode"
277
+ },
278
+ "color": "#322",
279
+ "bgcolor": "#533"
280
+ },
281
+ {
282
+ "id": 20,
283
+ "type": "SamplerCustom",
284
+ "pos": [
285
+ 1518.3539918681,
286
+ -304.93798438717295
287
+ ],
288
+ "size": [
289
+ 280,
290
+ 500
291
+ ],
292
+ "flags": {},
293
+ "order": 6,
294
+ "mode": 0,
295
+ "inputs": [
296
+ {
297
+ "name": "model",
298
+ "type": "MODEL",
299
+ "link": 5
300
+ },
301
+ {
302
+ "name": "positive",
303
+ "type": "CONDITIONING",
304
+ "link": 6
305
+ },
306
+ {
307
+ "name": "negative",
308
+ "type": "CONDITIONING",
309
+ "link": 7
310
+ },
311
+ {
312
+ "name": "sampler",
313
+ "type": "SAMPLER",
314
+ "link": 8
315
+ },
316
+ {
317
+ "name": "sigmas",
318
+ "type": "SIGMAS",
319
+ "link": 9
320
+ },
321
+ {
322
+ "name": "latent_image",
323
+ "type": "LATENT",
324
+ "link": 10
325
+ }
326
+ ],
327
+ "outputs": [
328
+ {
329
+ "name": "output",
330
+ "type": "LATENT",
331
+ "links": [
332
+ 3
333
+ ]
334
+ },
335
+ {
336
+ "name": "denoised_output",
337
+ "type": "LATENT",
338
+ "links": null
339
+ }
340
+ ],
341
+ "properties": {
342
+ "cnr_id": "comfy-core",
343
+ "ver": "0.44.0",
344
+ "Node name for S&R": "SamplerCustom"
345
+ },
346
+ "widgets_values": [
347
+ true,
348
+ 994529790995974,
349
+ "fixed",
350
+ 5
351
+ ],
352
+ "widgets_values_named": {
353
+ "add_noise": true,
354
+ "noise_seed": 994529790995974,
355
+ "control_after_generate": "fixed",
356
+ "cfg": 5.0
357
+ }
358
+ },
359
+ {
360
+ "id": 22,
361
+ "type": "BasicScheduler",
362
+ "pos": [
363
+ 1169.6140490851362,
364
+ 1.7699193319720672
365
+ ],
366
+ "size": [
367
+ 270,
368
+ 110
369
+ ],
370
+ "flags": {},
371
+ "order": 7,
372
+ "mode": 0,
373
+ "inputs": [
374
+ {
375
+ "name": "model",
376
+ "type": "MODEL",
377
+ "link": 11
378
+ }
379
+ ],
380
+ "outputs": [
381
+ {
382
+ "name": "SIGMAS",
383
+ "type": "SIGMAS",
384
+ "links": [
385
+ 9
386
+ ]
387
+ }
388
+ ],
389
+ "properties": {
390
+ "cnr_id": "comfy-core",
391
+ "ver": "0.44.0",
392
+ "Node name for S&R": "BasicScheduler"
393
+ },
394
+ "widgets_values": [
395
+ "sgm_uniform",
396
+ 15,
397
+ 1
398
+ ],
399
+ "widgets_values_named": {
400
+ "scheduler": "sgm_uniform",
401
+ "steps": 15,
402
+ "denoise": 1.0
403
+ }
404
+ },
405
+ {
406
+ "id": 36,
407
+ "type": "ModelSamplingAuraFlow",
408
+ "pos": [
409
+ 216.72263354772872,
410
+ -300
411
+ ],
412
+ "size": [
413
+ 400,
414
+ 90
415
+ ],
416
+ "flags": {},
417
+ "order": 8,
418
+ "mode": 0,
419
+ "inputs": [
420
+ {
421
+ "name": "model",
422
+ "type": "MODEL",
423
+ "link": 12
424
+ }
425
+ ],
426
+ "outputs": [
427
+ {
428
+ "name": "MODEL",
429
+ "type": "MODEL",
430
+ "links": [
431
+ 18
432
+ ]
433
+ }
434
+ ],
435
+ "properties": {
436
+ "cnr_id": "comfy-core",
437
+ "ver": "0.44.0",
438
+ "Node name for S&R": "ModelSamplingAuraFlow"
439
+ },
440
+ "widgets_values": [
441
+ 3,
442
+ "flow"
443
+ ],
444
+ "widgets_values_named": {
445
+ "shift": 3.0,
446
+ "sampling": "flow"
447
+ }
448
+ },
449
+ {
450
+ "id": 55,
451
+ "type": "LoadImage",
452
+ "pos": [
453
+ -400,
454
+ 630
455
+ ],
456
+ "size": [
457
+ 290,
458
+ 110
459
+ ],
460
+ "flags": {},
461
+ "order": 9,
462
+ "mode": 0,
463
+ "inputs": [],
464
+ "outputs": [
465
+ {
466
+ "name": "IMAGE",
467
+ "type": "IMAGE",
468
+ "links": [
469
+ 15
470
+ ]
471
+ },
472
+ {
473
+ "name": "MASK",
474
+ "type": "MASK",
475
+ "links": null
476
+ }
477
+ ],
478
+ "properties": {
479
+ "cnr_id": "comfy-core",
480
+ "ver": "0.44.0",
481
+ "Node name for S&R": "LoadImage"
482
+ },
483
+ "widgets_values": [
484
+ "flying-cat-input.png",
485
+ "image"
486
+ ],
487
+ "widgets_values_named": {
488
+ "image": "flying-cat-input.png",
489
+ "upload": "image"
490
+ },
491
+ "color": "#322",
492
+ "bgcolor": "#533"
493
+ },
494
+ {
495
+ "id": 85,
496
+ "type": "SaveImage",
497
+ "pos": [
498
+ 3307.317013871774,
499
+ 597.8636595391883
500
+ ],
501
+ "size": [
502
+ 420,
503
+ 470
504
+ ],
505
+ "flags": {},
506
+ "order": 10,
507
+ "mode": 0,
508
+ "inputs": [
509
+ {
510
+ "name": "images",
511
+ "type": "IMAGE",
512
+ "link": 13
513
+ }
514
+ ],
515
+ "outputs": [
516
+ {
517
+ "name": "images",
518
+ "type": "IMAGE",
519
+ "links": null
520
+ }
521
+ ],
522
+ "properties": {
523
+ "cnr_id": "comfy-core",
524
+ "ver": "0.44.0",
525
+ "Node name for S&R": "SaveImage"
526
+ },
527
+ "widgets_values": [
528
+ "MOSS-LLLite-V1/flying-cat"
529
+ ],
530
+ "widgets_values_named": {
531
+ "filename_prefix": "MOSS-LLLite-V1/flying-cat"
532
+ }
533
+ },
534
+ {
535
+ "id": 86,
536
+ "type": "KSamplerSelect",
537
+ "pos": [
538
+ 1170.4423572560272,
539
+ -260.582916386391
540
+ ],
541
+ "size": [
542
+ 270,
543
+ 58
544
+ ],
545
+ "flags": {},
546
+ "order": 11,
547
+ "mode": 0,
548
+ "inputs": [],
549
+ "outputs": [
550
+ {
551
+ "name": "SAMPLER",
552
+ "type": "SAMPLER",
553
+ "links": [
554
+ 8
555
+ ]
556
+ }
557
+ ],
558
+ "properties": {
559
+ "cnr_id": "comfy-core",
560
+ "ver": "0.44.0",
561
+ "Node name for S&R": "KSamplerSelect"
562
+ },
563
+ "widgets_values": [
564
+ "euler"
565
+ ],
566
+ "widgets_values_named": {
567
+ "sampler_name": "euler"
568
+ }
569
+ },
570
+ {
571
+ "id": 88,
572
+ "type": "LoraLoaderModelOnly",
573
+ "pos": [
574
+ -447.13559989481814,
575
+ -263.5408892596166
576
+ ],
577
+ "size": [
578
+ 500,
579
+ 120
580
+ ],
581
+ "flags": {},
582
+ "order": 12,
583
+ "mode": 0,
584
+ "inputs": [
585
+ {
586
+ "name": "model",
587
+ "type": "MODEL",
588
+ "link": 14
589
+ }
590
+ ],
591
+ "outputs": [
592
+ {
593
+ "name": "MODEL",
594
+ "type": "MODEL",
595
+ "links": [
596
+ 12
597
+ ]
598
+ }
599
+ ],
600
+ "properties": {
601
+ "cnr_id": "comfy-core",
602
+ "ver": "0.44.0",
603
+ "Node name for S&R": "LoraLoaderModelOnly",
604
+ "models": [
605
+ {
606
+ "name": "moss_image_color_depth_control.safetensors",
607
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
608
+ "directory": "loras"
609
+ }
610
+ ]
611
+ },
612
+ "widgets_values": [
613
+ "Coffmeteor-sft-15step-cfg3.safetensors",
614
+ 1
615
+ ],
616
+ "widgets_values_named": {
617
+ "lora_name": "Coffmeteor-sft-15step-cfg3.safetensors",
618
+ "strength_model": 1.0
619
+ }
620
+ },
621
+ {
622
+ "id": 90,
623
+ "type": "ImageScaleBy",
624
+ "pos": [
625
+ 321.18555329294736,
626
+ 683.4828563588695
627
+ ],
628
+ "size": [
629
+ 270,
630
+ 82
631
+ ],
632
+ "flags": {},
633
+ "order": 13,
634
+ "mode": 0,
635
+ "inputs": [
636
+ {
637
+ "name": "image",
638
+ "type": "IMAGE",
639
+ "link": 15
640
+ }
641
+ ],
642
+ "outputs": [
643
+ {
644
+ "name": "IMAGE",
645
+ "type": "IMAGE",
646
+ "links": [
647
+ 16,
648
+ 20
649
+ ]
650
+ }
651
+ ],
652
+ "properties": {
653
+ "cnr_id": "comfy-core",
654
+ "ver": "0.44.0",
655
+ "Node name for S&R": "ImageScaleBy"
656
+ },
657
+ "widgets_values": [
658
+ "lanczos",
659
+ 4
660
+ ],
661
+ "widgets_values_named": {
662
+ "upscale_method": "lanczos",
663
+ "scale_by": 4.0
664
+ }
665
+ },
666
+ {
667
+ "id": 124,
668
+ "type": "VAEEncode",
669
+ "pos": [
670
+ 1149.2279511667382,
671
+ 242.2625195138657
672
+ ],
673
+ "size": [
674
+ 140,
675
+ 46
676
+ ],
677
+ "flags": {},
678
+ "order": 14,
679
+ "mode": 0,
680
+ "inputs": [
681
+ {
682
+ "name": "pixels",
683
+ "type": "IMAGE",
684
+ "link": 16
685
+ },
686
+ {
687
+ "name": "vae",
688
+ "type": "VAE",
689
+ "link": 17
690
+ }
691
+ ],
692
+ "outputs": [
693
+ {
694
+ "name": "LATENT",
695
+ "type": "LATENT",
696
+ "links": [
697
+ 10
698
+ ]
699
+ }
700
+ ],
701
+ "properties": {
702
+ "cnr_id": "comfy-core",
703
+ "ver": "0.44.0",
704
+ "Node name for S&R": "VAEEncode"
705
+ }
706
+ },
707
+ {
708
+ "id": 130,
709
+ "type": "MossLLLiteApply",
710
+ "pos": [
711
+ 754.4053011472657,
712
+ 147.5866285701938
713
+ ],
714
+ "size": [
715
+ 270,
716
+ 146
717
+ ],
718
+ "flags": {},
719
+ "order": 15,
720
+ "mode": 0,
721
+ "inputs": [
722
+ {
723
+ "name": "model",
724
+ "type": "MODEL",
725
+ "link": 18
726
+ },
727
+ {
728
+ "name": "control",
729
+ "type": "MOSS_LLLITE_CONTROL",
730
+ "link": 19
731
+ },
732
+ {
733
+ "name": "image",
734
+ "type": "IMAGE",
735
+ "link": 20
736
+ }
737
+ ],
738
+ "outputs": [
739
+ {
740
+ "name": "MODEL",
741
+ "type": "MODEL",
742
+ "links": [
743
+ 5,
744
+ 11
745
+ ]
746
+ }
747
+ ],
748
+ "properties": {
749
+ "aux_id": "X-absurd/ComfyUI-MOSS-LLLite",
750
+ "ver": "20842b2b9d08511abd32404a0be24a51c9352109",
751
+ "Node name for S&R": "MossLLLiteApply"
752
+ },
753
+ "widgets_values": [
754
+ 1,
755
+ 0,
756
+ 1
757
+ ],
758
+ "widgets_values_named": {
759
+ "strength": 1.0,
760
+ "start_percent": 0.0,
761
+ "end_percent": 1.0
762
+ }
763
+ },
764
+ {
765
+ "id": 131,
766
+ "type": "MossLLLiteLoader",
767
+ "pos": [
768
+ 757.1064855291617,
769
+ -12.82076859408437
770
+ ],
771
+ "size": [
772
+ 270,
773
+ 58
774
+ ],
775
+ "flags": {},
776
+ "order": 16,
777
+ "mode": 0,
778
+ "inputs": [],
779
+ "outputs": [
780
+ {
781
+ "name": "MOSS_LLLITE_CONTROL",
782
+ "type": "MOSS_LLLITE_CONTROL",
783
+ "links": [
784
+ 19
785
+ ]
786
+ }
787
+ ],
788
+ "properties": {
789
+ "aux_id": "X-absurd/ComfyUI-MOSS-LLLite",
790
+ "ver": "20842b2b9d08511abd32404a0be24a51c9352109",
791
+ "Node name for S&R": "MossLLLiteLoader"
792
+ },
793
+ "widgets_values": [
794
+ "MOSS-LLLite-Tile-V1.safetensors"
795
+ ],
796
+ "widgets_values_named": {
797
+ "name": "MOSS-LLLite-Tile-V1.safetensors"
798
+ }
799
+ }
800
+ ],
801
+ "links": [
802
+ [
803
+ 1,
804
+ 3,
805
+ 0,
806
+ 5,
807
+ 0,
808
+ "CLIP"
809
+ ],
810
+ [
811
+ 2,
812
+ 3,
813
+ 0,
814
+ 6,
815
+ 0,
816
+ "CLIP"
817
+ ],
818
+ [
819
+ 3,
820
+ 20,
821
+ 0,
822
+ 9,
823
+ 0,
824
+ "LATENT"
825
+ ],
826
+ [
827
+ 4,
828
+ 4,
829
+ 0,
830
+ 9,
831
+ 1,
832
+ "VAE"
833
+ ],
834
+ [
835
+ 5,
836
+ 130,
837
+ 0,
838
+ 20,
839
+ 0,
840
+ "MODEL"
841
+ ],
842
+ [
843
+ 6,
844
+ 5,
845
+ 0,
846
+ 20,
847
+ 1,
848
+ "CONDITIONING"
849
+ ],
850
+ [
851
+ 7,
852
+ 6,
853
+ 0,
854
+ 20,
855
+ 2,
856
+ "CONDITIONING"
857
+ ],
858
+ [
859
+ 8,
860
+ 86,
861
+ 0,
862
+ 20,
863
+ 3,
864
+ "SAMPLER"
865
+ ],
866
+ [
867
+ 9,
868
+ 22,
869
+ 0,
870
+ 20,
871
+ 4,
872
+ "SIGMAS"
873
+ ],
874
+ [
875
+ 10,
876
+ 124,
877
+ 0,
878
+ 20,
879
+ 5,
880
+ "LATENT"
881
+ ],
882
+ [
883
+ 11,
884
+ 130,
885
+ 0,
886
+ 22,
887
+ 0,
888
+ "MODEL"
889
+ ],
890
+ [
891
+ 12,
892
+ 88,
893
+ 0,
894
+ 36,
895
+ 0,
896
+ "MODEL"
897
+ ],
898
+ [
899
+ 13,
900
+ 9,
901
+ 0,
902
+ 85,
903
+ 0,
904
+ "IMAGE"
905
+ ],
906
+ [
907
+ 14,
908
+ 1,
909
+ 0,
910
+ 88,
911
+ 0,
912
+ "MODEL"
913
+ ],
914
+ [
915
+ 15,
916
+ 55,
917
+ 0,
918
+ 90,
919
+ 0,
920
+ "IMAGE"
921
+ ],
922
+ [
923
+ 16,
924
+ 90,
925
+ 0,
926
+ 124,
927
+ 0,
928
+ "IMAGE"
929
+ ],
930
+ [
931
+ 17,
932
+ 4,
933
+ 0,
934
+ 124,
935
+ 1,
936
+ "VAE"
937
+ ],
938
+ [
939
+ 18,
940
+ 36,
941
+ 0,
942
+ 130,
943
+ 0,
944
+ "MODEL"
945
+ ],
946
+ [
947
+ 19,
948
+ 131,
949
+ 0,
950
+ 130,
951
+ 1,
952
+ "MOSS_LLLITE_CONTROL"
953
+ ],
954
+ [
955
+ 20,
956
+ 90,
957
+ 0,
958
+ 130,
959
+ 2,
960
+ "IMAGE"
961
+ ]
962
+ ],
963
+ "groups": [],
964
+ "config": {},
965
+ "extra": {
966
+ "ds": {
967
+ "scale": 0.55,
968
+ "offset": [
969
+ 550,
970
+ 150
971
+ ]
972
+ }
973
+ },
974
+ "version": 0.4
975
+ }
examples/v1/grass-control.png ADDED

Git LFS Details

  • SHA256: 14d3a2c02e67720949d41539d6db42e59c7595eb19a102da6fa4af8ba775ff76
  • Pointer size: 132 Bytes
  • Size of remote file: 2.26 MB
examples/v1/grass-input.png ADDED

Git LFS Details

  • SHA256: ac406ef7c8a3410bd199901366e83430a7aa6c49a82318e4e1b8d95b1b0b6585
  • Pointer size: 133 Bytes
  • Size of remote file: 11.6 MB
examples/v1/grass-v1.png ADDED

Git LFS Details

  • SHA256: c967a858169b2a8310990e708d03bb4c372475272af666d62d5aff3a276bfc8d
  • Pointer size: 132 Bytes
  • Size of remote file: 4.05 MB
examples/v1/grass.api.json ADDED
@@ -0,0 +1,205 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "1": {
3
+ "class_type": "UNETLoader",
4
+ "inputs": {
5
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
6
+ "weight_dtype": "default"
7
+ }
8
+ },
9
+ "3": {
10
+ "class_type": "CLIPLoader",
11
+ "inputs": {
12
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
13
+ "type": "moss_image",
14
+ "device": "default"
15
+ }
16
+ },
17
+ "4": {
18
+ "class_type": "VAELoader",
19
+ "inputs": {
20
+ "vae_name": "flux2-vae.safetensors"
21
+ }
22
+ },
23
+ "5": {
24
+ "class_type": "CLIPTextEncode",
25
+ "inputs": {
26
+ "text": "草地特写,清晨的阳光",
27
+ "clip": [
28
+ "3",
29
+ 0
30
+ ]
31
+ }
32
+ },
33
+ "6": {
34
+ "class_type": "CLIPTextEncode",
35
+ "inputs": {
36
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact",
37
+ "clip": [
38
+ "3",
39
+ 0
40
+ ]
41
+ }
42
+ },
43
+ "9": {
44
+ "class_type": "VAEDecode",
45
+ "inputs": {
46
+ "samples": [
47
+ "20",
48
+ 0
49
+ ],
50
+ "vae": [
51
+ "4",
52
+ 0
53
+ ]
54
+ }
55
+ },
56
+ "20": {
57
+ "class_type": "SamplerCustom",
58
+ "inputs": {
59
+ "add_noise": true,
60
+ "noise_seed": 994529790995974,
61
+ "cfg": 5.0,
62
+ "model": [
63
+ "130",
64
+ 0
65
+ ],
66
+ "positive": [
67
+ "5",
68
+ 0
69
+ ],
70
+ "negative": [
71
+ "6",
72
+ 0
73
+ ],
74
+ "sampler": [
75
+ "86",
76
+ 0
77
+ ],
78
+ "sigmas": [
79
+ "22",
80
+ 0
81
+ ],
82
+ "latent_image": [
83
+ "124",
84
+ 0
85
+ ]
86
+ }
87
+ },
88
+ "22": {
89
+ "class_type": "BasicScheduler",
90
+ "inputs": {
91
+ "scheduler": "sgm_uniform",
92
+ "steps": 15,
93
+ "denoise": 1.0,
94
+ "model": [
95
+ "130",
96
+ 0
97
+ ]
98
+ }
99
+ },
100
+ "36": {
101
+ "class_type": "ModelSamplingAuraFlow",
102
+ "inputs": {
103
+ "shift": 3.0,
104
+ "sampling": "flow",
105
+ "model": [
106
+ "88",
107
+ 0
108
+ ]
109
+ }
110
+ },
111
+ "55": {
112
+ "class_type": "LoadImage",
113
+ "inputs": {
114
+ "image": "grass-input.png"
115
+ }
116
+ },
117
+ "85": {
118
+ "class_type": "SaveImage",
119
+ "inputs": {
120
+ "filename_prefix": "MOSS-LLLite-V1/grass",
121
+ "images": [
122
+ "9",
123
+ 0
124
+ ]
125
+ }
126
+ },
127
+ "86": {
128
+ "class_type": "KSamplerSelect",
129
+ "inputs": {
130
+ "sampler_name": "euler"
131
+ }
132
+ },
133
+ "87": {
134
+ "class_type": "ImageScaleBy",
135
+ "inputs": {
136
+ "upscale_method": "lanczos",
137
+ "scale_by": 0.1,
138
+ "image": [
139
+ "55",
140
+ 0
141
+ ]
142
+ }
143
+ },
144
+ "88": {
145
+ "class_type": "LoraLoaderModelOnly",
146
+ "inputs": {
147
+ "lora_name": "Coffmeteor-sft-15step-cfg3.safetensors",
148
+ "strength_model": 1.0,
149
+ "model": [
150
+ "1",
151
+ 0
152
+ ]
153
+ }
154
+ },
155
+ "90": {
156
+ "class_type": "ImageScaleBy",
157
+ "inputs": {
158
+ "upscale_method": "lanczos",
159
+ "scale_by": 5.0,
160
+ "image": [
161
+ "87",
162
+ 0
163
+ ]
164
+ }
165
+ },
166
+ "124": {
167
+ "class_type": "VAEEncode",
168
+ "inputs": {
169
+ "pixels": [
170
+ "90",
171
+ 0
172
+ ],
173
+ "vae": [
174
+ "4",
175
+ 0
176
+ ]
177
+ }
178
+ },
179
+ "130": {
180
+ "class_type": "MossLLLiteApply",
181
+ "inputs": {
182
+ "strength": 1.0,
183
+ "start_percent": 0.0,
184
+ "end_percent": 1.0,
185
+ "model": [
186
+ "36",
187
+ 0
188
+ ],
189
+ "control": [
190
+ "131",
191
+ 0
192
+ ],
193
+ "image": [
194
+ "90",
195
+ 0
196
+ ]
197
+ }
198
+ },
199
+ "131": {
200
+ "class_type": "MossLLLiteLoader",
201
+ "inputs": {
202
+ "name": "MOSS-LLLite-Tile-V1.safetensors"
203
+ }
204
+ }
205
+ }
examples/v1/grass.workflow.json ADDED
@@ -0,0 +1,1027 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "id": "063f06a3-595f-4c31-87f1-02402ccd2d3b",
3
+ "revision": 0,
4
+ "last_node_id": 131,
5
+ "last_link_id": 21,
6
+ "nodes": [
7
+ {
8
+ "id": 1,
9
+ "type": "UNETLoader",
10
+ "pos": [
11
+ -423.26391629976365,
12
+ -689.4382881990866
13
+ ],
14
+ "size": [
15
+ 500,
16
+ 90
17
+ ],
18
+ "flags": {},
19
+ "order": 0,
20
+ "mode": 0,
21
+ "inputs": [],
22
+ "outputs": [
23
+ {
24
+ "name": "MODEL",
25
+ "type": "MODEL",
26
+ "slot_index": 0,
27
+ "links": [
28
+ 15
29
+ ]
30
+ }
31
+ ],
32
+ "properties": {
33
+ "cnr_id": "comfy-core",
34
+ "ver": "0.44.0",
35
+ "Node name for S&R": "UNETLoader",
36
+ "models": [
37
+ {
38
+ "name": "moss_image_sft_int8_convrot.safetensors",
39
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors",
40
+ "directory": "diffusion_models"
41
+ }
42
+ ]
43
+ },
44
+ "widgets_values": [
45
+ "moss-image/moss_image_sft_int8_convrot.safetensors",
46
+ "default"
47
+ ],
48
+ "widgets_values_named": {
49
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
50
+ "weight_dtype": "default"
51
+ }
52
+ },
53
+ {
54
+ "id": 3,
55
+ "type": "CLIPLoader",
56
+ "pos": [
57
+ -440,
58
+ -50
59
+ ],
60
+ "size": [
61
+ 500,
62
+ 110
63
+ ],
64
+ "flags": {},
65
+ "order": 1,
66
+ "mode": 0,
67
+ "inputs": [],
68
+ "outputs": [
69
+ {
70
+ "name": "CLIP",
71
+ "type": "CLIP",
72
+ "slot_index": 0,
73
+ "links": [
74
+ 1,
75
+ 2
76
+ ]
77
+ }
78
+ ],
79
+ "properties": {
80
+ "cnr_id": "comfy-core",
81
+ "ver": "0.44.0",
82
+ "Node name for S&R": "CLIPLoader",
83
+ "models": [
84
+ {
85
+ "name": "qwen3.5_9b_int8_convrot.safetensors",
86
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors",
87
+ "directory": "text_encoders"
88
+ }
89
+ ]
90
+ },
91
+ "widgets_values": [
92
+ "qwen3.5_9b_bf16.safetensors",
93
+ "moss_image",
94
+ "default"
95
+ ],
96
+ "widgets_values_named": {
97
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
98
+ "type": "moss_image",
99
+ "device": "default"
100
+ }
101
+ },
102
+ {
103
+ "id": 4,
104
+ "type": "VAELoader",
105
+ "pos": [
106
+ -440,
107
+ 130
108
+ ],
109
+ "size": [
110
+ 500,
111
+ 60
112
+ ],
113
+ "flags": {},
114
+ "order": 2,
115
+ "mode": 0,
116
+ "inputs": [],
117
+ "outputs": [
118
+ {
119
+ "name": "VAE",
120
+ "type": "VAE",
121
+ "slot_index": 0,
122
+ "links": [
123
+ 4,
124
+ 18
125
+ ]
126
+ }
127
+ ],
128
+ "properties": {
129
+ "cnr_id": "comfy-core",
130
+ "ver": "0.44.0",
131
+ "Node name for S&R": "VAELoader",
132
+ "models": [
133
+ {
134
+ "name": "flux2-vae_small.safetensors",
135
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors",
136
+ "directory": "vae"
137
+ }
138
+ ]
139
+ },
140
+ "widgets_values": [
141
+ "flux2-vae.safetensors"
142
+ ],
143
+ "widgets_values_named": {
144
+ "vae_name": "flux2-vae.safetensors"
145
+ }
146
+ },
147
+ {
148
+ "id": 5,
149
+ "type": "CLIPTextEncode",
150
+ "pos": [
151
+ 185.95627952212953,
152
+ -36.734098034606404
153
+ ],
154
+ "size": [
155
+ 470,
156
+ 240
157
+ ],
158
+ "flags": {},
159
+ "order": 3,
160
+ "mode": 0,
161
+ "inputs": [
162
+ {
163
+ "name": "clip",
164
+ "type": "CLIP",
165
+ "link": 1
166
+ }
167
+ ],
168
+ "outputs": [
169
+ {
170
+ "name": "CONDITIONING",
171
+ "type": "CONDITIONING",
172
+ "slot_index": 0,
173
+ "links": [
174
+ 6
175
+ ]
176
+ }
177
+ ],
178
+ "properties": {
179
+ "cnr_id": "comfy-core",
180
+ "ver": "0.44.0",
181
+ "Node name for S&R": "CLIPTextEncode"
182
+ },
183
+ "widgets_values": [
184
+ "草地特写,清晨的阳光"
185
+ ],
186
+ "widgets_values_named": {
187
+ "text": "草地特写,清晨的阳光"
188
+ },
189
+ "color": "#232",
190
+ "bgcolor": "#353"
191
+ },
192
+ {
193
+ "id": 6,
194
+ "type": "CLIPTextEncode",
195
+ "pos": [
196
+ 150,
197
+ 250
198
+ ],
199
+ "size": [
200
+ 480,
201
+ 110
202
+ ],
203
+ "flags": {},
204
+ "order": 4,
205
+ "mode": 0,
206
+ "inputs": [
207
+ {
208
+ "name": "clip",
209
+ "type": "CLIP",
210
+ "link": 2
211
+ }
212
+ ],
213
+ "outputs": [
214
+ {
215
+ "name": "CONDITIONING",
216
+ "type": "CONDITIONING",
217
+ "slot_index": 0,
218
+ "links": [
219
+ 7
220
+ ]
221
+ }
222
+ ],
223
+ "properties": {
224
+ "cnr_id": "comfy-core",
225
+ "ver": "0.44.0",
226
+ "Node name for S&R": "CLIPTextEncode"
227
+ },
228
+ "widgets_values": [
229
+ "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
230
+ ],
231
+ "widgets_values_named": {
232
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
233
+ },
234
+ "color": "#223",
235
+ "bgcolor": "#335"
236
+ },
237
+ {
238
+ "id": 9,
239
+ "type": "VAEDecode",
240
+ "pos": [
241
+ 1910,
242
+ -520
243
+ ],
244
+ "size": [
245
+ 210,
246
+ 50
247
+ ],
248
+ "flags": {},
249
+ "order": 5,
250
+ "mode": 0,
251
+ "inputs": [
252
+ {
253
+ "name": "samples",
254
+ "type": "LATENT",
255
+ "link": 3
256
+ },
257
+ {
258
+ "name": "vae",
259
+ "type": "VAE",
260
+ "link": 4
261
+ }
262
+ ],
263
+ "outputs": [
264
+ {
265
+ "name": "IMAGE",
266
+ "type": "IMAGE",
267
+ "slot_index": 0,
268
+ "links": [
269
+ 13
270
+ ]
271
+ }
272
+ ],
273
+ "properties": {
274
+ "cnr_id": "comfy-core",
275
+ "ver": "0.44.0",
276
+ "Node name for S&R": "VAEDecode"
277
+ },
278
+ "color": "#322",
279
+ "bgcolor": "#533"
280
+ },
281
+ {
282
+ "id": 20,
283
+ "type": "SamplerCustom",
284
+ "pos": [
285
+ 1518.3539918681,
286
+ -304.93798438717295
287
+ ],
288
+ "size": [
289
+ 280,
290
+ 500
291
+ ],
292
+ "flags": {},
293
+ "order": 6,
294
+ "mode": 0,
295
+ "inputs": [
296
+ {
297
+ "name": "model",
298
+ "type": "MODEL",
299
+ "link": 5
300
+ },
301
+ {
302
+ "name": "positive",
303
+ "type": "CONDITIONING",
304
+ "link": 6
305
+ },
306
+ {
307
+ "name": "negative",
308
+ "type": "CONDITIONING",
309
+ "link": 7
310
+ },
311
+ {
312
+ "name": "sampler",
313
+ "type": "SAMPLER",
314
+ "link": 8
315
+ },
316
+ {
317
+ "name": "sigmas",
318
+ "type": "SIGMAS",
319
+ "link": 9
320
+ },
321
+ {
322
+ "name": "latent_image",
323
+ "type": "LATENT",
324
+ "link": 10
325
+ }
326
+ ],
327
+ "outputs": [
328
+ {
329
+ "name": "output",
330
+ "type": "LATENT",
331
+ "links": [
332
+ 3
333
+ ]
334
+ },
335
+ {
336
+ "name": "denoised_output",
337
+ "type": "LATENT",
338
+ "links": null
339
+ }
340
+ ],
341
+ "properties": {
342
+ "cnr_id": "comfy-core",
343
+ "ver": "0.44.0",
344
+ "Node name for S&R": "SamplerCustom"
345
+ },
346
+ "widgets_values": [
347
+ true,
348
+ 994529790995974,
349
+ "fixed",
350
+ 5
351
+ ],
352
+ "widgets_values_named": {
353
+ "add_noise": true,
354
+ "noise_seed": 994529790995974,
355
+ "control_after_generate": "fixed",
356
+ "cfg": 5.0
357
+ }
358
+ },
359
+ {
360
+ "id": 22,
361
+ "type": "BasicScheduler",
362
+ "pos": [
363
+ 1169.6140490851362,
364
+ 1.7699193319720672
365
+ ],
366
+ "size": [
367
+ 270,
368
+ 110
369
+ ],
370
+ "flags": {},
371
+ "order": 7,
372
+ "mode": 0,
373
+ "inputs": [
374
+ {
375
+ "name": "model",
376
+ "type": "MODEL",
377
+ "link": 11
378
+ }
379
+ ],
380
+ "outputs": [
381
+ {
382
+ "name": "SIGMAS",
383
+ "type": "SIGMAS",
384
+ "links": [
385
+ 9
386
+ ]
387
+ }
388
+ ],
389
+ "properties": {
390
+ "cnr_id": "comfy-core",
391
+ "ver": "0.44.0",
392
+ "Node name for S&R": "BasicScheduler"
393
+ },
394
+ "widgets_values": [
395
+ "sgm_uniform",
396
+ 15,
397
+ 1
398
+ ],
399
+ "widgets_values_named": {
400
+ "scheduler": "sgm_uniform",
401
+ "steps": 15,
402
+ "denoise": 1.0
403
+ }
404
+ },
405
+ {
406
+ "id": 36,
407
+ "type": "ModelSamplingAuraFlow",
408
+ "pos": [
409
+ 216.72263354772872,
410
+ -300
411
+ ],
412
+ "size": [
413
+ 400,
414
+ 90
415
+ ],
416
+ "flags": {},
417
+ "order": 8,
418
+ "mode": 0,
419
+ "inputs": [
420
+ {
421
+ "name": "model",
422
+ "type": "MODEL",
423
+ "link": 12
424
+ }
425
+ ],
426
+ "outputs": [
427
+ {
428
+ "name": "MODEL",
429
+ "type": "MODEL",
430
+ "links": [
431
+ 19
432
+ ]
433
+ }
434
+ ],
435
+ "properties": {
436
+ "cnr_id": "comfy-core",
437
+ "ver": "0.44.0",
438
+ "Node name for S&R": "ModelSamplingAuraFlow"
439
+ },
440
+ "widgets_values": [
441
+ 3,
442
+ "flow"
443
+ ],
444
+ "widgets_values_named": {
445
+ "shift": 3.0,
446
+ "sampling": "flow"
447
+ }
448
+ },
449
+ {
450
+ "id": 55,
451
+ "type": "LoadImage",
452
+ "pos": [
453
+ -400,
454
+ 630
455
+ ],
456
+ "size": [
457
+ 290,
458
+ 110
459
+ ],
460
+ "flags": {},
461
+ "order": 9,
462
+ "mode": 0,
463
+ "inputs": [],
464
+ "outputs": [
465
+ {
466
+ "name": "IMAGE",
467
+ "type": "IMAGE",
468
+ "links": [
469
+ 14
470
+ ]
471
+ },
472
+ {
473
+ "name": "MASK",
474
+ "type": "MASK",
475
+ "links": null
476
+ }
477
+ ],
478
+ "properties": {
479
+ "cnr_id": "comfy-core",
480
+ "ver": "0.44.0",
481
+ "Node name for S&R": "LoadImage"
482
+ },
483
+ "widgets_values": [
484
+ "grass-input.png",
485
+ "image"
486
+ ],
487
+ "widgets_values_named": {
488
+ "image": "grass-input.png",
489
+ "upload": "image"
490
+ },
491
+ "color": "#322",
492
+ "bgcolor": "#533"
493
+ },
494
+ {
495
+ "id": 85,
496
+ "type": "SaveImage",
497
+ "pos": [
498
+ 3307.317013871774,
499
+ 597.8636595391883
500
+ ],
501
+ "size": [
502
+ 420,
503
+ 470
504
+ ],
505
+ "flags": {},
506
+ "order": 10,
507
+ "mode": 0,
508
+ "inputs": [
509
+ {
510
+ "name": "images",
511
+ "type": "IMAGE",
512
+ "link": 13
513
+ }
514
+ ],
515
+ "outputs": [
516
+ {
517
+ "name": "images",
518
+ "type": "IMAGE",
519
+ "links": null
520
+ }
521
+ ],
522
+ "properties": {
523
+ "cnr_id": "comfy-core",
524
+ "ver": "0.44.0",
525
+ "Node name for S&R": "SaveImage"
526
+ },
527
+ "widgets_values": [
528
+ "MOSS-LLLite-V1/grass"
529
+ ],
530
+ "widgets_values_named": {
531
+ "filename_prefix": "MOSS-LLLite-V1/grass"
532
+ }
533
+ },
534
+ {
535
+ "id": 86,
536
+ "type": "KSamplerSelect",
537
+ "pos": [
538
+ 1170.4423572560272,
539
+ -260.582916386391
540
+ ],
541
+ "size": [
542
+ 270,
543
+ 58
544
+ ],
545
+ "flags": {},
546
+ "order": 11,
547
+ "mode": 0,
548
+ "inputs": [],
549
+ "outputs": [
550
+ {
551
+ "name": "SAMPLER",
552
+ "type": "SAMPLER",
553
+ "links": [
554
+ 8
555
+ ]
556
+ }
557
+ ],
558
+ "properties": {
559
+ "cnr_id": "comfy-core",
560
+ "ver": "0.44.0",
561
+ "Node name for S&R": "KSamplerSelect"
562
+ },
563
+ "widgets_values": [
564
+ "euler"
565
+ ],
566
+ "widgets_values_named": {
567
+ "sampler_name": "euler"
568
+ }
569
+ },
570
+ {
571
+ "id": 87,
572
+ "type": "ImageScaleBy",
573
+ "pos": [
574
+ -7.178545924871845,
575
+ 681.7385897176745
576
+ ],
577
+ "size": [
578
+ 270,
579
+ 82
580
+ ],
581
+ "flags": {},
582
+ "order": 12,
583
+ "mode": 0,
584
+ "inputs": [
585
+ {
586
+ "name": "image",
587
+ "type": "IMAGE",
588
+ "link": 14
589
+ }
590
+ ],
591
+ "outputs": [
592
+ {
593
+ "name": "IMAGE",
594
+ "type": "IMAGE",
595
+ "links": [
596
+ 16
597
+ ]
598
+ }
599
+ ],
600
+ "properties": {
601
+ "cnr_id": "comfy-core",
602
+ "ver": "0.44.0",
603
+ "Node name for S&R": "ImageScaleBy"
604
+ },
605
+ "widgets_values": [
606
+ "lanczos",
607
+ 0.1
608
+ ],
609
+ "widgets_values_named": {
610
+ "upscale_method": "lanczos",
611
+ "scale_by": 0.1
612
+ }
613
+ },
614
+ {
615
+ "id": 88,
616
+ "type": "LoraLoaderModelOnly",
617
+ "pos": [
618
+ -447.13559989481814,
619
+ -263.5408892596166
620
+ ],
621
+ "size": [
622
+ 500,
623
+ 120
624
+ ],
625
+ "flags": {},
626
+ "order": 13,
627
+ "mode": 0,
628
+ "inputs": [
629
+ {
630
+ "name": "model",
631
+ "type": "MODEL",
632
+ "link": 15
633
+ }
634
+ ],
635
+ "outputs": [
636
+ {
637
+ "name": "MODEL",
638
+ "type": "MODEL",
639
+ "links": [
640
+ 12
641
+ ]
642
+ }
643
+ ],
644
+ "properties": {
645
+ "cnr_id": "comfy-core",
646
+ "ver": "0.44.0",
647
+ "Node name for S&R": "LoraLoaderModelOnly",
648
+ "models": [
649
+ {
650
+ "name": "moss_image_color_depth_control.safetensors",
651
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
652
+ "directory": "loras"
653
+ }
654
+ ]
655
+ },
656
+ "widgets_values": [
657
+ "Coffmeteor-sft-15step-cfg3.safetensors",
658
+ 1
659
+ ],
660
+ "widgets_values_named": {
661
+ "lora_name": "Coffmeteor-sft-15step-cfg3.safetensors",
662
+ "strength_model": 1.0
663
+ }
664
+ },
665
+ {
666
+ "id": 90,
667
+ "type": "ImageScaleBy",
668
+ "pos": [
669
+ 321.18555329294736,
670
+ 683.4828563588695
671
+ ],
672
+ "size": [
673
+ 270,
674
+ 82
675
+ ],
676
+ "flags": {},
677
+ "order": 14,
678
+ "mode": 0,
679
+ "inputs": [
680
+ {
681
+ "name": "image",
682
+ "type": "IMAGE",
683
+ "link": 16
684
+ }
685
+ ],
686
+ "outputs": [
687
+ {
688
+ "name": "IMAGE",
689
+ "type": "IMAGE",
690
+ "links": [
691
+ 17,
692
+ 21
693
+ ]
694
+ }
695
+ ],
696
+ "properties": {
697
+ "cnr_id": "comfy-core",
698
+ "ver": "0.44.0",
699
+ "Node name for S&R": "ImageScaleBy"
700
+ },
701
+ "widgets_values": [
702
+ "lanczos",
703
+ 5
704
+ ],
705
+ "widgets_values_named": {
706
+ "upscale_method": "lanczos",
707
+ "scale_by": 5.0
708
+ }
709
+ },
710
+ {
711
+ "id": 124,
712
+ "type": "VAEEncode",
713
+ "pos": [
714
+ 1149.2279511667382,
715
+ 242.2625195138657
716
+ ],
717
+ "size": [
718
+ 140,
719
+ 46
720
+ ],
721
+ "flags": {},
722
+ "order": 15,
723
+ "mode": 0,
724
+ "inputs": [
725
+ {
726
+ "name": "pixels",
727
+ "type": "IMAGE",
728
+ "link": 17
729
+ },
730
+ {
731
+ "name": "vae",
732
+ "type": "VAE",
733
+ "link": 18
734
+ }
735
+ ],
736
+ "outputs": [
737
+ {
738
+ "name": "LATENT",
739
+ "type": "LATENT",
740
+ "links": [
741
+ 10
742
+ ]
743
+ }
744
+ ],
745
+ "properties": {
746
+ "cnr_id": "comfy-core",
747
+ "ver": "0.44.0",
748
+ "Node name for S&R": "VAEEncode"
749
+ }
750
+ },
751
+ {
752
+ "id": 130,
753
+ "type": "MossLLLiteApply",
754
+ "pos": [
755
+ 754.4053011472657,
756
+ 147.5866285701938
757
+ ],
758
+ "size": [
759
+ 270,
760
+ 146
761
+ ],
762
+ "flags": {},
763
+ "order": 16,
764
+ "mode": 0,
765
+ "inputs": [
766
+ {
767
+ "name": "model",
768
+ "type": "MODEL",
769
+ "link": 19
770
+ },
771
+ {
772
+ "name": "control",
773
+ "type": "MOSS_LLLITE_CONTROL",
774
+ "link": 20
775
+ },
776
+ {
777
+ "name": "image",
778
+ "type": "IMAGE",
779
+ "link": 21
780
+ }
781
+ ],
782
+ "outputs": [
783
+ {
784
+ "name": "MODEL",
785
+ "type": "MODEL",
786
+ "links": [
787
+ 5,
788
+ 11
789
+ ]
790
+ }
791
+ ],
792
+ "properties": {
793
+ "aux_id": "X-absurd/ComfyUI-MOSS-LLLite",
794
+ "ver": "20842b2b9d08511abd32404a0be24a51c9352109",
795
+ "Node name for S&R": "MossLLLiteApply"
796
+ },
797
+ "widgets_values": [
798
+ 1,
799
+ 0,
800
+ 1
801
+ ],
802
+ "widgets_values_named": {
803
+ "strength": 1.0,
804
+ "start_percent": 0.0,
805
+ "end_percent": 1.0
806
+ }
807
+ },
808
+ {
809
+ "id": 131,
810
+ "type": "MossLLLiteLoader",
811
+ "pos": [
812
+ 757.1064855291617,
813
+ -12.82076859408437
814
+ ],
815
+ "size": [
816
+ 270,
817
+ 58
818
+ ],
819
+ "flags": {},
820
+ "order": 17,
821
+ "mode": 0,
822
+ "inputs": [],
823
+ "outputs": [
824
+ {
825
+ "name": "MOSS_LLLITE_CONTROL",
826
+ "type": "MOSS_LLLITE_CONTROL",
827
+ "links": [
828
+ 20
829
+ ]
830
+ }
831
+ ],
832
+ "properties": {
833
+ "aux_id": "X-absurd/ComfyUI-MOSS-LLLite",
834
+ "ver": "20842b2b9d08511abd32404a0be24a51c9352109",
835
+ "Node name for S&R": "MossLLLiteLoader"
836
+ },
837
+ "widgets_values": [
838
+ "MOSS-LLLite-Tile-V1.safetensors"
839
+ ],
840
+ "widgets_values_named": {
841
+ "name": "MOSS-LLLite-Tile-V1.safetensors"
842
+ }
843
+ }
844
+ ],
845
+ "links": [
846
+ [
847
+ 1,
848
+ 3,
849
+ 0,
850
+ 5,
851
+ 0,
852
+ "CLIP"
853
+ ],
854
+ [
855
+ 2,
856
+ 3,
857
+ 0,
858
+ 6,
859
+ 0,
860
+ "CLIP"
861
+ ],
862
+ [
863
+ 3,
864
+ 20,
865
+ 0,
866
+ 9,
867
+ 0,
868
+ "LATENT"
869
+ ],
870
+ [
871
+ 4,
872
+ 4,
873
+ 0,
874
+ 9,
875
+ 1,
876
+ "VAE"
877
+ ],
878
+ [
879
+ 5,
880
+ 130,
881
+ 0,
882
+ 20,
883
+ 0,
884
+ "MODEL"
885
+ ],
886
+ [
887
+ 6,
888
+ 5,
889
+ 0,
890
+ 20,
891
+ 1,
892
+ "CONDITIONING"
893
+ ],
894
+ [
895
+ 7,
896
+ 6,
897
+ 0,
898
+ 20,
899
+ 2,
900
+ "CONDITIONING"
901
+ ],
902
+ [
903
+ 8,
904
+ 86,
905
+ 0,
906
+ 20,
907
+ 3,
908
+ "SAMPLER"
909
+ ],
910
+ [
911
+ 9,
912
+ 22,
913
+ 0,
914
+ 20,
915
+ 4,
916
+ "SIGMAS"
917
+ ],
918
+ [
919
+ 10,
920
+ 124,
921
+ 0,
922
+ 20,
923
+ 5,
924
+ "LATENT"
925
+ ],
926
+ [
927
+ 11,
928
+ 130,
929
+ 0,
930
+ 22,
931
+ 0,
932
+ "MODEL"
933
+ ],
934
+ [
935
+ 12,
936
+ 88,
937
+ 0,
938
+ 36,
939
+ 0,
940
+ "MODEL"
941
+ ],
942
+ [
943
+ 13,
944
+ 9,
945
+ 0,
946
+ 85,
947
+ 0,
948
+ "IMAGE"
949
+ ],
950
+ [
951
+ 14,
952
+ 55,
953
+ 0,
954
+ 87,
955
+ 0,
956
+ "IMAGE"
957
+ ],
958
+ [
959
+ 15,
960
+ 1,
961
+ 0,
962
+ 88,
963
+ 0,
964
+ "MODEL"
965
+ ],
966
+ [
967
+ 16,
968
+ 87,
969
+ 0,
970
+ 90,
971
+ 0,
972
+ "IMAGE"
973
+ ],
974
+ [
975
+ 17,
976
+ 90,
977
+ 0,
978
+ 124,
979
+ 0,
980
+ "IMAGE"
981
+ ],
982
+ [
983
+ 18,
984
+ 4,
985
+ 0,
986
+ 124,
987
+ 1,
988
+ "VAE"
989
+ ],
990
+ [
991
+ 19,
992
+ 36,
993
+ 0,
994
+ 130,
995
+ 0,
996
+ "MODEL"
997
+ ],
998
+ [
999
+ 20,
1000
+ 131,
1001
+ 0,
1002
+ 130,
1003
+ 1,
1004
+ "MOSS_LLLITE_CONTROL"
1005
+ ],
1006
+ [
1007
+ 21,
1008
+ 90,
1009
+ 0,
1010
+ 130,
1011
+ 2,
1012
+ "IMAGE"
1013
+ ]
1014
+ ],
1015
+ "groups": [],
1016
+ "config": {},
1017
+ "extra": {
1018
+ "ds": {
1019
+ "scale": 0.55,
1020
+ "offset": [
1021
+ 550,
1022
+ 150
1023
+ ]
1024
+ }
1025
+ },
1026
+ "version": 0.4
1027
+ }
examples/v1/jetpack-cat-control.png ADDED

Git LFS Details

  • SHA256: 0ad6ce18316831a1908a66afc4f82a0b273e714b40c32505547dd70ce489d70a
  • Pointer size: 132 Bytes
  • Size of remote file: 2.59 MB
examples/v1/jetpack-cat-input.png ADDED

Git LFS Details

  • SHA256: 870075c52ec12b031794e0018a43aa0df74fe2ce4000b121ab0fc7faf2db1a93
  • Pointer size: 131 Bytes
  • Size of remote file: 292 kB
examples/v1/jetpack-cat-v1.png ADDED

Git LFS Details

  • SHA256: 3045fed4dbe7d1f4256bc6cb21f1beff387990b636bba0cb2694cca82979f090
  • Pointer size: 132 Bytes
  • Size of remote file: 3.09 MB
examples/v1/jetpack-cat.api.json ADDED
@@ -0,0 +1,194 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "1": {
3
+ "class_type": "UNETLoader",
4
+ "inputs": {
5
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
6
+ "weight_dtype": "default"
7
+ }
8
+ },
9
+ "3": {
10
+ "class_type": "CLIPLoader",
11
+ "inputs": {
12
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
13
+ "type": "moss_image",
14
+ "device": "default"
15
+ }
16
+ },
17
+ "4": {
18
+ "class_type": "VAELoader",
19
+ "inputs": {
20
+ "vae_name": "flux2-vae.safetensors"
21
+ }
22
+ },
23
+ "5": {
24
+ "class_type": "CLIPTextEncode",
25
+ "inputs": {
26
+ "text": "A cute orange tabby cat with white paws and chest sits upright, featuring large green eyes and blue goggles with green lenses on its head. The cat is riding a complex blue mechanical hoverboard, wearing a brown harness with a blue triangle emblem. Blue rocket engines with yellow and blue flames shoot from the sides, and yellow wheels with glowing centers are visible. The image is rendered in a vibrant 3D style, featuring dynamic lighting and soft shadows. The background is transparent black.",
27
+ "clip": [
28
+ "3",
29
+ 0
30
+ ]
31
+ }
32
+ },
33
+ "6": {
34
+ "class_type": "CLIPTextEncode",
35
+ "inputs": {
36
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact",
37
+ "clip": [
38
+ "3",
39
+ 0
40
+ ]
41
+ }
42
+ },
43
+ "9": {
44
+ "class_type": "VAEDecode",
45
+ "inputs": {
46
+ "samples": [
47
+ "20",
48
+ 0
49
+ ],
50
+ "vae": [
51
+ "4",
52
+ 0
53
+ ]
54
+ }
55
+ },
56
+ "20": {
57
+ "class_type": "SamplerCustom",
58
+ "inputs": {
59
+ "add_noise": true,
60
+ "noise_seed": 994529790995974,
61
+ "cfg": 5.0,
62
+ "model": [
63
+ "130",
64
+ 0
65
+ ],
66
+ "positive": [
67
+ "5",
68
+ 0
69
+ ],
70
+ "negative": [
71
+ "6",
72
+ 0
73
+ ],
74
+ "sampler": [
75
+ "86",
76
+ 0
77
+ ],
78
+ "sigmas": [
79
+ "22",
80
+ 0
81
+ ],
82
+ "latent_image": [
83
+ "124",
84
+ 0
85
+ ]
86
+ }
87
+ },
88
+ "22": {
89
+ "class_type": "BasicScheduler",
90
+ "inputs": {
91
+ "scheduler": "sgm_uniform",
92
+ "steps": 15,
93
+ "denoise": 1.0,
94
+ "model": [
95
+ "130",
96
+ 0
97
+ ]
98
+ }
99
+ },
100
+ "36": {
101
+ "class_type": "ModelSamplingAuraFlow",
102
+ "inputs": {
103
+ "shift": 3.0,
104
+ "sampling": "flow",
105
+ "model": [
106
+ "88",
107
+ 0
108
+ ]
109
+ }
110
+ },
111
+ "55": {
112
+ "class_type": "LoadImage",
113
+ "inputs": {
114
+ "image": "jetpack-cat-input.png"
115
+ }
116
+ },
117
+ "85": {
118
+ "class_type": "SaveImage",
119
+ "inputs": {
120
+ "filename_prefix": "MOSS-LLLite-V1/jetpack-cat",
121
+ "images": [
122
+ "9",
123
+ 0
124
+ ]
125
+ }
126
+ },
127
+ "86": {
128
+ "class_type": "KSamplerSelect",
129
+ "inputs": {
130
+ "sampler_name": "euler"
131
+ }
132
+ },
133
+ "88": {
134
+ "class_type": "LoraLoaderModelOnly",
135
+ "inputs": {
136
+ "lora_name": "Coffmeteor-sft-15step-cfg3.safetensors",
137
+ "strength_model": 1.0,
138
+ "model": [
139
+ "1",
140
+ 0
141
+ ]
142
+ }
143
+ },
144
+ "90": {
145
+ "class_type": "ImageScaleBy",
146
+ "inputs": {
147
+ "upscale_method": "lanczos",
148
+ "scale_by": 4.0,
149
+ "image": [
150
+ "55",
151
+ 0
152
+ ]
153
+ }
154
+ },
155
+ "124": {
156
+ "class_type": "VAEEncode",
157
+ "inputs": {
158
+ "pixels": [
159
+ "90",
160
+ 0
161
+ ],
162
+ "vae": [
163
+ "4",
164
+ 0
165
+ ]
166
+ }
167
+ },
168
+ "130": {
169
+ "class_type": "MossLLLiteApply",
170
+ "inputs": {
171
+ "strength": 1.0,
172
+ "start_percent": 0.0,
173
+ "end_percent": 1.0,
174
+ "model": [
175
+ "36",
176
+ 0
177
+ ],
178
+ "control": [
179
+ "131",
180
+ 0
181
+ ],
182
+ "image": [
183
+ "90",
184
+ 0
185
+ ]
186
+ }
187
+ },
188
+ "131": {
189
+ "class_type": "MossLLLiteLoader",
190
+ "inputs": {
191
+ "name": "MOSS-LLLite-Tile-V1.safetensors"
192
+ }
193
+ }
194
+ }
examples/v1/jetpack-cat.workflow.json ADDED
@@ -0,0 +1,975 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "id": "04b39ccf-4740-4ee1-a003-513bcdbc3b02",
3
+ "revision": 0,
4
+ "last_node_id": 131,
5
+ "last_link_id": 20,
6
+ "nodes": [
7
+ {
8
+ "id": 1,
9
+ "type": "UNETLoader",
10
+ "pos": [
11
+ -423.26391629976365,
12
+ -689.4382881990866
13
+ ],
14
+ "size": [
15
+ 500,
16
+ 90
17
+ ],
18
+ "flags": {},
19
+ "order": 0,
20
+ "mode": 0,
21
+ "inputs": [],
22
+ "outputs": [
23
+ {
24
+ "name": "MODEL",
25
+ "type": "MODEL",
26
+ "slot_index": 0,
27
+ "links": [
28
+ 14
29
+ ]
30
+ }
31
+ ],
32
+ "properties": {
33
+ "cnr_id": "comfy-core",
34
+ "ver": "0.44.0",
35
+ "Node name for S&R": "UNETLoader",
36
+ "models": [
37
+ {
38
+ "name": "moss_image_sft_int8_convrot.safetensors",
39
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors",
40
+ "directory": "diffusion_models"
41
+ }
42
+ ]
43
+ },
44
+ "widgets_values": [
45
+ "moss-image/moss_image_sft_int8_convrot.safetensors",
46
+ "default"
47
+ ],
48
+ "widgets_values_named": {
49
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
50
+ "weight_dtype": "default"
51
+ }
52
+ },
53
+ {
54
+ "id": 3,
55
+ "type": "CLIPLoader",
56
+ "pos": [
57
+ -440,
58
+ -50
59
+ ],
60
+ "size": [
61
+ 500,
62
+ 110
63
+ ],
64
+ "flags": {},
65
+ "order": 1,
66
+ "mode": 0,
67
+ "inputs": [],
68
+ "outputs": [
69
+ {
70
+ "name": "CLIP",
71
+ "type": "CLIP",
72
+ "slot_index": 0,
73
+ "links": [
74
+ 1,
75
+ 2
76
+ ]
77
+ }
78
+ ],
79
+ "properties": {
80
+ "cnr_id": "comfy-core",
81
+ "ver": "0.44.0",
82
+ "Node name for S&R": "CLIPLoader",
83
+ "models": [
84
+ {
85
+ "name": "qwen3.5_9b_int8_convrot.safetensors",
86
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors",
87
+ "directory": "text_encoders"
88
+ }
89
+ ]
90
+ },
91
+ "widgets_values": [
92
+ "qwen3.5_9b_bf16.safetensors",
93
+ "moss_image",
94
+ "default"
95
+ ],
96
+ "widgets_values_named": {
97
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
98
+ "type": "moss_image",
99
+ "device": "default"
100
+ }
101
+ },
102
+ {
103
+ "id": 4,
104
+ "type": "VAELoader",
105
+ "pos": [
106
+ -440,
107
+ 130
108
+ ],
109
+ "size": [
110
+ 500,
111
+ 60
112
+ ],
113
+ "flags": {},
114
+ "order": 2,
115
+ "mode": 0,
116
+ "inputs": [],
117
+ "outputs": [
118
+ {
119
+ "name": "VAE",
120
+ "type": "VAE",
121
+ "slot_index": 0,
122
+ "links": [
123
+ 4,
124
+ 17
125
+ ]
126
+ }
127
+ ],
128
+ "properties": {
129
+ "cnr_id": "comfy-core",
130
+ "ver": "0.44.0",
131
+ "Node name for S&R": "VAELoader",
132
+ "models": [
133
+ {
134
+ "name": "flux2-vae_small.safetensors",
135
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors",
136
+ "directory": "vae"
137
+ }
138
+ ]
139
+ },
140
+ "widgets_values": [
141
+ "flux2-vae.safetensors"
142
+ ],
143
+ "widgets_values_named": {
144
+ "vae_name": "flux2-vae.safetensors"
145
+ }
146
+ },
147
+ {
148
+ "id": 5,
149
+ "type": "CLIPTextEncode",
150
+ "pos": [
151
+ 185.95627952212953,
152
+ -36.734098034606404
153
+ ],
154
+ "size": [
155
+ 470,
156
+ 240
157
+ ],
158
+ "flags": {},
159
+ "order": 3,
160
+ "mode": 0,
161
+ "inputs": [
162
+ {
163
+ "name": "clip",
164
+ "type": "CLIP",
165
+ "link": 1
166
+ }
167
+ ],
168
+ "outputs": [
169
+ {
170
+ "name": "CONDITIONING",
171
+ "type": "CONDITIONING",
172
+ "slot_index": 0,
173
+ "links": [
174
+ 6
175
+ ]
176
+ }
177
+ ],
178
+ "properties": {
179
+ "cnr_id": "comfy-core",
180
+ "ver": "0.44.0",
181
+ "Node name for S&R": "CLIPTextEncode"
182
+ },
183
+ "widgets_values": [
184
+ "A cute orange tabby cat with white paws and chest sits upright, featuring large green eyes and blue goggles with green lenses on its head. The cat is riding a complex blue mechanical hoverboard, wearing a brown harness with a blue triangle emblem. Blue rocket engines with yellow and blue flames shoot from the sides, and yellow wheels with glowing centers are visible. The image is rendered in a vibrant 3D style, featuring dynamic lighting and soft shadows. The background is transparent black."
185
+ ],
186
+ "widgets_values_named": {
187
+ "text": "A cute orange tabby cat with white paws and chest sits upright, featuring large green eyes and blue goggles with green lenses on its head. The cat is riding a complex blue mechanical hoverboard, wearing a brown harness with a blue triangle emblem. Blue rocket engines with yellow and blue flames shoot from the sides, and yellow wheels with glowing centers are visible. The image is rendered in a vibrant 3D style, featuring dynamic lighting and soft shadows. The background is transparent black."
188
+ },
189
+ "color": "#232",
190
+ "bgcolor": "#353"
191
+ },
192
+ {
193
+ "id": 6,
194
+ "type": "CLIPTextEncode",
195
+ "pos": [
196
+ 150,
197
+ 250
198
+ ],
199
+ "size": [
200
+ 480,
201
+ 110
202
+ ],
203
+ "flags": {},
204
+ "order": 4,
205
+ "mode": 0,
206
+ "inputs": [
207
+ {
208
+ "name": "clip",
209
+ "type": "CLIP",
210
+ "link": 2
211
+ }
212
+ ],
213
+ "outputs": [
214
+ {
215
+ "name": "CONDITIONING",
216
+ "type": "CONDITIONING",
217
+ "slot_index": 0,
218
+ "links": [
219
+ 7
220
+ ]
221
+ }
222
+ ],
223
+ "properties": {
224
+ "cnr_id": "comfy-core",
225
+ "ver": "0.44.0",
226
+ "Node name for S&R": "CLIPTextEncode"
227
+ },
228
+ "widgets_values": [
229
+ "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
230
+ ],
231
+ "widgets_values_named": {
232
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
233
+ },
234
+ "color": "#223",
235
+ "bgcolor": "#335"
236
+ },
237
+ {
238
+ "id": 9,
239
+ "type": "VAEDecode",
240
+ "pos": [
241
+ 1910,
242
+ -520
243
+ ],
244
+ "size": [
245
+ 210,
246
+ 50
247
+ ],
248
+ "flags": {},
249
+ "order": 5,
250
+ "mode": 0,
251
+ "inputs": [
252
+ {
253
+ "name": "samples",
254
+ "type": "LATENT",
255
+ "link": 3
256
+ },
257
+ {
258
+ "name": "vae",
259
+ "type": "VAE",
260
+ "link": 4
261
+ }
262
+ ],
263
+ "outputs": [
264
+ {
265
+ "name": "IMAGE",
266
+ "type": "IMAGE",
267
+ "slot_index": 0,
268
+ "links": [
269
+ 13
270
+ ]
271
+ }
272
+ ],
273
+ "properties": {
274
+ "cnr_id": "comfy-core",
275
+ "ver": "0.44.0",
276
+ "Node name for S&R": "VAEDecode"
277
+ },
278
+ "color": "#322",
279
+ "bgcolor": "#533"
280
+ },
281
+ {
282
+ "id": 20,
283
+ "type": "SamplerCustom",
284
+ "pos": [
285
+ 1518.3539918681,
286
+ -304.93798438717295
287
+ ],
288
+ "size": [
289
+ 280,
290
+ 500
291
+ ],
292
+ "flags": {},
293
+ "order": 6,
294
+ "mode": 0,
295
+ "inputs": [
296
+ {
297
+ "name": "model",
298
+ "type": "MODEL",
299
+ "link": 5
300
+ },
301
+ {
302
+ "name": "positive",
303
+ "type": "CONDITIONING",
304
+ "link": 6
305
+ },
306
+ {
307
+ "name": "negative",
308
+ "type": "CONDITIONING",
309
+ "link": 7
310
+ },
311
+ {
312
+ "name": "sampler",
313
+ "type": "SAMPLER",
314
+ "link": 8
315
+ },
316
+ {
317
+ "name": "sigmas",
318
+ "type": "SIGMAS",
319
+ "link": 9
320
+ },
321
+ {
322
+ "name": "latent_image",
323
+ "type": "LATENT",
324
+ "link": 10
325
+ }
326
+ ],
327
+ "outputs": [
328
+ {
329
+ "name": "output",
330
+ "type": "LATENT",
331
+ "links": [
332
+ 3
333
+ ]
334
+ },
335
+ {
336
+ "name": "denoised_output",
337
+ "type": "LATENT",
338
+ "links": null
339
+ }
340
+ ],
341
+ "properties": {
342
+ "cnr_id": "comfy-core",
343
+ "ver": "0.44.0",
344
+ "Node name for S&R": "SamplerCustom"
345
+ },
346
+ "widgets_values": [
347
+ true,
348
+ 994529790995974,
349
+ "fixed",
350
+ 5
351
+ ],
352
+ "widgets_values_named": {
353
+ "add_noise": true,
354
+ "noise_seed": 994529790995974,
355
+ "control_after_generate": "fixed",
356
+ "cfg": 5.0
357
+ }
358
+ },
359
+ {
360
+ "id": 22,
361
+ "type": "BasicScheduler",
362
+ "pos": [
363
+ 1169.6140490851362,
364
+ 1.7699193319720672
365
+ ],
366
+ "size": [
367
+ 270,
368
+ 110
369
+ ],
370
+ "flags": {},
371
+ "order": 7,
372
+ "mode": 0,
373
+ "inputs": [
374
+ {
375
+ "name": "model",
376
+ "type": "MODEL",
377
+ "link": 11
378
+ }
379
+ ],
380
+ "outputs": [
381
+ {
382
+ "name": "SIGMAS",
383
+ "type": "SIGMAS",
384
+ "links": [
385
+ 9
386
+ ]
387
+ }
388
+ ],
389
+ "properties": {
390
+ "cnr_id": "comfy-core",
391
+ "ver": "0.44.0",
392
+ "Node name for S&R": "BasicScheduler"
393
+ },
394
+ "widgets_values": [
395
+ "sgm_uniform",
396
+ 15,
397
+ 1
398
+ ],
399
+ "widgets_values_named": {
400
+ "scheduler": "sgm_uniform",
401
+ "steps": 15,
402
+ "denoise": 1.0
403
+ }
404
+ },
405
+ {
406
+ "id": 36,
407
+ "type": "ModelSamplingAuraFlow",
408
+ "pos": [
409
+ 216.72263354772872,
410
+ -300
411
+ ],
412
+ "size": [
413
+ 400,
414
+ 90
415
+ ],
416
+ "flags": {},
417
+ "order": 8,
418
+ "mode": 0,
419
+ "inputs": [
420
+ {
421
+ "name": "model",
422
+ "type": "MODEL",
423
+ "link": 12
424
+ }
425
+ ],
426
+ "outputs": [
427
+ {
428
+ "name": "MODEL",
429
+ "type": "MODEL",
430
+ "links": [
431
+ 18
432
+ ]
433
+ }
434
+ ],
435
+ "properties": {
436
+ "cnr_id": "comfy-core",
437
+ "ver": "0.44.0",
438
+ "Node name for S&R": "ModelSamplingAuraFlow"
439
+ },
440
+ "widgets_values": [
441
+ 3,
442
+ "flow"
443
+ ],
444
+ "widgets_values_named": {
445
+ "shift": 3.0,
446
+ "sampling": "flow"
447
+ }
448
+ },
449
+ {
450
+ "id": 55,
451
+ "type": "LoadImage",
452
+ "pos": [
453
+ -400,
454
+ 630
455
+ ],
456
+ "size": [
457
+ 290,
458
+ 110
459
+ ],
460
+ "flags": {},
461
+ "order": 9,
462
+ "mode": 0,
463
+ "inputs": [],
464
+ "outputs": [
465
+ {
466
+ "name": "IMAGE",
467
+ "type": "IMAGE",
468
+ "links": [
469
+ 15
470
+ ]
471
+ },
472
+ {
473
+ "name": "MASK",
474
+ "type": "MASK",
475
+ "links": null
476
+ }
477
+ ],
478
+ "properties": {
479
+ "cnr_id": "comfy-core",
480
+ "ver": "0.44.0",
481
+ "Node name for S&R": "LoadImage"
482
+ },
483
+ "widgets_values": [
484
+ "jetpack-cat-input.png",
485
+ "image"
486
+ ],
487
+ "widgets_values_named": {
488
+ "image": "jetpack-cat-input.png",
489
+ "upload": "image"
490
+ },
491
+ "color": "#322",
492
+ "bgcolor": "#533"
493
+ },
494
+ {
495
+ "id": 85,
496
+ "type": "SaveImage",
497
+ "pos": [
498
+ 3307.317013871774,
499
+ 597.8636595391883
500
+ ],
501
+ "size": [
502
+ 420,
503
+ 470
504
+ ],
505
+ "flags": {},
506
+ "order": 10,
507
+ "mode": 0,
508
+ "inputs": [
509
+ {
510
+ "name": "images",
511
+ "type": "IMAGE",
512
+ "link": 13
513
+ }
514
+ ],
515
+ "outputs": [
516
+ {
517
+ "name": "images",
518
+ "type": "IMAGE",
519
+ "links": null
520
+ }
521
+ ],
522
+ "properties": {
523
+ "cnr_id": "comfy-core",
524
+ "ver": "0.44.0",
525
+ "Node name for S&R": "SaveImage"
526
+ },
527
+ "widgets_values": [
528
+ "MOSS-LLLite-V1/jetpack-cat"
529
+ ],
530
+ "widgets_values_named": {
531
+ "filename_prefix": "MOSS-LLLite-V1/jetpack-cat"
532
+ }
533
+ },
534
+ {
535
+ "id": 86,
536
+ "type": "KSamplerSelect",
537
+ "pos": [
538
+ 1170.4423572560272,
539
+ -260.582916386391
540
+ ],
541
+ "size": [
542
+ 270,
543
+ 58
544
+ ],
545
+ "flags": {},
546
+ "order": 11,
547
+ "mode": 0,
548
+ "inputs": [],
549
+ "outputs": [
550
+ {
551
+ "name": "SAMPLER",
552
+ "type": "SAMPLER",
553
+ "links": [
554
+ 8
555
+ ]
556
+ }
557
+ ],
558
+ "properties": {
559
+ "cnr_id": "comfy-core",
560
+ "ver": "0.44.0",
561
+ "Node name for S&R": "KSamplerSelect"
562
+ },
563
+ "widgets_values": [
564
+ "euler"
565
+ ],
566
+ "widgets_values_named": {
567
+ "sampler_name": "euler"
568
+ }
569
+ },
570
+ {
571
+ "id": 88,
572
+ "type": "LoraLoaderModelOnly",
573
+ "pos": [
574
+ -447.13559989481814,
575
+ -263.5408892596166
576
+ ],
577
+ "size": [
578
+ 500,
579
+ 120
580
+ ],
581
+ "flags": {},
582
+ "order": 12,
583
+ "mode": 0,
584
+ "inputs": [
585
+ {
586
+ "name": "model",
587
+ "type": "MODEL",
588
+ "link": 14
589
+ }
590
+ ],
591
+ "outputs": [
592
+ {
593
+ "name": "MODEL",
594
+ "type": "MODEL",
595
+ "links": [
596
+ 12
597
+ ]
598
+ }
599
+ ],
600
+ "properties": {
601
+ "cnr_id": "comfy-core",
602
+ "ver": "0.44.0",
603
+ "Node name for S&R": "LoraLoaderModelOnly",
604
+ "models": [
605
+ {
606
+ "name": "moss_image_color_depth_control.safetensors",
607
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
608
+ "directory": "loras"
609
+ }
610
+ ]
611
+ },
612
+ "widgets_values": [
613
+ "Coffmeteor-sft-15step-cfg3.safetensors",
614
+ 1
615
+ ],
616
+ "widgets_values_named": {
617
+ "lora_name": "Coffmeteor-sft-15step-cfg3.safetensors",
618
+ "strength_model": 1.0
619
+ }
620
+ },
621
+ {
622
+ "id": 90,
623
+ "type": "ImageScaleBy",
624
+ "pos": [
625
+ 321.18555329294736,
626
+ 683.4828563588695
627
+ ],
628
+ "size": [
629
+ 270,
630
+ 82
631
+ ],
632
+ "flags": {},
633
+ "order": 13,
634
+ "mode": 0,
635
+ "inputs": [
636
+ {
637
+ "name": "image",
638
+ "type": "IMAGE",
639
+ "link": 15
640
+ }
641
+ ],
642
+ "outputs": [
643
+ {
644
+ "name": "IMAGE",
645
+ "type": "IMAGE",
646
+ "links": [
647
+ 16,
648
+ 20
649
+ ]
650
+ }
651
+ ],
652
+ "properties": {
653
+ "cnr_id": "comfy-core",
654
+ "ver": "0.44.0",
655
+ "Node name for S&R": "ImageScaleBy"
656
+ },
657
+ "widgets_values": [
658
+ "lanczos",
659
+ 4
660
+ ],
661
+ "widgets_values_named": {
662
+ "upscale_method": "lanczos",
663
+ "scale_by": 4.0
664
+ }
665
+ },
666
+ {
667
+ "id": 124,
668
+ "type": "VAEEncode",
669
+ "pos": [
670
+ 1149.2279511667382,
671
+ 242.2625195138657
672
+ ],
673
+ "size": [
674
+ 140,
675
+ 46
676
+ ],
677
+ "flags": {},
678
+ "order": 14,
679
+ "mode": 0,
680
+ "inputs": [
681
+ {
682
+ "name": "pixels",
683
+ "type": "IMAGE",
684
+ "link": 16
685
+ },
686
+ {
687
+ "name": "vae",
688
+ "type": "VAE",
689
+ "link": 17
690
+ }
691
+ ],
692
+ "outputs": [
693
+ {
694
+ "name": "LATENT",
695
+ "type": "LATENT",
696
+ "links": [
697
+ 10
698
+ ]
699
+ }
700
+ ],
701
+ "properties": {
702
+ "cnr_id": "comfy-core",
703
+ "ver": "0.44.0",
704
+ "Node name for S&R": "VAEEncode"
705
+ }
706
+ },
707
+ {
708
+ "id": 130,
709
+ "type": "MossLLLiteApply",
710
+ "pos": [
711
+ 754.4053011472657,
712
+ 147.5866285701938
713
+ ],
714
+ "size": [
715
+ 270,
716
+ 146
717
+ ],
718
+ "flags": {},
719
+ "order": 15,
720
+ "mode": 0,
721
+ "inputs": [
722
+ {
723
+ "name": "model",
724
+ "type": "MODEL",
725
+ "link": 18
726
+ },
727
+ {
728
+ "name": "control",
729
+ "type": "MOSS_LLLITE_CONTROL",
730
+ "link": 19
731
+ },
732
+ {
733
+ "name": "image",
734
+ "type": "IMAGE",
735
+ "link": 20
736
+ }
737
+ ],
738
+ "outputs": [
739
+ {
740
+ "name": "MODEL",
741
+ "type": "MODEL",
742
+ "links": [
743
+ 5,
744
+ 11
745
+ ]
746
+ }
747
+ ],
748
+ "properties": {
749
+ "aux_id": "X-absurd/ComfyUI-MOSS-LLLite",
750
+ "ver": "20842b2b9d08511abd32404a0be24a51c9352109",
751
+ "Node name for S&R": "MossLLLiteApply"
752
+ },
753
+ "widgets_values": [
754
+ 1,
755
+ 0,
756
+ 1
757
+ ],
758
+ "widgets_values_named": {
759
+ "strength": 1.0,
760
+ "start_percent": 0.0,
761
+ "end_percent": 1.0
762
+ }
763
+ },
764
+ {
765
+ "id": 131,
766
+ "type": "MossLLLiteLoader",
767
+ "pos": [
768
+ 757.1064855291617,
769
+ -12.82076859408437
770
+ ],
771
+ "size": [
772
+ 270,
773
+ 58
774
+ ],
775
+ "flags": {},
776
+ "order": 16,
777
+ "mode": 0,
778
+ "inputs": [],
779
+ "outputs": [
780
+ {
781
+ "name": "MOSS_LLLITE_CONTROL",
782
+ "type": "MOSS_LLLITE_CONTROL",
783
+ "links": [
784
+ 19
785
+ ]
786
+ }
787
+ ],
788
+ "properties": {
789
+ "aux_id": "X-absurd/ComfyUI-MOSS-LLLite",
790
+ "ver": "20842b2b9d08511abd32404a0be24a51c9352109",
791
+ "Node name for S&R": "MossLLLiteLoader"
792
+ },
793
+ "widgets_values": [
794
+ "MOSS-LLLite-Tile-V1.safetensors"
795
+ ],
796
+ "widgets_values_named": {
797
+ "name": "MOSS-LLLite-Tile-V1.safetensors"
798
+ }
799
+ }
800
+ ],
801
+ "links": [
802
+ [
803
+ 1,
804
+ 3,
805
+ 0,
806
+ 5,
807
+ 0,
808
+ "CLIP"
809
+ ],
810
+ [
811
+ 2,
812
+ 3,
813
+ 0,
814
+ 6,
815
+ 0,
816
+ "CLIP"
817
+ ],
818
+ [
819
+ 3,
820
+ 20,
821
+ 0,
822
+ 9,
823
+ 0,
824
+ "LATENT"
825
+ ],
826
+ [
827
+ 4,
828
+ 4,
829
+ 0,
830
+ 9,
831
+ 1,
832
+ "VAE"
833
+ ],
834
+ [
835
+ 5,
836
+ 130,
837
+ 0,
838
+ 20,
839
+ 0,
840
+ "MODEL"
841
+ ],
842
+ [
843
+ 6,
844
+ 5,
845
+ 0,
846
+ 20,
847
+ 1,
848
+ "CONDITIONING"
849
+ ],
850
+ [
851
+ 7,
852
+ 6,
853
+ 0,
854
+ 20,
855
+ 2,
856
+ "CONDITIONING"
857
+ ],
858
+ [
859
+ 8,
860
+ 86,
861
+ 0,
862
+ 20,
863
+ 3,
864
+ "SAMPLER"
865
+ ],
866
+ [
867
+ 9,
868
+ 22,
869
+ 0,
870
+ 20,
871
+ 4,
872
+ "SIGMAS"
873
+ ],
874
+ [
875
+ 10,
876
+ 124,
877
+ 0,
878
+ 20,
879
+ 5,
880
+ "LATENT"
881
+ ],
882
+ [
883
+ 11,
884
+ 130,
885
+ 0,
886
+ 22,
887
+ 0,
888
+ "MODEL"
889
+ ],
890
+ [
891
+ 12,
892
+ 88,
893
+ 0,
894
+ 36,
895
+ 0,
896
+ "MODEL"
897
+ ],
898
+ [
899
+ 13,
900
+ 9,
901
+ 0,
902
+ 85,
903
+ 0,
904
+ "IMAGE"
905
+ ],
906
+ [
907
+ 14,
908
+ 1,
909
+ 0,
910
+ 88,
911
+ 0,
912
+ "MODEL"
913
+ ],
914
+ [
915
+ 15,
916
+ 55,
917
+ 0,
918
+ 90,
919
+ 0,
920
+ "IMAGE"
921
+ ],
922
+ [
923
+ 16,
924
+ 90,
925
+ 0,
926
+ 124,
927
+ 0,
928
+ "IMAGE"
929
+ ],
930
+ [
931
+ 17,
932
+ 4,
933
+ 0,
934
+ 124,
935
+ 1,
936
+ "VAE"
937
+ ],
938
+ [
939
+ 18,
940
+ 36,
941
+ 0,
942
+ 130,
943
+ 0,
944
+ "MODEL"
945
+ ],
946
+ [
947
+ 19,
948
+ 131,
949
+ 0,
950
+ 130,
951
+ 1,
952
+ "MOSS_LLLITE_CONTROL"
953
+ ],
954
+ [
955
+ 20,
956
+ 90,
957
+ 0,
958
+ 130,
959
+ 2,
960
+ "IMAGE"
961
+ ]
962
+ ],
963
+ "groups": [],
964
+ "config": {},
965
+ "extra": {
966
+ "ds": {
967
+ "scale": 0.55,
968
+ "offset": [
969
+ 550,
970
+ 150
971
+ ]
972
+ }
973
+ },
974
+ "version": 0.4
975
+ }
release-manifest.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "release": "V1",
3
+ "weights": {
4
+ "path": "V1/MOSS-LLLite-Tile-V1.safetensors",
5
+ "sha256": "1dc931f3f5533fd96768c64886d87537e7bd19f4f16c1893fb40a459758ae601",
6
+ "bytes": 103462080
7
+ },
8
+ "node_package": {
9
+ "path": "ComfyUI-MOSS-LLLite.zip",
10
+ "version": "0.1.2",
11
+ "source_commit": "c8c076a7d82a4b5f11542899d0c5011c16e1e825",
12
+ "sha256": "7c2fa276e04b576ba2bb46c114ea8aef6f86fa7435fc300864d391b95d7a3999"
13
+ },
14
+ "preserved_legacy_weights": {
15
+ "beta2/epoch1/adapter_model.safetensors": {
16
+ "sha256": "21856898240e6f21e2aee8e856726948cedfde35a8d899c11b05c2105d4af4b9",
17
+ "bytes": 268221784
18
+ },
19
+ "beta3/epoch1/adapter_model.safetensors": {
20
+ "sha256": "45d81afa8b5afbac9b8813106d195d9778013c7b10f8c535db80e7da96eacd0c",
21
+ "bytes": 71148664
22
+ },
23
+ "beta3/epoch2/adapter_model.safetensors": {
24
+ "sha256": "e2d4de5d18cba07cd94d999c0d60ad6849676991a996f69f7a9b3b2b192164d8",
25
+ "bytes": 71148664
26
+ }
27
+ },
28
+ "development_iterations": 4,
29
+ "examples": {
30
+ "directory": "examples/v1",
31
+ "count": 4,
32
+ "settings": "examples/v1/examples.json",
33
+ "checksums": "examples/v1/SHA256SUMS.txt",
34
+ "embedded_workflows": true
35
+ }
36
+ }
workflow/moss_image_tile-beta3-color-match.json ADDED
@@ -0,0 +1,1739 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "id": "5140b129-e927-49a6-816e-668c69d43008",
3
+ "revision": 0,
4
+ "last_node_id": 129,
5
+ "last_link_id": 268,
6
+ "nodes": [
7
+ {
8
+ "id": 1,
9
+ "type": "UNETLoader",
10
+ "pos": [
11
+ -423.26391629976365,
12
+ -689.4382881990866
13
+ ],
14
+ "size": [
15
+ 500,
16
+ 90
17
+ ],
18
+ "flags": {},
19
+ "order": 0,
20
+ "mode": 0,
21
+ "inputs": [],
22
+ "outputs": [
23
+ {
24
+ "name": "MODEL",
25
+ "type": "MODEL",
26
+ "slot_index": 0,
27
+ "links": [
28
+ 91
29
+ ]
30
+ }
31
+ ],
32
+ "properties": {
33
+ "cnr_id": "comfy-core",
34
+ "ver": "0.44.0",
35
+ "Node name for S&R": "UNETLoader",
36
+ "models": [
37
+ {
38
+ "name": "moss_image_sft_int8_convrot.safetensors",
39
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors",
40
+ "directory": "diffusion_models"
41
+ }
42
+ ]
43
+ },
44
+ "widgets_values": [
45
+ "moss-image/moss_image_sft_int8_convrot.safetensors",
46
+ "default"
47
+ ],
48
+ "widgets_values_named": {
49
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
50
+ "weight_dtype": "default"
51
+ }
52
+ },
53
+ {
54
+ "id": 2,
55
+ "type": "SkipLayerGuidanceDiTSimple",
56
+ "pos": [
57
+ 620,
58
+ -460
59
+ ],
60
+ "size": [
61
+ 400,
62
+ 130
63
+ ],
64
+ "flags": {},
65
+ "order": 21,
66
+ "mode": 4,
67
+ "showAdvanced": true,
68
+ "inputs": [
69
+ {
70
+ "name": "model",
71
+ "type": "MODEL",
72
+ "link": 69
73
+ }
74
+ ],
75
+ "outputs": [
76
+ {
77
+ "name": "MODEL",
78
+ "type": "MODEL",
79
+ "slot_index": 0,
80
+ "links": [
81
+ 255
82
+ ]
83
+ }
84
+ ],
85
+ "title": "SkipLayerGuidanceDiTSimple (uncond skips block 22)",
86
+ "properties": {
87
+ "cnr_id": "comfy-core",
88
+ "ver": "0.44.0",
89
+ "Node name for S&R": "SkipLayerGuidanceDiTSimple"
90
+ },
91
+ "widgets_values": [
92
+ "",
93
+ "24",
94
+ 0,
95
+ 1
96
+ ],
97
+ "widgets_values_named": {
98
+ "double_layers": "",
99
+ "single_layers": "24",
100
+ "start_percent": 0,
101
+ "end_percent": 1
102
+ }
103
+ },
104
+ {
105
+ "id": 3,
106
+ "type": "CLIPLoader",
107
+ "pos": [
108
+ -440,
109
+ -50
110
+ ],
111
+ "size": [
112
+ 500,
113
+ 110
114
+ ],
115
+ "flags": {},
116
+ "order": 1,
117
+ "mode": 0,
118
+ "inputs": [],
119
+ "outputs": [
120
+ {
121
+ "name": "CLIP",
122
+ "type": "CLIP",
123
+ "slot_index": 0,
124
+ "links": [
125
+ 3,
126
+ 4
127
+ ]
128
+ }
129
+ ],
130
+ "properties": {
131
+ "cnr_id": "comfy-core",
132
+ "ver": "0.44.0",
133
+ "Node name for S&R": "CLIPLoader",
134
+ "models": [
135
+ {
136
+ "name": "qwen3.5_9b_int8_convrot.safetensors",
137
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors",
138
+ "directory": "text_encoders"
139
+ }
140
+ ]
141
+ },
142
+ "widgets_values": [
143
+ "qwen3.5_9b_bf16.safetensors",
144
+ "moss_image",
145
+ "default"
146
+ ],
147
+ "widgets_values_named": {
148
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
149
+ "type": "moss_image",
150
+ "device": "default"
151
+ }
152
+ },
153
+ {
154
+ "id": 4,
155
+ "type": "VAELoader",
156
+ "pos": [
157
+ -440,
158
+ 130
159
+ ],
160
+ "size": [
161
+ 500,
162
+ 60
163
+ ],
164
+ "flags": {},
165
+ "order": 2,
166
+ "mode": 0,
167
+ "inputs": [],
168
+ "outputs": [
169
+ {
170
+ "name": "VAE",
171
+ "type": "VAE",
172
+ "slot_index": 0,
173
+ "links": [
174
+ 9,
175
+ 246,
176
+ 256
177
+ ]
178
+ }
179
+ ],
180
+ "properties": {
181
+ "cnr_id": "comfy-core",
182
+ "ver": "0.44.0",
183
+ "Node name for S&R": "VAELoader",
184
+ "models": [
185
+ {
186
+ "name": "flux2-vae_small.safetensors",
187
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors",
188
+ "directory": "vae"
189
+ }
190
+ ]
191
+ },
192
+ "widgets_values": [
193
+ "flux2-vae.safetensors"
194
+ ],
195
+ "widgets_values_named": {
196
+ "vae_name": "flux2-vae.safetensors"
197
+ }
198
+ },
199
+ {
200
+ "id": 5,
201
+ "type": "CLIPTextEncode",
202
+ "pos": [
203
+ 185.95627952212953,
204
+ -36.734098034606404
205
+ ],
206
+ "size": [
207
+ 470,
208
+ 240
209
+ ],
210
+ "flags": {},
211
+ "order": 11,
212
+ "mode": 0,
213
+ "inputs": [
214
+ {
215
+ "name": "clip",
216
+ "type": "CLIP",
217
+ "link": 3
218
+ }
219
+ ],
220
+ "outputs": [
221
+ {
222
+ "name": "CONDITIONING",
223
+ "type": "CONDITIONING",
224
+ "slot_index": 0,
225
+ "links": [
226
+ 250
227
+ ]
228
+ }
229
+ ],
230
+ "title": "Positive Prompt",
231
+ "properties": {
232
+ "cnr_id": "comfy-core",
233
+ "ver": "0.44.0",
234
+ "Node name for S&R": "CLIPTextEncode"
235
+ },
236
+ "widgets_values": [
237
+ "A highly realistic and detailed image that blends realistic photography with digital painting, featuring a grey tabby cat as the main subject. The cat is in a spread-eagle pose, flying through the air, with its front paws and hind legs stretched out. The cat is strapped into a flying drone. There are four propellers, two attached to the front paws and two to the hind legs, featuring silver hubs, black shafts, and bright orange blades. The drone frame is made of black metal rods, and red straps hold the cat securely in place. The cat's head is tilted upward, with wide eyes, small pupils, and a slightly open mouth, looking shocked or surprised. The fur is highly detailed, rendered with a painterly style that combines realism with digital art. The background is a soft, out-of-focus bokeh of green and white, suggesting a bright outdoor setting. The lighting is bright and natural, casting soft shadows that give the scene a solid, 3D illustration quality."
238
+ ],
239
+ "widgets_values_named": {
240
+ "text": "A highly realistic and detailed image that blends realistic photography with digital painting, featuring a grey tabby cat as the main subject. The cat is in a spread-eagle pose, flying through the air, with its front paws and hind legs stretched out. The cat is strapped into a flying drone. There are four propellers, two attached to the front paws and two to the hind legs, featuring silver hubs, black shafts, and bright orange blades. The drone frame is made of black metal rods, and red straps hold the cat securely in place. The cat's head is tilted upward, with wide eyes, small pupils, and a slightly open mouth, looking shocked or surprised. The fur is highly detailed, rendered with a painterly style that combines realism with digital art. The background is a soft, out-of-focus bokeh of green and white, suggesting a bright outdoor setting. The lighting is bright and natural, casting soft shadows that give the scene a solid, 3D illustration quality."
241
+ },
242
+ "color": "#232",
243
+ "bgcolor": "#353"
244
+ },
245
+ {
246
+ "id": 6,
247
+ "type": "CLIPTextEncode",
248
+ "pos": [
249
+ 150,
250
+ 250
251
+ ],
252
+ "size": [
253
+ 480,
254
+ 110
255
+ ],
256
+ "flags": {},
257
+ "order": 12,
258
+ "mode": 0,
259
+ "inputs": [
260
+ {
261
+ "name": "clip",
262
+ "type": "CLIP",
263
+ "link": 4
264
+ }
265
+ ],
266
+ "outputs": [
267
+ {
268
+ "name": "CONDITIONING",
269
+ "type": "CONDITIONING",
270
+ "slot_index": 0,
271
+ "links": [
272
+ 251
273
+ ]
274
+ }
275
+ ],
276
+ "title": "Negative Prompt",
277
+ "properties": {
278
+ "cnr_id": "comfy-core",
279
+ "ver": "0.44.0",
280
+ "Node name for S&R": "CLIPTextEncode"
281
+ },
282
+ "widgets_values": [
283
+ "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
284
+ ],
285
+ "widgets_values_named": {
286
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
287
+ },
288
+ "color": "#223",
289
+ "bgcolor": "#335"
290
+ },
291
+ {
292
+ "id": 9,
293
+ "type": "VAEDecode",
294
+ "pos": [
295
+ 1910,
296
+ -520
297
+ ],
298
+ "size": [
299
+ 210,
300
+ 50
301
+ ],
302
+ "flags": {},
303
+ "order": 25,
304
+ "mode": 0,
305
+ "inputs": [
306
+ {
307
+ "name": "samples",
308
+ "type": "LATENT",
309
+ "link": 25
310
+ },
311
+ {
312
+ "name": "vae",
313
+ "type": "VAE",
314
+ "link": 9
315
+ }
316
+ ],
317
+ "outputs": [
318
+ {
319
+ "name": "IMAGE",
320
+ "type": "IMAGE",
321
+ "slot_index": 0,
322
+ "links": [
323
+ 266
324
+ ]
325
+ }
326
+ ],
327
+ "properties": {
328
+ "cnr_id": "comfy-core",
329
+ "ver": "0.44.0",
330
+ "Node name for S&R": "VAEDecode"
331
+ },
332
+ "color": "#322",
333
+ "bgcolor": "#533"
334
+ },
335
+ {
336
+ "id": 20,
337
+ "type": "SamplerCustom",
338
+ "pos": [
339
+ 1518.3539918681,
340
+ -304.93798438717295
341
+ ],
342
+ "size": [
343
+ 280,
344
+ 500
345
+ ],
346
+ "flags": {},
347
+ "order": 24,
348
+ "mode": 0,
349
+ "inputs": [
350
+ {
351
+ "name": "model",
352
+ "type": "MODEL",
353
+ "link": 253
354
+ },
355
+ {
356
+ "name": "positive",
357
+ "type": "CONDITIONING",
358
+ "link": 250
359
+ },
360
+ {
361
+ "name": "negative",
362
+ "type": "CONDITIONING",
363
+ "link": 251
364
+ },
365
+ {
366
+ "name": "sampler",
367
+ "type": "SAMPLER",
368
+ "link": 154
369
+ },
370
+ {
371
+ "name": "sigmas",
372
+ "type": "SIGMAS",
373
+ "link": 27
374
+ },
375
+ {
376
+ "name": "latent_image",
377
+ "type": "LATENT",
378
+ "link": 247
379
+ }
380
+ ],
381
+ "outputs": [
382
+ {
383
+ "name": "output",
384
+ "type": "LATENT",
385
+ "links": [
386
+ 25
387
+ ]
388
+ },
389
+ {
390
+ "name": "denoised_output",
391
+ "type": "LATENT",
392
+ "links": null
393
+ }
394
+ ],
395
+ "properties": {
396
+ "cnr_id": "comfy-core",
397
+ "ver": "0.44.0",
398
+ "Node name for S&R": "SamplerCustom"
399
+ },
400
+ "widgets_values": [
401
+ true,
402
+ 994529790995974,
403
+ "fixed",
404
+ 5
405
+ ],
406
+ "widgets_values_named": {
407
+ "add_noise": true,
408
+ "noise_seed": 994529790995974,
409
+ "control_after_generate": "fixed",
410
+ "cfg": 5
411
+ }
412
+ },
413
+ {
414
+ "id": 22,
415
+ "type": "BasicScheduler",
416
+ "pos": [
417
+ 1169.6140490851362,
418
+ 1.7699193319720672
419
+ ],
420
+ "size": [
421
+ 270,
422
+ 110
423
+ ],
424
+ "flags": {},
425
+ "order": 23,
426
+ "mode": 0,
427
+ "inputs": [
428
+ {
429
+ "name": "model",
430
+ "type": "MODEL",
431
+ "link": 254
432
+ }
433
+ ],
434
+ "outputs": [
435
+ {
436
+ "name": "SIGMAS",
437
+ "type": "SIGMAS",
438
+ "links": [
439
+ 27
440
+ ]
441
+ }
442
+ ],
443
+ "properties": {
444
+ "cnr_id": "comfy-core",
445
+ "ver": "0.44.0",
446
+ "Node name for S&R": "BasicScheduler"
447
+ },
448
+ "widgets_values": [
449
+ "sgm_uniform",
450
+ 15,
451
+ 1
452
+ ],
453
+ "widgets_values_named": {
454
+ "scheduler": "sgm_uniform",
455
+ "steps": 15,
456
+ "denoise": 1
457
+ }
458
+ },
459
+ {
460
+ "id": 36,
461
+ "type": "ModelSamplingAuraFlow",
462
+ "pos": [
463
+ 216.72263354772872,
464
+ -300
465
+ ],
466
+ "size": [
467
+ 400,
468
+ 90
469
+ ],
470
+ "flags": {},
471
+ "order": 18,
472
+ "mode": 0,
473
+ "inputs": [
474
+ {
475
+ "name": "model",
476
+ "type": "MODEL",
477
+ "link": 260
478
+ }
479
+ ],
480
+ "outputs": [
481
+ {
482
+ "name": "MODEL",
483
+ "type": "MODEL",
484
+ "links": [
485
+ 69
486
+ ]
487
+ }
488
+ ],
489
+ "properties": {
490
+ "cnr_id": "comfy-core",
491
+ "ver": "0.44.0",
492
+ "Node name for S&R": "ModelSamplingAuraFlow"
493
+ },
494
+ "widgets_values": [
495
+ 3,
496
+ "flow"
497
+ ],
498
+ "widgets_values_named": {
499
+ "shift": 3,
500
+ "sampling": "flow"
501
+ }
502
+ },
503
+ {
504
+ "id": 42,
505
+ "type": "PreviewImage",
506
+ "pos": [
507
+ 2988.050239438019,
508
+ -319.4113609916968
509
+ ],
510
+ "size": [
511
+ 890,
512
+ 780
513
+ ],
514
+ "flags": {},
515
+ "order": 27,
516
+ "mode": 0,
517
+ "inputs": [
518
+ {
519
+ "name": "images",
520
+ "type": "IMAGE",
521
+ "link": 268
522
+ }
523
+ ],
524
+ "outputs": [
525
+ {
526
+ "name": "images",
527
+ "type": "IMAGE",
528
+ "links": [
529
+ 149,
530
+ 228
531
+ ]
532
+ }
533
+ ],
534
+ "properties": {
535
+ "cnr_id": "comfy-core",
536
+ "ver": "0.44.0",
537
+ "Node name for S&R": "PreviewImage"
538
+ }
539
+ },
540
+ {
541
+ "id": 49,
542
+ "type": "LoraLoaderModelOnly",
543
+ "pos": [
544
+ -423.26391629976365,
545
+ -439.4382881990866
546
+ ],
547
+ "size": [
548
+ 500,
549
+ 120
550
+ ],
551
+ "flags": {},
552
+ "order": 10,
553
+ "mode": 4,
554
+ "inputs": [
555
+ {
556
+ "name": "model",
557
+ "type": "MODEL",
558
+ "link": 91
559
+ }
560
+ ],
561
+ "outputs": [
562
+ {
563
+ "name": "MODEL",
564
+ "type": "MODEL",
565
+ "links": [
566
+ 157
567
+ ]
568
+ }
569
+ ],
570
+ "properties": {
571
+ "cnr_id": "comfy-core",
572
+ "ver": "0.44.0",
573
+ "Node name for S&R": "LoraLoaderModelOnly",
574
+ "models": [
575
+ {
576
+ "name": "moss_image_color_depth_control.safetensors",
577
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
578
+ "directory": "loras"
579
+ }
580
+ ]
581
+ },
582
+ "widgets_values": [
583
+ "moss/moss-tile-control/beta2/epoch1/adapter_model.safetensors",
584
+ 1
585
+ ],
586
+ "widgets_values_named": {
587
+ "lora_name": "moss/moss-tile-control/beta2/epoch1/adapter_model.safetensors",
588
+ "strength_model": 1
589
+ }
590
+ },
591
+ {
592
+ "id": 55,
593
+ "type": "LoadImage",
594
+ "pos": [
595
+ -400,
596
+ 630
597
+ ],
598
+ "size": [
599
+ 290,
600
+ 110
601
+ ],
602
+ "flags": {},
603
+ "order": 3,
604
+ "mode": 0,
605
+ "inputs": [],
606
+ "outputs": [
607
+ {
608
+ "name": "IMAGE",
609
+ "type": "IMAGE",
610
+ "links": [
611
+ 155,
612
+ 248
613
+ ]
614
+ },
615
+ {
616
+ "name": "MASK",
617
+ "type": "MASK",
618
+ "links": null
619
+ }
620
+ ],
621
+ "properties": {
622
+ "cnr_id": "comfy-core",
623
+ "ver": "0.44.0",
624
+ "Node name for S&R": "LoadImage"
625
+ },
626
+ "widgets_values": [
627
+ "3794962993127468234.jpg",
628
+ "image"
629
+ ],
630
+ "widgets_values_named": {
631
+ "image": "3794962993127468234.jpg",
632
+ "upload": "image"
633
+ },
634
+ "color": "#322",
635
+ "bgcolor": "#533"
636
+ },
637
+ {
638
+ "id": 79,
639
+ "type": "PreviewImage",
640
+ "pos": [
641
+ 733.854443722757,
642
+ 622.1325731950062
643
+ ],
644
+ "size": [
645
+ 140,
646
+ 30
647
+ ],
648
+ "flags": {},
649
+ "order": 19,
650
+ "mode": 0,
651
+ "inputs": [
652
+ {
653
+ "name": "images",
654
+ "type": "IMAGE",
655
+ "link": 264
656
+ }
657
+ ],
658
+ "outputs": [
659
+ {
660
+ "name": "images",
661
+ "type": "IMAGE",
662
+ "links": [
663
+ 150,
664
+ 257,
665
+ 267
666
+ ]
667
+ }
668
+ ],
669
+ "properties": {
670
+ "cnr_id": "comfy-core",
671
+ "ver": "0.44.0",
672
+ "Node name for S&R": "PreviewImage"
673
+ }
674
+ },
675
+ {
676
+ "id": 80,
677
+ "type": "ImageCompare",
678
+ "pos": [
679
+ 3912.6659329544727,
680
+ -282.59682725374586
681
+ ],
682
+ "size": [
683
+ 930,
684
+ 780
685
+ ],
686
+ "flags": {},
687
+ "order": 28,
688
+ "mode": 0,
689
+ "inputs": [
690
+ {
691
+ "name": "image_a",
692
+ "shape": 7,
693
+ "type": "IMAGE",
694
+ "link": 150
695
+ },
696
+ {
697
+ "name": "image_b",
698
+ "shape": 7,
699
+ "type": "IMAGE",
700
+ "link": 149
701
+ }
702
+ ],
703
+ "outputs": [],
704
+ "properties": {
705
+ "cnr_id": "comfy-core",
706
+ "ver": "0.44.0",
707
+ "Node name for S&R": "ImageCompare"
708
+ },
709
+ "widgets_values": [],
710
+ "widgets_values_named": {}
711
+ },
712
+ {
713
+ "id": 83,
714
+ "type": "MarkdownNote",
715
+ "pos": [
716
+ -1230,
717
+ -520
718
+ ],
719
+ "size": [
720
+ 710,
721
+ 890
722
+ ],
723
+ "flags": {},
724
+ "order": 4,
725
+ "mode": 0,
726
+ "inputs": [],
727
+ "outputs": [],
728
+ "title": "Note: About",
729
+ "properties": {},
730
+ "widgets_values": [
731
+ "## MOSS-Image: Color & Depth Control from a Photo\n\nDrop in a photo and the graph estimates its geometry, then generates a new image that keeps the same pose and framing. No map to prepare by hand.\n\n**How it works**\n\n1. **Load Image** (node 55) - the character reference. It is scaled to 1.0 MP, 16 px aligned (node 77), and drives the output size through **Get Image Size** (node 61) -> **Empty Latent** (node 54).\n2. **Geometry Estimation (MoGe 3)** subgraph - `Load MoGe Model` (`moge_3_vitg_fp16`) -> `Run MoGe Inference` -> `Render MoGe Geometry`. Its output is the `depth_map` of **MOSS-Image Color/Depth Control** (node 53).\n3. **Positive Prompt** (node 5) describes the character and the scene. Geometry comes from the render, everything else - identity, wardrobe, material, light, style - comes from the text.\n4. **MOSS-Image Color/Depth Control** (node 53) injects the map into the image tokens; the LoRA loaded just before it (`moss_image_color_depth_control.safetensors`, strength 1.0) is what makes the model read it. Without the LoRA the node is a no-op.\n5. **Save / Compare**: `VAE Decode` -> preview, plus **Compare Images** (node 80) holding the generated frame next to the rendered map.\n\n**Settings**\n\n`strength` 1.0 is the trained strength; `start_percent` / `end_percent` pick the part of the run that is controlled (0 to 0.6 here: the layout is fixed, the last steps render freely). Sampling follows the MOSS-Image control test graph: 25 steps, CFG 3, `er_sde` / `simple`, shift 5, skip layer guidance on layer 24, temporal score rescaling 0.95.\n\n**Useful knobs**\n\n- `Render MoGe Geometry` in the subgraph is set to `mask`: only the silhouette is used. Switch it to `depth` (`normalization = min_max`) when you want the face profile, shoulders and background falloff to be controlled too - the pose gets tighter, the prompt has less room.\n- `end_percent` 1.0 pins geometry to the last step; 0.5 frees the surface and lighting. For a mask-derived map 0.6 is a good middle.\n- The reference has **pointed elf ears**, and the geometry keeps them. Either keep that in the prompt (`pointed elf ears`) or hide them (`ears covered by hair`).\n- **Painter** (node 81) is in the graph for painting a map by hand: white near / black far for depth, one flat colour per region for colour. Connect its `IMAGE` to `depth_map` or `color_map` to override the estimate.\n- `color_map` is optional and currently unconnected. One flat colour per region, colours only act as region IDs, black means \"unspecified\"; it must match the output size (1024 x 1024 here).\n"
732
+ ],
733
+ "widgets_values_named": {
734
+ "text": "## MOSS-Image: Color & Depth Control from a Photo\n\nDrop in a photo and the graph estimates its geometry, then generates a new image that keeps the same pose and framing. No map to prepare by hand.\n\n**How it works**\n\n1. **Load Image** (node 55) - the character reference. It is scaled to 1.0 MP, 16 px aligned (node 77), and drives the output size through **Get Image Size** (node 61) -> **Empty Latent** (node 54).\n2. **Geometry Estimation (MoGe 3)** subgraph - `Load MoGe Model` (`moge_3_vitg_fp16`) -> `Run MoGe Inference` -> `Render MoGe Geometry`. Its output is the `depth_map` of **MOSS-Image Color/Depth Control** (node 53).\n3. **Positive Prompt** (node 5) describes the character and the scene. Geometry comes from the render, everything else - identity, wardrobe, material, light, style - comes from the text.\n4. **MOSS-Image Color/Depth Control** (node 53) injects the map into the image tokens; the LoRA loaded just before it (`moss_image_color_depth_control.safetensors`, strength 1.0) is what makes the model read it. Without the LoRA the node is a no-op.\n5. **Save / Compare**: `VAE Decode` -> preview, plus **Compare Images** (node 80) holding the generated frame next to the rendered map.\n\n**Settings**\n\n`strength` 1.0 is the trained strength; `start_percent` / `end_percent` pick the part of the run that is controlled (0 to 0.6 here: the layout is fixed, the last steps render freely). Sampling follows the MOSS-Image control test graph: 25 steps, CFG 3, `er_sde` / `simple`, shift 5, skip layer guidance on layer 24, temporal score rescaling 0.95.\n\n**Useful knobs**\n\n- `Render MoGe Geometry` in the subgraph is set to `mask`: only the silhouette is used. Switch it to `depth` (`normalization = min_max`) when you want the face profile, shoulders and background falloff to be controlled too - the pose gets tighter, the prompt has less room.\n- `end_percent` 1.0 pins geometry to the last step; 0.5 frees the surface and lighting. For a mask-derived map 0.6 is a good middle.\n- The reference has **pointed elf ears**, and the geometry keeps them. Either keep that in the prompt (`pointed elf ears`) or hide them (`ears covered by hair`).\n- **Painter** (node 81) is in the graph for painting a map by hand: white near / black far for depth, one flat colour per region for colour. Connect its `IMAGE` to `depth_map` or `color_map` to override the estimate.\n- `color_map` is optional and currently unconnected. One flat colour per region, colours only act as region IDs, black means \"unspecified\"; it must match the output size (1024 x 1024 here).\n"
735
+ },
736
+ "color": "#222",
737
+ "bgcolor": "#000"
738
+ },
739
+ {
740
+ "id": 84,
741
+ "type": "MarkdownNote",
742
+ "pos": [
743
+ -1970,
744
+ -520
745
+ ],
746
+ "size": [
747
+ 520,
748
+ 1250
749
+ ],
750
+ "flags": {},
751
+ "order": 5,
752
+ "mode": 0,
753
+ "inputs": [],
754
+ "outputs": [],
755
+ "title": "Note: Model Links",
756
+ "properties": {},
757
+ "widgets_values": [
758
+ "## Model Links\n\n- [Hugging Face: Comfy-Org/MOSS-Image](https://huggingface.co/Comfy-Org/MOSS-Image)\n- [Hugging Face: Comfy-Org/MoGe](https://huggingface.co/Comfy-Org/MoGe) (geometry estimation)\n- [Upstream control adapter: darask0/MOSS-Image-color-depth-control](https://huggingface.co/darask0/MOSS-Image-color-depth-control)\n\n**diffusion_models**\n\n- [moss_image_sft_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors) - 5.41 GB, the checkpoint this graph loads\n- [moss_image_sft_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_bf16.safetensors) - 10.16 GB, higher precision, slower\n- [moss_image_base_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_base_int8_convrot.safetensors) - 5.41 GB, the base checkpoint runs in the same graph\n\n**text_encoders**\n\n- [qwen3.5_9b_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors) - 9.75 GB, the encoder this graph loads\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) - 7.34 GB, drop-in swap, saves 2.4 GB\n- [qwen3.5_9b_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_bf16.safetensors) - 19.31 GB, full precision\n\n**vae**\n\n- [flux2-vae_small.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors) - 249 MB, the one this graph loads\n- [flux2-vae.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae.safetensors) - 336 MB, full size, pixel-level differences only\n\n**loras**\n\n- [moss_image_color_depth_control.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors) - 313 MB, `LoraLoaderModelOnly` at strength 1.0\n\n**geometry_estimation**\n\n- [moge_3_vitg_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitg_fp16.safetensors) - 2.50 GB, the depth estimator used by the subgraph\n- [moge_3_vitl_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitl_fp16.safetensors) - 741 MB, the lighter ViT-L variant\n\n**Disk footprint**: the graph as shipped needs **18.23 GB** (18,226,010,396 bytes) (5.41 + 9.75 + 0.25 + 0.31 + 2.50). Swapping the text encoder to `w4a8_gptq` brings it to **15.82 GB**; a bf16 checkpoint and encoder instead push it past 32 GB.\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 diffusion_models/\n│ │ └── moss_image_sft_int8_convrot.safetensors 5.41 GB\n│ ├── 📂 text_encoders/\n│ │ └── qwen3.5_9b_int8_convrot.safetensors 9.75 GB\n│ ├── 📂 vae/\n│ │ └── flux2-vae_small.safetensors 249 MB\n│ ├── 📂 loras/\n│ │ └── moss_image_color_depth_control.safetensors 313 MB\n│ └── 📂 geometry_estimation/\n│ └── moge_3_vitg_fp16.safetensors 2.50 GB\n```\n\nCore support (`MOSSImageControl`, MoGe nodes) ships from ComfyUI **0.43.0**.\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
759
+ ],
760
+ "widgets_values_named": {
761
+ "text": "## Model Links\n\n- [Hugging Face: Comfy-Org/MOSS-Image](https://huggingface.co/Comfy-Org/MOSS-Image)\n- [Hugging Face: Comfy-Org/MoGe](https://huggingface.co/Comfy-Org/MoGe) (geometry estimation)\n- [Upstream control adapter: darask0/MOSS-Image-color-depth-control](https://huggingface.co/darask0/MOSS-Image-color-depth-control)\n\n**diffusion_models**\n\n- [moss_image_sft_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors) - 5.41 GB, the checkpoint this graph loads\n- [moss_image_sft_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_bf16.safetensors) - 10.16 GB, higher precision, slower\n- [moss_image_base_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_base_int8_convrot.safetensors) - 5.41 GB, the base checkpoint runs in the same graph\n\n**text_encoders**\n\n- [qwen3.5_9b_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors) - 9.75 GB, the encoder this graph loads\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) - 7.34 GB, drop-in swap, saves 2.4 GB\n- [qwen3.5_9b_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_bf16.safetensors) - 19.31 GB, full precision\n\n**vae**\n\n- [flux2-vae_small.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors) - 249 MB, the one this graph loads\n- [flux2-vae.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae.safetensors) - 336 MB, full size, pixel-level differences only\n\n**loras**\n\n- [moss_image_color_depth_control.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors) - 313 MB, `LoraLoaderModelOnly` at strength 1.0\n\n**geometry_estimation**\n\n- [moge_3_vitg_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitg_fp16.safetensors) - 2.50 GB, the depth estimator used by the subgraph\n- [moge_3_vitl_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitl_fp16.safetensors) - 741 MB, the lighter ViT-L variant\n\n**Disk footprint**: the graph as shipped needs **18.23 GB** (18,226,010,396 bytes) (5.41 + 9.75 + 0.25 + 0.31 + 2.50). Swapping the text encoder to `w4a8_gptq` brings it to **15.82 GB**; a bf16 checkpoint and encoder instead push it past 32 GB.\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 diffusion_models/\n│ │ └── moss_image_sft_int8_convrot.safetensors 5.41 GB\n│ ├── 📂 text_encoders/\n│ │ └── qwen3.5_9b_int8_convrot.safetensors 9.75 GB\n│ ├── 📂 vae/\n│ │ └── flux2-vae_small.safetensors 249 MB\n│ ├── 📂 loras/\n│ │ └── moss_image_color_depth_control.safetensors 313 MB\n│ └── 📂 geometry_estimation/\n│ └── moge_3_vitg_fp16.safetensors 2.50 GB\n```\n\nCore support (`MOSSImageControl`, MoGe nodes) ships from ComfyUI **0.43.0**.\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
762
+ },
763
+ "color": "#222",
764
+ "bgcolor": "#000"
765
+ },
766
+ {
767
+ "id": 85,
768
+ "type": "SaveImage",
769
+ "pos": [
770
+ 3307.317013871774,
771
+ 597.8636595391883
772
+ ],
773
+ "size": [
774
+ 420,
775
+ 470
776
+ ],
777
+ "flags": {},
778
+ "order": 29,
779
+ "mode": 0,
780
+ "inputs": [
781
+ {
782
+ "name": "images",
783
+ "type": "IMAGE",
784
+ "link": 228
785
+ }
786
+ ],
787
+ "outputs": [
788
+ {
789
+ "name": "images",
790
+ "type": "IMAGE",
791
+ "links": null
792
+ }
793
+ ],
794
+ "properties": {
795
+ "cnr_id": "comfy-core",
796
+ "ver": "0.44.0",
797
+ "Node name for S&R": "SaveImage"
798
+ },
799
+ "widgets_values": [
800
+ "image_moss_image_color_depth_control"
801
+ ],
802
+ "widgets_values_named": {
803
+ "filename_prefix": "image_moss_image_color_depth_control"
804
+ }
805
+ },
806
+ {
807
+ "id": 86,
808
+ "type": "KSamplerSelect",
809
+ "pos": [
810
+ 1170.4423572560272,
811
+ -260.582916386391
812
+ ],
813
+ "size": [
814
+ 270,
815
+ 58
816
+ ],
817
+ "flags": {},
818
+ "order": 6,
819
+ "mode": 0,
820
+ "inputs": [],
821
+ "outputs": [
822
+ {
823
+ "name": "SAMPLER",
824
+ "type": "SAMPLER",
825
+ "links": [
826
+ 154
827
+ ]
828
+ }
829
+ ],
830
+ "properties": {
831
+ "cnr_id": "comfy-core",
832
+ "ver": "0.44.0",
833
+ "Node name for S&R": "KSamplerSelect"
834
+ },
835
+ "widgets_values": [
836
+ "euler"
837
+ ],
838
+ "widgets_values_named": {
839
+ "sampler_name": "euler"
840
+ }
841
+ },
842
+ {
843
+ "id": 87,
844
+ "type": "ImageScaleBy",
845
+ "pos": [
846
+ -7.178545924871845,
847
+ 681.7385897176745
848
+ ],
849
+ "size": [
850
+ 270,
851
+ 82
852
+ ],
853
+ "flags": {},
854
+ "order": 13,
855
+ "mode": 4,
856
+ "inputs": [
857
+ {
858
+ "name": "image",
859
+ "type": "IMAGE",
860
+ "link": 155
861
+ }
862
+ ],
863
+ "outputs": [
864
+ {
865
+ "name": "IMAGE",
866
+ "type": "IMAGE",
867
+ "links": [
868
+ 161
869
+ ]
870
+ }
871
+ ],
872
+ "properties": {
873
+ "cnr_id": "comfy-core",
874
+ "ver": "0.44.0",
875
+ "Node name for S&R": "ImageScaleBy"
876
+ },
877
+ "widgets_values": [
878
+ "lanczos",
879
+ 0.5
880
+ ],
881
+ "widgets_values_named": {
882
+ "upscale_method": "lanczos",
883
+ "scale_by": 0.5
884
+ }
885
+ },
886
+ {
887
+ "id": 88,
888
+ "type": "LoraLoaderModelOnly",
889
+ "pos": [
890
+ -447.13559989481814,
891
+ -263.5408892596166
892
+ ],
893
+ "size": [
894
+ 500,
895
+ 120
896
+ ],
897
+ "flags": {},
898
+ "order": 15,
899
+ "mode": 0,
900
+ "inputs": [
901
+ {
902
+ "name": "model",
903
+ "type": "MODEL",
904
+ "link": 157
905
+ }
906
+ ],
907
+ "outputs": [
908
+ {
909
+ "name": "MODEL",
910
+ "type": "MODEL",
911
+ "links": [
912
+ 260
913
+ ]
914
+ }
915
+ ],
916
+ "properties": {
917
+ "cnr_id": "comfy-core",
918
+ "ver": "0.44.0",
919
+ "Node name for S&R": "LoraLoaderModelOnly",
920
+ "models": [
921
+ {
922
+ "name": "moss_image_color_depth_control.safetensors",
923
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
924
+ "directory": "loras"
925
+ }
926
+ ]
927
+ },
928
+ "widgets_values": [
929
+ "moss/download/Coffmeteor-sft-15step-cfg3.safetensors",
930
+ 1
931
+ ],
932
+ "widgets_values_named": {
933
+ "lora_name": "moss/download/Coffmeteor-sft-15step-cfg3.safetensors",
934
+ "strength_model": 1
935
+ }
936
+ },
937
+ {
938
+ "id": 90,
939
+ "type": "ImageScaleBy",
940
+ "pos": [
941
+ 321.18555329294736,
942
+ 683.4828563588695
943
+ ],
944
+ "size": [
945
+ 270,
946
+ 82
947
+ ],
948
+ "flags": {},
949
+ "order": 16,
950
+ "mode": 0,
951
+ "inputs": [
952
+ {
953
+ "name": "image",
954
+ "type": "IMAGE",
955
+ "link": 161
956
+ }
957
+ ],
958
+ "outputs": [
959
+ {
960
+ "name": "IMAGE",
961
+ "type": "IMAGE",
962
+ "links": [
963
+ 264,
964
+ 265
965
+ ]
966
+ }
967
+ ],
968
+ "properties": {
969
+ "cnr_id": "comfy-core",
970
+ "ver": "0.44.0",
971
+ "Node name for S&R": "ImageScaleBy"
972
+ },
973
+ "widgets_values": [
974
+ "lanczos",
975
+ 4
976
+ ],
977
+ "widgets_values_named": {
978
+ "upscale_method": "lanczos",
979
+ "scale_by": 4
980
+ }
981
+ },
982
+ {
983
+ "id": 91,
984
+ "type": "PrimitiveStringMultiline",
985
+ "pos": [
986
+ 375.1526388094519,
987
+ -2150.2072456067854
988
+ ],
989
+ "size": [
990
+ 470,
991
+ 270
992
+ ],
993
+ "flags": {
994
+ "collapsed": false
995
+ },
996
+ "order": 7,
997
+ "mode": 4,
998
+ "inputs": [],
999
+ "outputs": [
1000
+ {
1001
+ "name": "STRING",
1002
+ "type": "STRING",
1003
+ "links": [
1004
+ 163
1005
+ ]
1006
+ }
1007
+ ],
1008
+ "title": "Text (System prompt)",
1009
+ "properties": {
1010
+ "cnr_id": "comfy-core",
1011
+ "ver": "0.42.0",
1012
+ "Node name for S&R": "PrimitiveStringMultiline"
1013
+ },
1014
+ "widgets_values": [
1015
+ "\n你是一名顶级视觉逆向分析专家、数字绘画研究员和 AI 图像生成提示词工程师。\n\n你的唯一任务是:对用户提供的参考图像进行极其细致的视觉分析,并将所有具有复现价值的视觉特征整合为一段高精度、可直接用于 AI 图像生成模型的英文提示词。\n\n【核心要求】\n\n你必须在内部完成充分、深入的图像分析,但绝对不能向用户展示分析过程、分析结果、推理步骤、分类标题或任何解释。\n\n最终回答只能包含英文 Positive Prompt 本身。\n\n禁止输出:\n- 任何中文内容\n- 开场白或结束语\n- 标题、编号、分类标签\n- Markdown 格式或代码块\n- 分析过程和解释\n- Negative Prompt\n- 参数建议\n- 额外说明\n- 与提示词无关的任何文字\n\n【分析深度】\n\n在内部对图像进行像素级视觉特征观察,尽可能细致地识别:\n\n1. 整体艺术风格、媒介特征、风格融合方式、二维或三维表现倾向。\n\n2. 线稿系统:外轮廓粗细、内部线条粗细、线宽变化、线条收尖、线条转折、曲率、闭合程度、线条颜色、线条疏密、边缘平滑度。\n\n3. 形状语言:几何概括方式、曲线与直线比例、轮廓简化程度、局部夸张、形状节奏、视觉体积的建立方式。\n\n4. 上色方法:平涂、赛璐璐、厚涂、渐变、色块分层、色彩边界、硬边与软边过渡、局部混合渲染。\n\n5. 阴影结构:阴影边界锐利度、阴影形状、阴影色相、明暗层级、结构性阴影、接触阴影、装饰性阴影、阴影与形体的关系。\n\n6. 高光系统:高光形状、大小、分布、方向、边缘硬度、色彩、亮度、反射效果、块状高光、条状高光、局部辉光。\n\n7. 配色逻辑:主色调、辅助色、强调色、色彩饱和度、明度分布、冷暖对比、亮暗部色相偏移、色彩统一性。\n\n8. 光照处理:主光方向、环境光、轮廓光、光照对比度、局部泛光、光影风格化程度。\n\n9. 渲染质感:数字绘画感、笔触可见度、纹理密度、边缘锐度、画面平滑度、颗粒感、印刷感、类似矢量图的视觉质感。\n\n10. 细节组织:视觉焦点、细节密度、局部简化、主次关系、大中小形状的组合、画面精致程度。\n\n11. 构图与画面语言:景别、视角、透视、裁切、空间层次、主体位置、视觉动线。\n\n12. 画面内容:主体外观、服装、表情、姿态、道具、背景和场景元素。\n\n【反推规则】\n\n1. 默认采用纯画风反推模式:只提取能够迁移到其他主体的绘画风格,不描述具体人物、体型、性别、肌肉、服装、动作或场景。\n\n2. 如果用户明确要求完整反推,则同时还原主体、场景、构图和绘画风格。\n\n3. 如果提供多张参考图,提取所有图像共同、稳定且具有辨识度的视觉特征,不把单张图片独有的元素误认为共同画风。\n\n4. 不允许只用 anime style、beautiful illustration、detailed shading、high quality 等笼统词汇代替具体的视觉描述。\n\n5. 优先使用能够控制画面视觉效果的具体英文短语,准确描述线条、阴影、高光、色块、色彩、边缘和渲染方式。\n\n6. 区分硬边阴影与柔和渐变、结构性高光与装饰性高光、真实材质质感与绘画渲染质感。\n\n7. 不得凭空猜测原作者、绘画软件、具体模型、LoRA 或原始提示词。\n\n8. 仅描述有视觉证据支持的特征,不虚构不可见的细节。\n\n9. 将最有辨识度、最影响生成结果的特征放在提示词前部。\n\n10. 避免重复堆砌近义词,避免相互矛盾的描述,避免无意义的质量标签。\n\n11. 允许详细、丰富、具体的提示词,但每个短语都必须具有实际的视觉控制价值。\n\n12. 将所有分析结果重新组织为流畅、准确、信息密度高的英文生图提示词,而不是逐条罗列分析项目。\n\n13. 优先使用适合 MOSS-Image 等自然语言生图模型的英文描述。若用户明确指定 Anima 或 Danbooru 标签格式,则转换为适合该模型的英文标签式提示词。\n\n14. 对无法确认的视觉细节不进行猜测,不为了增加长度而编造特征。\n\n【最终输出规范——最高优先级】\n\n最终输出必须严格遵守以下要求:\n\n只输出一段英文 Positive Prompt。\n\n不允许出现任何标题。\n不允许出现任何分析内容。\n不允许解释为什么选择这些词。\n不允许输出 Negative Prompt。\n不允许附加任何建议���\n不允许添加引号或代码块。\n不允许输出第二个版本。\n\n所有视觉分析仅在内部完成,最终仅保留经过筛选、整合和优化的高精度提示词。\n\n无论分析多么复杂,输出形式都不能改变。\n\n用户上传图像后,直接输出最终英文提示词,不要进行任何说明。\n"
1016
+ ],
1017
+ "widgets_values_named": {
1018
+ "value": "\n你是一名顶级视觉逆向分析专家、数字绘画研究员和 AI 图像生成提示词工程师。\n\n你的唯一任务是:对用户提供的参考图像进行极其细致的视觉分析,并将所有具有复现价值的视觉特征整合为一段高精度、可直接用于 AI 图像生成模型的英文提示词。\n\n【核心要求】\n\n你必须在内部完成充分、深入的图像分析,但绝对不能向用户展示分析过程、分析结果、推理步骤、分类标题或任何解释。\n\n最终回答只能包含英文 Positive Prompt 本身。\n\n禁止输出:\n- 任何中文内容\n- 开场白或结束语\n- 标题、编号、分类标签\n- Markdown 格式或代码块\n- 分析过程和解释\n- Negative Prompt\n- 参数建议\n- 额外说明\n- 与提示词无关的任何文字\n\n【分析深度】\n\n在内部对图像进行像素级视觉特征观察,尽可能细致地识别:\n\n1. 整体艺术风格、媒介特征、风格融合方式、二维或三维表现倾向。\n\n2. 线稿系统:外轮廓粗细、内部线条粗细、线宽变化、线条收尖、线条转折、曲率、闭合程度、线条颜色、线条疏密、边缘平滑度。\n\n3. 形状语言:几何概括方式、曲线与直线比例、轮廓简化程度、局部夸张、形状节奏、视觉体积的建立方式。\n\n4. 上色方法:平涂、赛璐璐、厚涂、渐变、色块分层、色彩边界、硬边与软边过渡、局部混合渲染。\n\n5. 阴影结构:阴影边界锐利度、阴影形状、阴影色相、明暗层级、结构性阴影、接触阴影、装饰性阴影、阴影与形体的关系。\n\n6. 高光系统:高光形状、大小、分布、方向、边缘硬度、色彩、亮度、反射效果、块状高光、条状高光、局部辉光。\n\n7. 配色逻辑:主色调、辅助色、强调色、色彩饱和度、明度分布、冷暖对比、亮暗部色相偏移、色彩统一性。\n\n8. 光照处理:主光方向、环境光、轮廓光、光照对比度、局部泛光、光影风格化程度。\n\n9. 渲染质感:数字绘画感、笔触可见度、纹理密度、边缘锐度、画面平滑度、颗粒感、印刷感、类似矢量图的视觉质感。\n\n10. 细节组织:视觉焦点、细节密度、局部简化、主次关系、大中小形状的组合、画面精致程度。\n\n11. 构图与画面语言:景别、视角、透视、裁切、空间层次、主体位置、视觉动线。\n\n12. 画面内容:主体外观、服装、表情、姿态、道具、背景和场景元素。\n\n【反推规则】\n\n1. 默认采用纯画风反推模式:只提取能够迁移到其他主体的绘画风格,不描述具体人物、体型、性别、肌肉、服装、动作或场景。\n\n2. 如果用户明确要求完整反推,则同时还原主体、场景、构图和绘画风格。\n\n3. 如果提供多张参考图,提取所有图像共同、稳定且具有辨识度的视觉特征,不把单张图片独有的元素误认为共同画风。\n\n4. 不允许只用 anime style、beautiful illustration、detailed shading、high quality 等笼统词汇代替具体的视觉描述。\n\n5. 优先使用能够控制画面视觉效果的具体英文短语,准确描述线条、阴影、高光、色块、色彩、边缘和渲染方式。\n\n6. 区分硬边阴影与柔和渐变、结构性高光与装饰性高光、真实材质质感与绘画渲染质感。\n\n7. 不得凭空猜测原作者、绘画软件、具体模型、LoRA 或原始提示词。\n\n8. 仅描述有视觉证据支持的特征,不虚构不可见的细节。\n\n9. 将最有辨识度、最影响生成结果的特征放在提示词前部。\n\n10. 避免重复堆砌近义词,避免相互矛盾的描述,避免无意义的质量标签。\n\n11. 允许详细、丰富、具体的提示词,但每个短语都必须具有实际的视觉控制价值。\n\n12. 将所有分析结果重新组织为流畅、准确、信息密度高的英文生图提示词,而不是逐条罗列分析项目。\n\n13. 优先使用适合 MOSS-Image 等自然语言生图模型的英文描述。若用户明确指定 Anima 或 Danbooru 标签格式,则转换为适合该模型的英文标签式提示词。\n\n14. 对无法确认的视觉细节不进行猜测,不为了增加长度而编造特征。\n\n【最终输出规范——最高优先级】\n\n最终输出必须严格遵守以下要求:\n\n只输出一段英文 Positive Prompt。\n\n不允许出现任何标题。\n不允许出现任何分析内容。\n不允许解释为什么选择这些词。\n不允许输出 Negative Prompt。\n不允许附加任何建议。\n不允许添加引号或代码块。\n不允许输出第二个版本。\n\n所有视觉分析仅在内部完成,最终仅保留经过筛选、整合和优化的高精度提示词。\n\n无论分析多么复杂,输出形式都不能改变。\n\n用户上传图像后,直接输出最终英文提示词,不要进行任何说明。\n"
1019
+ }
1020
+ },
1021
+ {
1022
+ "id": 92,
1023
+ "type": "TextGenerate",
1024
+ "pos": [
1025
+ 1171.0822008806106,
1026
+ -2365.5035628809155
1027
+ ],
1028
+ "size": [
1029
+ 400,
1030
+ 550
1031
+ ],
1032
+ "flags": {},
1033
+ "order": 14,
1034
+ "mode": 4,
1035
+ "inputs": [
1036
+ {
1037
+ "name": "clip",
1038
+ "type": "CLIP",
1039
+ "link": 165
1040
+ },
1041
+ {
1042
+ "name": "image",
1043
+ "shape": 7,
1044
+ "type": "IMAGE",
1045
+ "link": 248
1046
+ },
1047
+ {
1048
+ "name": "video",
1049
+ "shape": 7,
1050
+ "type": "IMAGE",
1051
+ "link": null
1052
+ },
1053
+ {
1054
+ "name": "audio",
1055
+ "shape": 7,
1056
+ "type": "AUDIO",
1057
+ "link": null
1058
+ },
1059
+ {
1060
+ "name": "system_prompt",
1061
+ "shape": 7,
1062
+ "type": "STRING",
1063
+ "link": 163
1064
+ }
1065
+ ],
1066
+ "outputs": [
1067
+ {
1068
+ "name": "generated_text",
1069
+ "type": "STRING",
1070
+ "links": [
1071
+ 166
1072
+ ]
1073
+ },
1074
+ {
1075
+ "name": "thinking",
1076
+ "type": "STRING",
1077
+ "links": []
1078
+ }
1079
+ ],
1080
+ "properties": {
1081
+ "cnr_id": "comfy-core",
1082
+ "ver": "0.42.0",
1083
+ "Node name for S&R": "TextGenerate"
1084
+ },
1085
+ "widgets_values": [
1086
+ "",
1087
+ 4096,
1088
+ "on",
1089
+ 0.7,
1090
+ 64,
1091
+ 0.95,
1092
+ 0.05,
1093
+ 1,
1094
+ 0,
1095
+ 0,
1096
+ true,
1097
+ true,
1098
+ "auto"
1099
+ ],
1100
+ "widgets_values_named": {
1101
+ "prompt": "",
1102
+ "max_length": 4096,
1103
+ "sampling_mode": "on",
1104
+ "sampling_mode.temperature": 0.7,
1105
+ "sampling_mode.top_k": 64,
1106
+ "sampling_mode.top_p": 0.95,
1107
+ "sampling_mode.min_p": 0.05,
1108
+ "sampling_mode.repetition_penalty": 1,
1109
+ "sampling_mode.seed": 0,
1110
+ "sampling_mode.presence_penalty": 0,
1111
+ "thinking": true,
1112
+ "use_default_template": true,
1113
+ "mtp": "auto"
1114
+ }
1115
+ },
1116
+ {
1117
+ "id": 93,
1118
+ "type": "CLIPLoader",
1119
+ "pos": [
1120
+ 468.0082400074318,
1121
+ -2370.6899919511707
1122
+ ],
1123
+ "size": [
1124
+ 270,
1125
+ 106
1126
+ ],
1127
+ "flags": {},
1128
+ "order": 8,
1129
+ "mode": 4,
1130
+ "inputs": [],
1131
+ "outputs": [
1132
+ {
1133
+ "name": "CLIP",
1134
+ "type": "CLIP",
1135
+ "links": [
1136
+ 165
1137
+ ]
1138
+ }
1139
+ ],
1140
+ "properties": {
1141
+ "cnr_id": "comfy-core",
1142
+ "ver": "0.44.0",
1143
+ "Node name for S&R": "CLIPLoader"
1144
+ },
1145
+ "widgets_values": [
1146
+ "qwen3.5-9b-nsfw-captioning-v5.safetensors",
1147
+ "moss_image",
1148
+ "default"
1149
+ ],
1150
+ "widgets_values_named": {
1151
+ "clip_name": "qwen3.5-9b-nsfw-captioning-v5.safetensors",
1152
+ "type": "moss_image",
1153
+ "device": "default"
1154
+ }
1155
+ },
1156
+ {
1157
+ "id": 94,
1158
+ "type": "PreviewAny",
1159
+ "pos": [
1160
+ 1713.012250094057,
1161
+ -2364.526445608352
1162
+ ],
1163
+ "size": [
1164
+ 470.96127217369775,
1165
+ 425.59604314573403
1166
+ ],
1167
+ "flags": {},
1168
+ "order": 17,
1169
+ "mode": 4,
1170
+ "inputs": [
1171
+ {
1172
+ "name": "source",
1173
+ "type": "*",
1174
+ "link": 166
1175
+ }
1176
+ ],
1177
+ "outputs": [
1178
+ {
1179
+ "name": "STRING",
1180
+ "type": "STRING",
1181
+ "links": []
1182
+ }
1183
+ ],
1184
+ "properties": {
1185
+ "cnr_id": "comfy-core",
1186
+ "ver": "0.44.0",
1187
+ "Node name for S&R": "PreviewAny"
1188
+ },
1189
+ "widgets_values": [],
1190
+ "widgets_values_named": {}
1191
+ },
1192
+ {
1193
+ "id": 124,
1194
+ "type": "VAEEncode",
1195
+ "pos": [
1196
+ 1149.2279511667382,
1197
+ 242.2625195138657
1198
+ ],
1199
+ "size": [
1200
+ 140,
1201
+ 46
1202
+ ],
1203
+ "flags": {},
1204
+ "order": 20,
1205
+ "mode": 0,
1206
+ "inputs": [
1207
+ {
1208
+ "name": "pixels",
1209
+ "type": "IMAGE",
1210
+ "link": 265
1211
+ },
1212
+ {
1213
+ "name": "vae",
1214
+ "type": "VAE",
1215
+ "link": 246
1216
+ }
1217
+ ],
1218
+ "outputs": [
1219
+ {
1220
+ "name": "LATENT",
1221
+ "type": "LATENT",
1222
+ "links": [
1223
+ 247
1224
+ ]
1225
+ }
1226
+ ],
1227
+ "properties": {
1228
+ "cnr_id": "comfy-core",
1229
+ "ver": "0.44.0",
1230
+ "Node name for S&R": "VAEEncode"
1231
+ }
1232
+ },
1233
+ {
1234
+ "id": 125,
1235
+ "type": "MossTileControlLoader",
1236
+ "pos": [
1237
+ 758.3282271676215,
1238
+ -61.473909743927265
1239
+ ],
1240
+ "size": [
1241
+ 270,
1242
+ 58
1243
+ ],
1244
+ "flags": {},
1245
+ "order": 9,
1246
+ "mode": 0,
1247
+ "inputs": [],
1248
+ "outputs": [
1249
+ {
1250
+ "name": "MOSS_TILE_CONTROL",
1251
+ "type": "MOSS_TILE_CONTROL",
1252
+ "links": [
1253
+ 249
1254
+ ]
1255
+ }
1256
+ ],
1257
+ "properties": {
1258
+ "Node name for S&R": "MossTileControlLoader"
1259
+ },
1260
+ "widgets_values": [
1261
+ "moss-tile-control/beta3/epoch2/adapter_model.safetensors"
1262
+ ],
1263
+ "widgets_values_named": {
1264
+ "name": "moss-tile-control/beta3/epoch2/adapter_model.safetensors"
1265
+ }
1266
+ },
1267
+ {
1268
+ "id": 126,
1269
+ "type": "MossTileControlApply",
1270
+ "pos": [
1271
+ 759.0227970689675,
1272
+ 111.48495135711164
1273
+ ],
1274
+ "size": [
1275
+ 270,
1276
+ 166
1277
+ ],
1278
+ "flags": {},
1279
+ "order": 22,
1280
+ "mode": 0,
1281
+ "inputs": [
1282
+ {
1283
+ "name": "model",
1284
+ "type": "MODEL",
1285
+ "link": 255
1286
+ },
1287
+ {
1288
+ "name": "control",
1289
+ "type": "MOSS_TILE_CONTROL",
1290
+ "link": 249
1291
+ },
1292
+ {
1293
+ "name": "vae",
1294
+ "type": "VAE",
1295
+ "link": 256
1296
+ },
1297
+ {
1298
+ "name": "image",
1299
+ "type": "IMAGE",
1300
+ "link": 257
1301
+ }
1302
+ ],
1303
+ "outputs": [
1304
+ {
1305
+ "name": "MODEL",
1306
+ "type": "MODEL",
1307
+ "links": [
1308
+ 253,
1309
+ 254
1310
+ ]
1311
+ }
1312
+ ],
1313
+ "properties": {
1314
+ "Node name for S&R": "MossTileControlApply"
1315
+ },
1316
+ "widgets_values": [
1317
+ 0.5,
1318
+ 0,
1319
+ 1
1320
+ ],
1321
+ "widgets_values_named": {
1322
+ "strength": 0.5,
1323
+ "start_percent": 0,
1324
+ "end_percent": 1
1325
+ }
1326
+ },
1327
+ {
1328
+ "id": 129,
1329
+ "type": "ColorMatchV2",
1330
+ "pos": [
1331
+ 2216.9443861363466,
1332
+ -184.51602335792467
1333
+ ],
1334
+ "size": [
1335
+ 270,
1336
+ 126
1337
+ ],
1338
+ "flags": {},
1339
+ "order": 26,
1340
+ "mode": 0,
1341
+ "inputs": [
1342
+ {
1343
+ "name": "image_target",
1344
+ "type": "IMAGE",
1345
+ "link": 266
1346
+ },
1347
+ {
1348
+ "name": "image_ref",
1349
+ "type": "IMAGE",
1350
+ "link": 267
1351
+ }
1352
+ ],
1353
+ "outputs": [
1354
+ {
1355
+ "name": "image",
1356
+ "type": "IMAGE",
1357
+ "links": [
1358
+ 268
1359
+ ]
1360
+ }
1361
+ ],
1362
+ "properties": {
1363
+ "cnr_id": "comfyui-kjnodes",
1364
+ "ver": "37ca7aff7ed14dde6ba3c0b25713e21c18b28dcb",
1365
+ "Node name for S&R": "ColorMatchV2"
1366
+ },
1367
+ "widgets_values": [
1368
+ "mkl",
1369
+ 1,
1370
+ true
1371
+ ],
1372
+ "widgets_values_named": {
1373
+ "method": "mkl",
1374
+ "strength": 1,
1375
+ "multithread": true
1376
+ }
1377
+ }
1378
+ ],
1379
+ "links": [
1380
+ [
1381
+ 3,
1382
+ 3,
1383
+ 0,
1384
+ 5,
1385
+ 0,
1386
+ "CLIP"
1387
+ ],
1388
+ [
1389
+ 4,
1390
+ 3,
1391
+ 0,
1392
+ 6,
1393
+ 0,
1394
+ "CLIP"
1395
+ ],
1396
+ [
1397
+ 9,
1398
+ 4,
1399
+ 0,
1400
+ 9,
1401
+ 1,
1402
+ "VAE"
1403
+ ],
1404
+ [
1405
+ 25,
1406
+ 20,
1407
+ 0,
1408
+ 9,
1409
+ 0,
1410
+ "LATENT"
1411
+ ],
1412
+ [
1413
+ 27,
1414
+ 22,
1415
+ 0,
1416
+ 20,
1417
+ 4,
1418
+ "SIGMAS"
1419
+ ],
1420
+ [
1421
+ 69,
1422
+ 36,
1423
+ 0,
1424
+ 2,
1425
+ 0,
1426
+ "MODEL"
1427
+ ],
1428
+ [
1429
+ 91,
1430
+ 1,
1431
+ 0,
1432
+ 49,
1433
+ 0,
1434
+ "MODEL"
1435
+ ],
1436
+ [
1437
+ 149,
1438
+ 42,
1439
+ 0,
1440
+ 80,
1441
+ 1,
1442
+ "IMAGE"
1443
+ ],
1444
+ [
1445
+ 150,
1446
+ 79,
1447
+ 0,
1448
+ 80,
1449
+ 0,
1450
+ "IMAGE"
1451
+ ],
1452
+ [
1453
+ 154,
1454
+ 86,
1455
+ 0,
1456
+ 20,
1457
+ 3,
1458
+ "SAMPLER"
1459
+ ],
1460
+ [
1461
+ 155,
1462
+ 55,
1463
+ 0,
1464
+ 87,
1465
+ 0,
1466
+ "IMAGE"
1467
+ ],
1468
+ [
1469
+ 157,
1470
+ 49,
1471
+ 0,
1472
+ 88,
1473
+ 0,
1474
+ "MODEL"
1475
+ ],
1476
+ [
1477
+ 161,
1478
+ 87,
1479
+ 0,
1480
+ 90,
1481
+ 0,
1482
+ "IMAGE"
1483
+ ],
1484
+ [
1485
+ 166,
1486
+ 92,
1487
+ 0,
1488
+ 94,
1489
+ 0,
1490
+ "STRING"
1491
+ ],
1492
+ [
1493
+ 228,
1494
+ 42,
1495
+ 0,
1496
+ 85,
1497
+ 0,
1498
+ "IMAGE"
1499
+ ],
1500
+ [
1501
+ 246,
1502
+ 4,
1503
+ 0,
1504
+ 124,
1505
+ 1,
1506
+ "VAE"
1507
+ ],
1508
+ [
1509
+ 247,
1510
+ 124,
1511
+ 0,
1512
+ 20,
1513
+ 5,
1514
+ "LATENT"
1515
+ ],
1516
+ [
1517
+ 249,
1518
+ 125,
1519
+ 0,
1520
+ 126,
1521
+ 1,
1522
+ "MOSS_TILE_CONTROL"
1523
+ ],
1524
+ [
1525
+ 250,
1526
+ 5,
1527
+ 0,
1528
+ 20,
1529
+ 1,
1530
+ "CONDITIONING"
1531
+ ],
1532
+ [
1533
+ 251,
1534
+ 6,
1535
+ 0,
1536
+ 20,
1537
+ 2,
1538
+ "CONDITIONING"
1539
+ ],
1540
+ [
1541
+ 253,
1542
+ 126,
1543
+ 0,
1544
+ 20,
1545
+ 0,
1546
+ "MODEL"
1547
+ ],
1548
+ [
1549
+ 254,
1550
+ 126,
1551
+ 0,
1552
+ 22,
1553
+ 0,
1554
+ "MODEL"
1555
+ ],
1556
+ [
1557
+ 255,
1558
+ 2,
1559
+ 0,
1560
+ 126,
1561
+ 0,
1562
+ "MODEL"
1563
+ ],
1564
+ [
1565
+ 256,
1566
+ 4,
1567
+ 0,
1568
+ 126,
1569
+ 2,
1570
+ "VAE"
1571
+ ],
1572
+ [
1573
+ 257,
1574
+ 79,
1575
+ 0,
1576
+ 126,
1577
+ 3,
1578
+ "IMAGE"
1579
+ ],
1580
+ [
1581
+ 260,
1582
+ 88,
1583
+ 0,
1584
+ 36,
1585
+ 0,
1586
+ "MODEL"
1587
+ ],
1588
+ [
1589
+ 264,
1590
+ 90,
1591
+ 0,
1592
+ 79,
1593
+ 0,
1594
+ "IMAGE"
1595
+ ],
1596
+ [
1597
+ 265,
1598
+ 90,
1599
+ 0,
1600
+ 124,
1601
+ 0,
1602
+ "IMAGE"
1603
+ ],
1604
+ [
1605
+ 165,
1606
+ 93,
1607
+ 0,
1608
+ 92,
1609
+ 0,
1610
+ "CLIP"
1611
+ ],
1612
+ [
1613
+ 248,
1614
+ 55,
1615
+ 0,
1616
+ 92,
1617
+ 1,
1618
+ "IMAGE"
1619
+ ],
1620
+ [
1621
+ 163,
1622
+ 91,
1623
+ 0,
1624
+ 92,
1625
+ 4,
1626
+ "STRING"
1627
+ ],
1628
+ [
1629
+ 266,
1630
+ 9,
1631
+ 0,
1632
+ 129,
1633
+ 0,
1634
+ "IMAGE"
1635
+ ],
1636
+ [
1637
+ 267,
1638
+ 79,
1639
+ 0,
1640
+ 129,
1641
+ 1,
1642
+ "IMAGE"
1643
+ ],
1644
+ [
1645
+ 268,
1646
+ 129,
1647
+ 0,
1648
+ 42,
1649
+ 0,
1650
+ "IMAGE"
1651
+ ]
1652
+ ],
1653
+ "groups": [
1654
+ {
1655
+ "id": 1,
1656
+ "title": "Models",
1657
+ "bounding": [
1658
+ -460,
1659
+ -570,
1660
+ 550,
1661
+ 950
1662
+ ],
1663
+ "color": "#3f789e",
1664
+ "flags": {}
1665
+ },
1666
+ {
1667
+ "id": 2,
1668
+ "title": "Conditioning",
1669
+ "bounding": [
1670
+ 120,
1671
+ -140,
1672
+ 970,
1673
+ 520
1674
+ ],
1675
+ "color": "#A88",
1676
+ "flags": {}
1677
+ },
1678
+ {
1679
+ "id": 3,
1680
+ "title": "Group",
1681
+ "bounding": [
1682
+ 120,
1683
+ -570,
1684
+ 970,
1685
+ 400
1686
+ ],
1687
+ "color": "#3f789e",
1688
+ "flags": {}
1689
+ },
1690
+ {
1691
+ "id": 4,
1692
+ "title": "Sampling",
1693
+ "bounding": [
1694
+ 1106.0160595404368,
1695
+ -500.08001764249127,
1696
+ 740,
1697
+ 950
1698
+ ],
1699
+ "color": "#3f789e",
1700
+ "flags": {}
1701
+ },
1702
+ {
1703
+ "id": 5,
1704
+ "title": "Control Images",
1705
+ "bounding": [
1706
+ -460,
1707
+ 410,
1708
+ 1550,
1709
+ 640
1710
+ ],
1711
+ "color": "#A88",
1712
+ "flags": {}
1713
+ }
1714
+ ],
1715
+ "config": {},
1716
+ "extra": {
1717
+ "ds": {
1718
+ "scale": 0.5730855330117224,
1719
+ "offset": [
1720
+ 385.95525826847523,
1721
+ 1191.4153089462757
1722
+ ]
1723
+ },
1724
+ "frontendVersion": "1.58.5",
1725
+ "VHS_latentpreview": false,
1726
+ "VHS_latentpreviewrate": 0,
1727
+ "VHS_MetadataImage": true,
1728
+ "VHS_KeepIntermediate": true
1729
+ },
1730
+ "extensions": {
1731
+ "seed_widgets": {
1732
+ "20": 1
1733
+ }
1734
+ },
1735
+ "version": 0.4,
1736
+ "seed_widgets": {
1737
+ "20": 1
1738
+ }
1739
+ }
workflow/moss_image_tile-beta3.json ADDED
@@ -0,0 +1,1666 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "id": "5140b129-e927-49a6-816e-668c69d43008",
3
+ "revision": 0,
4
+ "last_node_id": 128,
5
+ "last_link_id": 265,
6
+ "nodes": [
7
+ {
8
+ "id": 1,
9
+ "type": "UNETLoader",
10
+ "pos": [
11
+ -423.26391629976365,
12
+ -689.4382881990866
13
+ ],
14
+ "size": [
15
+ 500,
16
+ 90
17
+ ],
18
+ "flags": {},
19
+ "order": 0,
20
+ "mode": 0,
21
+ "inputs": [],
22
+ "outputs": [
23
+ {
24
+ "name": "MODEL",
25
+ "type": "MODEL",
26
+ "slot_index": 0,
27
+ "links": [
28
+ 91
29
+ ]
30
+ }
31
+ ],
32
+ "properties": {
33
+ "cnr_id": "comfy-core",
34
+ "ver": "0.44.0",
35
+ "Node name for S&R": "UNETLoader",
36
+ "models": [
37
+ {
38
+ "name": "moss_image_sft_int8_convrot.safetensors",
39
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors",
40
+ "directory": "diffusion_models"
41
+ }
42
+ ]
43
+ },
44
+ "widgets_values": [
45
+ "moss-image/moss_image_sft_int8_convrot.safetensors",
46
+ "default"
47
+ ],
48
+ "widgets_values_named": {
49
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
50
+ "weight_dtype": "default"
51
+ }
52
+ },
53
+ {
54
+ "id": 2,
55
+ "type": "SkipLayerGuidanceDiTSimple",
56
+ "pos": [
57
+ 620,
58
+ -460
59
+ ],
60
+ "size": [
61
+ 400,
62
+ 130
63
+ ],
64
+ "flags": {},
65
+ "order": 21,
66
+ "mode": 4,
67
+ "showAdvanced": true,
68
+ "inputs": [
69
+ {
70
+ "name": "model",
71
+ "type": "MODEL",
72
+ "link": 69
73
+ }
74
+ ],
75
+ "outputs": [
76
+ {
77
+ "name": "MODEL",
78
+ "type": "MODEL",
79
+ "slot_index": 0,
80
+ "links": [
81
+ 255
82
+ ]
83
+ }
84
+ ],
85
+ "title": "SkipLayerGuidanceDiTSimple (uncond skips block 22)",
86
+ "properties": {
87
+ "cnr_id": "comfy-core",
88
+ "ver": "0.44.0",
89
+ "Node name for S&R": "SkipLayerGuidanceDiTSimple"
90
+ },
91
+ "widgets_values": [
92
+ "",
93
+ "24",
94
+ 0,
95
+ 1
96
+ ],
97
+ "widgets_values_named": {
98
+ "double_layers": "",
99
+ "single_layers": "24",
100
+ "start_percent": 0,
101
+ "end_percent": 1
102
+ }
103
+ },
104
+ {
105
+ "id": 3,
106
+ "type": "CLIPLoader",
107
+ "pos": [
108
+ -440,
109
+ -50
110
+ ],
111
+ "size": [
112
+ 500,
113
+ 110
114
+ ],
115
+ "flags": {},
116
+ "order": 1,
117
+ "mode": 0,
118
+ "inputs": [],
119
+ "outputs": [
120
+ {
121
+ "name": "CLIP",
122
+ "type": "CLIP",
123
+ "slot_index": 0,
124
+ "links": [
125
+ 3,
126
+ 4
127
+ ]
128
+ }
129
+ ],
130
+ "properties": {
131
+ "cnr_id": "comfy-core",
132
+ "ver": "0.44.0",
133
+ "Node name for S&R": "CLIPLoader",
134
+ "models": [
135
+ {
136
+ "name": "qwen3.5_9b_int8_convrot.safetensors",
137
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors",
138
+ "directory": "text_encoders"
139
+ }
140
+ ]
141
+ },
142
+ "widgets_values": [
143
+ "qwen3.5_9b_bf16.safetensors",
144
+ "moss_image",
145
+ "default"
146
+ ],
147
+ "widgets_values_named": {
148
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
149
+ "type": "moss_image",
150
+ "device": "default"
151
+ }
152
+ },
153
+ {
154
+ "id": 4,
155
+ "type": "VAELoader",
156
+ "pos": [
157
+ -440,
158
+ 130
159
+ ],
160
+ "size": [
161
+ 500,
162
+ 60
163
+ ],
164
+ "flags": {},
165
+ "order": 2,
166
+ "mode": 0,
167
+ "inputs": [],
168
+ "outputs": [
169
+ {
170
+ "name": "VAE",
171
+ "type": "VAE",
172
+ "slot_index": 0,
173
+ "links": [
174
+ 9,
175
+ 246,
176
+ 256
177
+ ]
178
+ }
179
+ ],
180
+ "properties": {
181
+ "cnr_id": "comfy-core",
182
+ "ver": "0.44.0",
183
+ "Node name for S&R": "VAELoader",
184
+ "models": [
185
+ {
186
+ "name": "flux2-vae_small.safetensors",
187
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors",
188
+ "directory": "vae"
189
+ }
190
+ ]
191
+ },
192
+ "widgets_values": [
193
+ "flux2-vae.safetensors"
194
+ ],
195
+ "widgets_values_named": {
196
+ "vae_name": "flux2-vae.safetensors"
197
+ }
198
+ },
199
+ {
200
+ "id": 5,
201
+ "type": "CLIPTextEncode",
202
+ "pos": [
203
+ 185.95627952212953,
204
+ -36.734098034606404
205
+ ],
206
+ "size": [
207
+ 470,
208
+ 240
209
+ ],
210
+ "flags": {},
211
+ "order": 11,
212
+ "mode": 0,
213
+ "inputs": [
214
+ {
215
+ "name": "clip",
216
+ "type": "CLIP",
217
+ "link": 3
218
+ }
219
+ ],
220
+ "outputs": [
221
+ {
222
+ "name": "CONDITIONING",
223
+ "type": "CONDITIONING",
224
+ "slot_index": 0,
225
+ "links": [
226
+ 250
227
+ ]
228
+ }
229
+ ],
230
+ "title": "Positive Prompt",
231
+ "properties": {
232
+ "cnr_id": "comfy-core",
233
+ "ver": "0.44.0",
234
+ "Node name for S&R": "CLIPTextEncode"
235
+ },
236
+ "widgets_values": [
237
+ "A highly realistic and detailed image that blends realistic photography with digital painting, featuring a grey tabby cat as the main subject. The cat is in a spread-eagle pose, flying through the air, with its front paws and hind legs stretched out. The cat is strapped into a flying drone. There are four propellers, two attached to the front paws and two to the hind legs, featuring silver hubs, black shafts, and bright orange blades. The drone frame is made of black metal rods, and red straps hold the cat securely in place. The cat's head is tilted upward, with wide eyes, small pupils, and a slightly open mouth, looking shocked or surprised. The fur is highly detailed, rendered with a painterly style that combines realism with digital art. The background is a soft, out-of-focus bokeh of green and white, suggesting a bright outdoor setting. The lighting is bright and natural, casting soft shadows that give the scene a solid, 3D illustration quality."
238
+ ],
239
+ "widgets_values_named": {
240
+ "text": "A highly realistic and detailed image that blends realistic photography with digital painting, featuring a grey tabby cat as the main subject. The cat is in a spread-eagle pose, flying through the air, with its front paws and hind legs stretched out. The cat is strapped into a flying drone. There are four propellers, two attached to the front paws and two to the hind legs, featuring silver hubs, black shafts, and bright orange blades. The drone frame is made of black metal rods, and red straps hold the cat securely in place. The cat's head is tilted upward, with wide eyes, small pupils, and a slightly open mouth, looking shocked or surprised. The fur is highly detailed, rendered with a painterly style that combines realism with digital art. The background is a soft, out-of-focus bokeh of green and white, suggesting a bright outdoor setting. The lighting is bright and natural, casting soft shadows that give the scene a solid, 3D illustration quality."
241
+ },
242
+ "color": "#232",
243
+ "bgcolor": "#353"
244
+ },
245
+ {
246
+ "id": 6,
247
+ "type": "CLIPTextEncode",
248
+ "pos": [
249
+ 150,
250
+ 250
251
+ ],
252
+ "size": [
253
+ 480,
254
+ 110
255
+ ],
256
+ "flags": {},
257
+ "order": 12,
258
+ "mode": 0,
259
+ "inputs": [
260
+ {
261
+ "name": "clip",
262
+ "type": "CLIP",
263
+ "link": 4
264
+ }
265
+ ],
266
+ "outputs": [
267
+ {
268
+ "name": "CONDITIONING",
269
+ "type": "CONDITIONING",
270
+ "slot_index": 0,
271
+ "links": [
272
+ 251
273
+ ]
274
+ }
275
+ ],
276
+ "title": "Negative Prompt",
277
+ "properties": {
278
+ "cnr_id": "comfy-core",
279
+ "ver": "0.44.0",
280
+ "Node name for S&R": "CLIPTextEncode"
281
+ },
282
+ "widgets_values": [
283
+ "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
284
+ ],
285
+ "widgets_values_named": {
286
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
287
+ },
288
+ "color": "#223",
289
+ "bgcolor": "#335"
290
+ },
291
+ {
292
+ "id": 9,
293
+ "type": "VAEDecode",
294
+ "pos": [
295
+ 1910,
296
+ -520
297
+ ],
298
+ "size": [
299
+ 210,
300
+ 50
301
+ ],
302
+ "flags": {},
303
+ "order": 25,
304
+ "mode": 0,
305
+ "inputs": [
306
+ {
307
+ "name": "samples",
308
+ "type": "LATENT",
309
+ "link": 25
310
+ },
311
+ {
312
+ "name": "vae",
313
+ "type": "VAE",
314
+ "link": 9
315
+ }
316
+ ],
317
+ "outputs": [
318
+ {
319
+ "name": "IMAGE",
320
+ "type": "IMAGE",
321
+ "slot_index": 0,
322
+ "links": [
323
+ 243
324
+ ]
325
+ }
326
+ ],
327
+ "properties": {
328
+ "cnr_id": "comfy-core",
329
+ "ver": "0.44.0",
330
+ "Node name for S&R": "VAEDecode"
331
+ },
332
+ "color": "#322",
333
+ "bgcolor": "#533"
334
+ },
335
+ {
336
+ "id": 20,
337
+ "type": "SamplerCustom",
338
+ "pos": [
339
+ 1518.3539918681,
340
+ -304.93798438717295
341
+ ],
342
+ "size": [
343
+ 280,
344
+ 500
345
+ ],
346
+ "flags": {},
347
+ "order": 24,
348
+ "mode": 0,
349
+ "inputs": [
350
+ {
351
+ "name": "model",
352
+ "type": "MODEL",
353
+ "link": 253
354
+ },
355
+ {
356
+ "name": "positive",
357
+ "type": "CONDITIONING",
358
+ "link": 250
359
+ },
360
+ {
361
+ "name": "negative",
362
+ "type": "CONDITIONING",
363
+ "link": 251
364
+ },
365
+ {
366
+ "name": "sampler",
367
+ "type": "SAMPLER",
368
+ "link": 154
369
+ },
370
+ {
371
+ "name": "sigmas",
372
+ "type": "SIGMAS",
373
+ "link": 27
374
+ },
375
+ {
376
+ "name": "latent_image",
377
+ "type": "LATENT",
378
+ "link": 247
379
+ }
380
+ ],
381
+ "outputs": [
382
+ {
383
+ "name": "output",
384
+ "type": "LATENT",
385
+ "links": [
386
+ 25
387
+ ]
388
+ },
389
+ {
390
+ "name": "denoised_output",
391
+ "type": "LATENT",
392
+ "links": null
393
+ }
394
+ ],
395
+ "properties": {
396
+ "cnr_id": "comfy-core",
397
+ "ver": "0.44.0",
398
+ "Node name for S&R": "SamplerCustom"
399
+ },
400
+ "widgets_values": [
401
+ true,
402
+ 994529790995974,
403
+ "fixed",
404
+ 5
405
+ ],
406
+ "widgets_values_named": {
407
+ "add_noise": true,
408
+ "noise_seed": 994529790995974,
409
+ "control_after_generate": "fixed",
410
+ "cfg": 5
411
+ }
412
+ },
413
+ {
414
+ "id": 22,
415
+ "type": "BasicScheduler",
416
+ "pos": [
417
+ 1169.6140490851362,
418
+ 1.7699193319720672
419
+ ],
420
+ "size": [
421
+ 270,
422
+ 110
423
+ ],
424
+ "flags": {},
425
+ "order": 23,
426
+ "mode": 0,
427
+ "inputs": [
428
+ {
429
+ "name": "model",
430
+ "type": "MODEL",
431
+ "link": 254
432
+ }
433
+ ],
434
+ "outputs": [
435
+ {
436
+ "name": "SIGMAS",
437
+ "type": "SIGMAS",
438
+ "links": [
439
+ 27
440
+ ]
441
+ }
442
+ ],
443
+ "properties": {
444
+ "cnr_id": "comfy-core",
445
+ "ver": "0.44.0",
446
+ "Node name for S&R": "BasicScheduler"
447
+ },
448
+ "widgets_values": [
449
+ "sgm_uniform",
450
+ 15,
451
+ 1
452
+ ],
453
+ "widgets_values_named": {
454
+ "scheduler": "sgm_uniform",
455
+ "steps": 15,
456
+ "denoise": 1
457
+ }
458
+ },
459
+ {
460
+ "id": 36,
461
+ "type": "ModelSamplingAuraFlow",
462
+ "pos": [
463
+ 216.72263354772872,
464
+ -300
465
+ ],
466
+ "size": [
467
+ 400,
468
+ 90
469
+ ],
470
+ "flags": {},
471
+ "order": 18,
472
+ "mode": 0,
473
+ "inputs": [
474
+ {
475
+ "name": "model",
476
+ "type": "MODEL",
477
+ "link": 260
478
+ }
479
+ ],
480
+ "outputs": [
481
+ {
482
+ "name": "MODEL",
483
+ "type": "MODEL",
484
+ "links": [
485
+ 69
486
+ ]
487
+ }
488
+ ],
489
+ "properties": {
490
+ "cnr_id": "comfy-core",
491
+ "ver": "0.44.0",
492
+ "Node name for S&R": "ModelSamplingAuraFlow"
493
+ },
494
+ "widgets_values": [
495
+ 3,
496
+ "flow"
497
+ ],
498
+ "widgets_values_named": {
499
+ "shift": 3,
500
+ "sampling": "flow"
501
+ }
502
+ },
503
+ {
504
+ "id": 42,
505
+ "type": "PreviewImage",
506
+ "pos": [
507
+ 1884.503594126516,
508
+ -378.5299358786798
509
+ ],
510
+ "size": [
511
+ 890,
512
+ 780
513
+ ],
514
+ "flags": {},
515
+ "order": 26,
516
+ "mode": 0,
517
+ "inputs": [
518
+ {
519
+ "name": "images",
520
+ "type": "IMAGE",
521
+ "link": 243
522
+ }
523
+ ],
524
+ "outputs": [
525
+ {
526
+ "name": "images",
527
+ "type": "IMAGE",
528
+ "links": [
529
+ 149,
530
+ 228
531
+ ]
532
+ }
533
+ ],
534
+ "properties": {
535
+ "cnr_id": "comfy-core",
536
+ "ver": "0.44.0",
537
+ "Node name for S&R": "PreviewImage"
538
+ }
539
+ },
540
+ {
541
+ "id": 49,
542
+ "type": "LoraLoaderModelOnly",
543
+ "pos": [
544
+ -423.26391629976365,
545
+ -439.4382881990866
546
+ ],
547
+ "size": [
548
+ 500,
549
+ 120
550
+ ],
551
+ "flags": {},
552
+ "order": 10,
553
+ "mode": 4,
554
+ "inputs": [
555
+ {
556
+ "name": "model",
557
+ "type": "MODEL",
558
+ "link": 91
559
+ }
560
+ ],
561
+ "outputs": [
562
+ {
563
+ "name": "MODEL",
564
+ "type": "MODEL",
565
+ "links": [
566
+ 157
567
+ ]
568
+ }
569
+ ],
570
+ "properties": {
571
+ "cnr_id": "comfy-core",
572
+ "ver": "0.44.0",
573
+ "Node name for S&R": "LoraLoaderModelOnly",
574
+ "models": [
575
+ {
576
+ "name": "moss_image_color_depth_control.safetensors",
577
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
578
+ "directory": "loras"
579
+ }
580
+ ]
581
+ },
582
+ "widgets_values": [
583
+ "moss/moss-tile-control/beta2/epoch1/adapter_model.safetensors",
584
+ 1
585
+ ],
586
+ "widgets_values_named": {
587
+ "lora_name": "moss/moss-tile-control/beta2/epoch1/adapter_model.safetensors",
588
+ "strength_model": 1
589
+ }
590
+ },
591
+ {
592
+ "id": 55,
593
+ "type": "LoadImage",
594
+ "pos": [
595
+ -400,
596
+ 630
597
+ ],
598
+ "size": [
599
+ 290,
600
+ 110
601
+ ],
602
+ "flags": {},
603
+ "order": 3,
604
+ "mode": 0,
605
+ "inputs": [],
606
+ "outputs": [
607
+ {
608
+ "name": "IMAGE",
609
+ "type": "IMAGE",
610
+ "links": [
611
+ 155,
612
+ 248
613
+ ]
614
+ },
615
+ {
616
+ "name": "MASK",
617
+ "type": "MASK",
618
+ "links": null
619
+ }
620
+ ],
621
+ "properties": {
622
+ "cnr_id": "comfy-core",
623
+ "ver": "0.44.0",
624
+ "Node name for S&R": "LoadImage"
625
+ },
626
+ "widgets_values": [
627
+ "3794962993127468234.jpg",
628
+ "image"
629
+ ],
630
+ "widgets_values_named": {
631
+ "image": "3794962993127468234.jpg",
632
+ "upload": "image"
633
+ },
634
+ "color": "#322",
635
+ "bgcolor": "#533"
636
+ },
637
+ {
638
+ "id": 79,
639
+ "type": "PreviewImage",
640
+ "pos": [
641
+ 733.854443722757,
642
+ 622.1325731950062
643
+ ],
644
+ "size": [
645
+ 140,
646
+ 30
647
+ ],
648
+ "flags": {},
649
+ "order": 19,
650
+ "mode": 0,
651
+ "inputs": [
652
+ {
653
+ "name": "images",
654
+ "type": "IMAGE",
655
+ "link": 264
656
+ }
657
+ ],
658
+ "outputs": [
659
+ {
660
+ "name": "images",
661
+ "type": "IMAGE",
662
+ "links": [
663
+ 150,
664
+ 257
665
+ ]
666
+ }
667
+ ],
668
+ "properties": {
669
+ "cnr_id": "comfy-core",
670
+ "ver": "0.44.0",
671
+ "Node name for S&R": "PreviewImage"
672
+ }
673
+ },
674
+ {
675
+ "id": 80,
676
+ "type": "ImageCompare",
677
+ "pos": [
678
+ 2809.1192876429695,
679
+ -341.7154021407288
680
+ ],
681
+ "size": [
682
+ 930,
683
+ 780
684
+ ],
685
+ "flags": {},
686
+ "order": 27,
687
+ "mode": 0,
688
+ "inputs": [
689
+ {
690
+ "name": "image_a",
691
+ "shape": 7,
692
+ "type": "IMAGE",
693
+ "link": 150
694
+ },
695
+ {
696
+ "name": "image_b",
697
+ "shape": 7,
698
+ "type": "IMAGE",
699
+ "link": 149
700
+ }
701
+ ],
702
+ "outputs": [],
703
+ "properties": {
704
+ "cnr_id": "comfy-core",
705
+ "ver": "0.44.0",
706
+ "Node name for S&R": "ImageCompare"
707
+ },
708
+ "widgets_values": [],
709
+ "widgets_values_named": {}
710
+ },
711
+ {
712
+ "id": 83,
713
+ "type": "MarkdownNote",
714
+ "pos": [
715
+ -1230,
716
+ -520
717
+ ],
718
+ "size": [
719
+ 710,
720
+ 890
721
+ ],
722
+ "flags": {},
723
+ "order": 4,
724
+ "mode": 0,
725
+ "inputs": [],
726
+ "outputs": [],
727
+ "title": "Note: About",
728
+ "properties": {},
729
+ "widgets_values": [
730
+ "## MOSS-Image: Color & Depth Control from a Photo\n\nDrop in a photo and the graph estimates its geometry, then generates a new image that keeps the same pose and framing. No map to prepare by hand.\n\n**How it works**\n\n1. **Load Image** (node 55) - the character reference. It is scaled to 1.0 MP, 16 px aligned (node 77), and drives the output size through **Get Image Size** (node 61) -> **Empty Latent** (node 54).\n2. **Geometry Estimation (MoGe 3)** subgraph - `Load MoGe Model` (`moge_3_vitg_fp16`) -> `Run MoGe Inference` -> `Render MoGe Geometry`. Its output is the `depth_map` of **MOSS-Image Color/Depth Control** (node 53).\n3. **Positive Prompt** (node 5) describes the character and the scene. Geometry comes from the render, everything else - identity, wardrobe, material, light, style - comes from the text.\n4. **MOSS-Image Color/Depth Control** (node 53) injects the map into the image tokens; the LoRA loaded just before it (`moss_image_color_depth_control.safetensors`, strength 1.0) is what makes the model read it. Without the LoRA the node is a no-op.\n5. **Save / Compare**: `VAE Decode` -> preview, plus **Compare Images** (node 80) holding the generated frame next to the rendered map.\n\n**Settings**\n\n`strength` 1.0 is the trained strength; `start_percent` / `end_percent` pick the part of the run that is controlled (0 to 0.6 here: the layout is fixed, the last steps render freely). Sampling follows the MOSS-Image control test graph: 25 steps, CFG 3, `er_sde` / `simple`, shift 5, skip layer guidance on layer 24, temporal score rescaling 0.95.\n\n**Useful knobs**\n\n- `Render MoGe Geometry` in the subgraph is set to `mask`: only the silhouette is used. Switch it to `depth` (`normalization = min_max`) when you want the face profile, shoulders and background falloff to be controlled too - the pose gets tighter, the prompt has less room.\n- `end_percent` 1.0 pins geometry to the last step; 0.5 frees the surface and lighting. For a mask-derived map 0.6 is a good middle.\n- The reference has **pointed elf ears**, and the geometry keeps them. Either keep that in the prompt (`pointed elf ears`) or hide them (`ears covered by hair`).\n- **Painter** (node 81) is in the graph for painting a map by hand: white near / black far for depth, one flat colour per region for colour. Connect its `IMAGE` to `depth_map` or `color_map` to override the estimate.\n- `color_map` is optional and currently unconnected. One flat colour per region, colours only act as region IDs, black means \"unspecified\"; it must match the output size (1024 x 1024 here).\n"
731
+ ],
732
+ "widgets_values_named": {
733
+ "text": "## MOSS-Image: Color & Depth Control from a Photo\n\nDrop in a photo and the graph estimates its geometry, then generates a new image that keeps the same pose and framing. No map to prepare by hand.\n\n**How it works**\n\n1. **Load Image** (node 55) - the character reference. It is scaled to 1.0 MP, 16 px aligned (node 77), and drives the output size through **Get Image Size** (node 61) -> **Empty Latent** (node 54).\n2. **Geometry Estimation (MoGe 3)** subgraph - `Load MoGe Model` (`moge_3_vitg_fp16`) -> `Run MoGe Inference` -> `Render MoGe Geometry`. Its output is the `depth_map` of **MOSS-Image Color/Depth Control** (node 53).\n3. **Positive Prompt** (node 5) describes the character and the scene. Geometry comes from the render, everything else - identity, wardrobe, material, light, style - comes from the text.\n4. **MOSS-Image Color/Depth Control** (node 53) injects the map into the image tokens; the LoRA loaded just before it (`moss_image_color_depth_control.safetensors`, strength 1.0) is what makes the model read it. Without the LoRA the node is a no-op.\n5. **Save / Compare**: `VAE Decode` -> preview, plus **Compare Images** (node 80) holding the generated frame next to the rendered map.\n\n**Settings**\n\n`strength` 1.0 is the trained strength; `start_percent` / `end_percent` pick the part of the run that is controlled (0 to 0.6 here: the layout is fixed, the last steps render freely). Sampling follows the MOSS-Image control test graph: 25 steps, CFG 3, `er_sde` / `simple`, shift 5, skip layer guidance on layer 24, temporal score rescaling 0.95.\n\n**Useful knobs**\n\n- `Render MoGe Geometry` in the subgraph is set to `mask`: only the silhouette is used. Switch it to `depth` (`normalization = min_max`) when you want the face profile, shoulders and background falloff to be controlled too - the pose gets tighter, the prompt has less room.\n- `end_percent` 1.0 pins geometry to the last step; 0.5 frees the surface and lighting. For a mask-derived map 0.6 is a good middle.\n- The reference has **pointed elf ears**, and the geometry keeps them. Either keep that in the prompt (`pointed elf ears`) or hide them (`ears covered by hair`).\n- **Painter** (node 81) is in the graph for painting a map by hand: white near / black far for depth, one flat colour per region for colour. Connect its `IMAGE` to `depth_map` or `color_map` to override the estimate.\n- `color_map` is optional and currently unconnected. One flat colour per region, colours only act as region IDs, black means \"unspecified\"; it must match the output size (1024 x 1024 here).\n"
734
+ },
735
+ "color": "#222",
736
+ "bgcolor": "#000"
737
+ },
738
+ {
739
+ "id": 84,
740
+ "type": "MarkdownNote",
741
+ "pos": [
742
+ -1970,
743
+ -520
744
+ ],
745
+ "size": [
746
+ 520,
747
+ 1250
748
+ ],
749
+ "flags": {},
750
+ "order": 5,
751
+ "mode": 0,
752
+ "inputs": [],
753
+ "outputs": [],
754
+ "title": "Note: Model Links",
755
+ "properties": {},
756
+ "widgets_values": [
757
+ "## Model Links\n\n- [Hugging Face: Comfy-Org/MOSS-Image](https://huggingface.co/Comfy-Org/MOSS-Image)\n- [Hugging Face: Comfy-Org/MoGe](https://huggingface.co/Comfy-Org/MoGe) (geometry estimation)\n- [Upstream control adapter: darask0/MOSS-Image-color-depth-control](https://huggingface.co/darask0/MOSS-Image-color-depth-control)\n\n**diffusion_models**\n\n- [moss_image_sft_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors) - 5.41 GB, the checkpoint this graph loads\n- [moss_image_sft_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_bf16.safetensors) - 10.16 GB, higher precision, slower\n- [moss_image_base_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_base_int8_convrot.safetensors) - 5.41 GB, the base checkpoint runs in the same graph\n\n**text_encoders**\n\n- [qwen3.5_9b_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors) - 9.75 GB, the encoder this graph loads\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) - 7.34 GB, drop-in swap, saves 2.4 GB\n- [qwen3.5_9b_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_bf16.safetensors) - 19.31 GB, full precision\n\n**vae**\n\n- [flux2-vae_small.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors) - 249 MB, the one this graph loads\n- [flux2-vae.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae.safetensors) - 336 MB, full size, pixel-level differences only\n\n**loras**\n\n- [moss_image_color_depth_control.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors) - 313 MB, `LoraLoaderModelOnly` at strength 1.0\n\n**geometry_estimation**\n\n- [moge_3_vitg_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitg_fp16.safetensors) - 2.50 GB, the depth estimator used by the subgraph\n- [moge_3_vitl_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitl_fp16.safetensors) - 741 MB, the lighter ViT-L variant\n\n**Disk footprint**: the graph as shipped needs **18.23 GB** (18,226,010,396 bytes) (5.41 + 9.75 + 0.25 + 0.31 + 2.50). Swapping the text encoder to `w4a8_gptq` brings it to **15.82 GB**; a bf16 checkpoint and encoder instead push it past 32 GB.\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 diffusion_models/\n│ │ └── moss_image_sft_int8_convrot.safetensors 5.41 GB\n│ ├── 📂 text_encoders/\n│ │ └── qwen3.5_9b_int8_convrot.safetensors 9.75 GB\n│ ├── 📂 vae/\n│ │ └── flux2-vae_small.safetensors 249 MB\n│ ├── 📂 loras/\n│ │ └── moss_image_color_depth_control.safetensors 313 MB\n│ └── 📂 geometry_estimation/\n│ └── moge_3_vitg_fp16.safetensors 2.50 GB\n```\n\nCore support (`MOSSImageControl`, MoGe nodes) ships from ComfyUI **0.43.0**.\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
758
+ ],
759
+ "widgets_values_named": {
760
+ "text": "## Model Links\n\n- [Hugging Face: Comfy-Org/MOSS-Image](https://huggingface.co/Comfy-Org/MOSS-Image)\n- [Hugging Face: Comfy-Org/MoGe](https://huggingface.co/Comfy-Org/MoGe) (geometry estimation)\n- [Upstream control adapter: darask0/MOSS-Image-color-depth-control](https://huggingface.co/darask0/MOSS-Image-color-depth-control)\n\n**diffusion_models**\n\n- [moss_image_sft_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors) - 5.41 GB, the checkpoint this graph loads\n- [moss_image_sft_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_bf16.safetensors) - 10.16 GB, higher precision, slower\n- [moss_image_base_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_base_int8_convrot.safetensors) - 5.41 GB, the base checkpoint runs in the same graph\n\n**text_encoders**\n\n- [qwen3.5_9b_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors) - 9.75 GB, the encoder this graph loads\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) - 7.34 GB, drop-in swap, saves 2.4 GB\n- [qwen3.5_9b_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_bf16.safetensors) - 19.31 GB, full precision\n\n**vae**\n\n- [flux2-vae_small.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors) - 249 MB, the one this graph loads\n- [flux2-vae.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae.safetensors) - 336 MB, full size, pixel-level differences only\n\n**loras**\n\n- [moss_image_color_depth_control.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors) - 313 MB, `LoraLoaderModelOnly` at strength 1.0\n\n**geometry_estimation**\n\n- [moge_3_vitg_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitg_fp16.safetensors) - 2.50 GB, the depth estimator used by the subgraph\n- [moge_3_vitl_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitl_fp16.safetensors) - 741 MB, the lighter ViT-L variant\n\n**Disk footprint**: the graph as shipped needs **18.23 GB** (18,226,010,396 bytes) (5.41 + 9.75 + 0.25 + 0.31 + 2.50). Swapping the text encoder to `w4a8_gptq` brings it to **15.82 GB**; a bf16 checkpoint and encoder instead push it past 32 GB.\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 diffusion_models/\n│ │ └── moss_image_sft_int8_convrot.safetensors 5.41 GB\n│ ├── 📂 text_encoders/\n│ │ └── qwen3.5_9b_int8_convrot.safetensors 9.75 GB\n│ ├── 📂 vae/\n│ │ └── flux2-vae_small.safetensors 249 MB\n│ ├── 📂 loras/\n│ │ └── moss_image_color_depth_control.safetensors 313 MB\n│ └── 📂 geometry_estimation/\n│ └── moge_3_vitg_fp16.safetensors 2.50 GB\n```\n\nCore support (`MOSSImageControl`, MoGe nodes) ships from ComfyUI **0.43.0**.\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
761
+ },
762
+ "color": "#222",
763
+ "bgcolor": "#000"
764
+ },
765
+ {
766
+ "id": 85,
767
+ "type": "SaveImage",
768
+ "pos": [
769
+ 2203.770368560271,
770
+ 538.7450846522054
771
+ ],
772
+ "size": [
773
+ 420,
774
+ 470
775
+ ],
776
+ "flags": {},
777
+ "order": 28,
778
+ "mode": 0,
779
+ "inputs": [
780
+ {
781
+ "name": "images",
782
+ "type": "IMAGE",
783
+ "link": 228
784
+ }
785
+ ],
786
+ "outputs": [
787
+ {
788
+ "name": "images",
789
+ "type": "IMAGE",
790
+ "links": null
791
+ }
792
+ ],
793
+ "properties": {
794
+ "cnr_id": "comfy-core",
795
+ "ver": "0.44.0",
796
+ "Node name for S&R": "SaveImage"
797
+ },
798
+ "widgets_values": [
799
+ "image_moss_image_color_depth_control"
800
+ ],
801
+ "widgets_values_named": {
802
+ "filename_prefix": "image_moss_image_color_depth_control"
803
+ }
804
+ },
805
+ {
806
+ "id": 86,
807
+ "type": "KSamplerSelect",
808
+ "pos": [
809
+ 1170.4423572560272,
810
+ -260.582916386391
811
+ ],
812
+ "size": [
813
+ 270,
814
+ 58
815
+ ],
816
+ "flags": {},
817
+ "order": 6,
818
+ "mode": 0,
819
+ "inputs": [],
820
+ "outputs": [
821
+ {
822
+ "name": "SAMPLER",
823
+ "type": "SAMPLER",
824
+ "links": [
825
+ 154
826
+ ]
827
+ }
828
+ ],
829
+ "properties": {
830
+ "cnr_id": "comfy-core",
831
+ "ver": "0.44.0",
832
+ "Node name for S&R": "KSamplerSelect"
833
+ },
834
+ "widgets_values": [
835
+ "euler"
836
+ ],
837
+ "widgets_values_named": {
838
+ "sampler_name": "euler"
839
+ }
840
+ },
841
+ {
842
+ "id": 87,
843
+ "type": "ImageScaleBy",
844
+ "pos": [
845
+ -7.178545924871845,
846
+ 681.7385897176745
847
+ ],
848
+ "size": [
849
+ 270,
850
+ 82
851
+ ],
852
+ "flags": {},
853
+ "order": 13,
854
+ "mode": 4,
855
+ "inputs": [
856
+ {
857
+ "name": "image",
858
+ "type": "IMAGE",
859
+ "link": 155
860
+ }
861
+ ],
862
+ "outputs": [
863
+ {
864
+ "name": "IMAGE",
865
+ "type": "IMAGE",
866
+ "links": [
867
+ 161
868
+ ]
869
+ }
870
+ ],
871
+ "properties": {
872
+ "cnr_id": "comfy-core",
873
+ "ver": "0.44.0",
874
+ "Node name for S&R": "ImageScaleBy"
875
+ },
876
+ "widgets_values": [
877
+ "lanczos",
878
+ 0.5
879
+ ],
880
+ "widgets_values_named": {
881
+ "upscale_method": "lanczos",
882
+ "scale_by": 0.5
883
+ }
884
+ },
885
+ {
886
+ "id": 88,
887
+ "type": "LoraLoaderModelOnly",
888
+ "pos": [
889
+ -447.13559989481814,
890
+ -263.5408892596166
891
+ ],
892
+ "size": [
893
+ 500,
894
+ 120
895
+ ],
896
+ "flags": {},
897
+ "order": 15,
898
+ "mode": 0,
899
+ "inputs": [
900
+ {
901
+ "name": "model",
902
+ "type": "MODEL",
903
+ "link": 157
904
+ }
905
+ ],
906
+ "outputs": [
907
+ {
908
+ "name": "MODEL",
909
+ "type": "MODEL",
910
+ "links": [
911
+ 260
912
+ ]
913
+ }
914
+ ],
915
+ "properties": {
916
+ "cnr_id": "comfy-core",
917
+ "ver": "0.44.0",
918
+ "Node name for S&R": "LoraLoaderModelOnly",
919
+ "models": [
920
+ {
921
+ "name": "moss_image_color_depth_control.safetensors",
922
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
923
+ "directory": "loras"
924
+ }
925
+ ]
926
+ },
927
+ "widgets_values": [
928
+ "moss/download/Coffmeteor-sft-15step-cfg3.safetensors",
929
+ 1
930
+ ],
931
+ "widgets_values_named": {
932
+ "lora_name": "moss/download/Coffmeteor-sft-15step-cfg3.safetensors",
933
+ "strength_model": 1
934
+ }
935
+ },
936
+ {
937
+ "id": 90,
938
+ "type": "ImageScaleBy",
939
+ "pos": [
940
+ 321.18555329294736,
941
+ 683.4828563588695
942
+ ],
943
+ "size": [
944
+ 270,
945
+ 82
946
+ ],
947
+ "flags": {},
948
+ "order": 16,
949
+ "mode": 0,
950
+ "inputs": [
951
+ {
952
+ "name": "image",
953
+ "type": "IMAGE",
954
+ "link": 161
955
+ }
956
+ ],
957
+ "outputs": [
958
+ {
959
+ "name": "IMAGE",
960
+ "type": "IMAGE",
961
+ "links": [
962
+ 264,
963
+ 265
964
+ ]
965
+ }
966
+ ],
967
+ "properties": {
968
+ "cnr_id": "comfy-core",
969
+ "ver": "0.44.0",
970
+ "Node name for S&R": "ImageScaleBy"
971
+ },
972
+ "widgets_values": [
973
+ "lanczos",
974
+ 4
975
+ ],
976
+ "widgets_values_named": {
977
+ "upscale_method": "lanczos",
978
+ "scale_by": 4
979
+ }
980
+ },
981
+ {
982
+ "id": 91,
983
+ "type": "PrimitiveStringMultiline",
984
+ "pos": [
985
+ 375.1526388094519,
986
+ -2150.2072456067854
987
+ ],
988
+ "size": [
989
+ 470,
990
+ 270
991
+ ],
992
+ "flags": {
993
+ "collapsed": false
994
+ },
995
+ "order": 7,
996
+ "mode": 4,
997
+ "inputs": [],
998
+ "outputs": [
999
+ {
1000
+ "name": "STRING",
1001
+ "type": "STRING",
1002
+ "links": [
1003
+ 163
1004
+ ]
1005
+ }
1006
+ ],
1007
+ "title": "Text (System prompt)",
1008
+ "properties": {
1009
+ "cnr_id": "comfy-core",
1010
+ "ver": "0.42.0",
1011
+ "Node name for S&R": "PrimitiveStringMultiline"
1012
+ },
1013
+ "widgets_values": [
1014
+ "\n你是一名顶级视觉逆向分析专家、数字绘画研究员和 AI 图像生成提示词工程师。\n\n你的唯一任务是:对用户提供的参考图像进行极其细致的视觉分析,并将所有具有复现价值的视觉特征整合为一段高精度、可直接用于 AI 图像生成模型的英文提示词。\n\n【核心要求】\n\n你必须在内部完成充分、深入的图像分析,但绝对不能向用户展示分析过程、分析结果、推理步骤、分类标题或任何解释。\n\n最终回答只能包含英文 Positive Prompt 本身。\n\n禁止输出:\n- 任何中文内容\n- 开场白或结束语\n- 标题、编号、分类标签\n- Markdown 格式或代码块\n- 分析过程和解释\n- Negative Prompt\n- 参数建议\n- 额外说明\n- 与提示词无关的任何文字\n\n【分析深度】\n\n在内部对图像进行像素级视觉特征观察,尽可能细致地识别:\n\n1. 整体艺术风格、媒介特征、风格融合方式、二维或三维表现倾向。\n\n2. 线稿系统:外轮廓粗细、内部线条粗细、线宽变化、线条收尖、线条转折、曲率、闭合程度、线条颜色、线条疏密、边缘平滑度。\n\n3. 形状语言:几何概括方式、曲线与直线比例、轮廓简化程度、局部夸张、形状节奏、视觉体积的建立方式。\n\n4. 上色方法:平涂、赛璐璐、厚涂、渐变、色块分层、色彩边界、硬边与软边过渡、局部混合渲染。\n\n5. 阴影结构:阴影边界锐利度、阴影形状、阴影色相、明暗层级、结构性阴影、接触阴影、装饰性阴影、阴影与形体的关系。\n\n6. 高光系统:高光形状、大小、分布、方向、边缘硬度、色彩、亮度、反射效果、块状高光、条状高光、局部辉光。\n\n7. 配色逻辑:主色调、辅助色、强调色、色彩饱和度、明度分布、冷暖对比、亮暗部色相偏移、色彩统一性。\n\n8. 光照处理:主光方向、环境光、轮廓光、光照对比度、局部泛光、光影风格化程度。\n\n9. 渲染质感:数字绘画感、笔触可见度、纹理密度、边缘锐度、画面平滑度、颗粒感、印刷感、类似矢量图的视觉质感。\n\n10. 细节组织:视觉焦点、细节密度、局部简化、主次关系、大中小形状的组合、画面精致程度。\n\n11. 构图与画面语言:景别、视角、透视、裁切、空间层次、主体位置、视觉动线。\n\n12. 画面内容:主体外观、服装、表情、姿态、道具、背景和场景元素。\n\n【反推规则】\n\n1. 默认采用纯画风反推模式:只提取能够迁移到其他主体的绘画风格,不描述具体人物、体型、性别、肌肉、服装、动作或场景。\n\n2. 如果用户明确要求完整反推,则同时还原主体、场景、构图和绘画风格。\n\n3. 如果提供多张参考图,提取所有图像共同、稳定且具有辨识度的视觉特征,不把单张图片独有的元素误认为共同画风。\n\n4. 不允许只用 anime style、beautiful illustration、detailed shading、high quality 等笼统词汇代替具体的视觉描述。\n\n5. 优先使用能够控制画面视觉效果的具体英文短语,准确描述线条、阴影、高光、色块、色彩、边缘和渲染方式。\n\n6. 区分硬边阴影与柔和渐变、结构性高光与装饰性高光、真实材质质感与绘画渲染质感。\n\n7. 不得凭空猜测原作者、绘画软件、具体模型、LoRA 或原始提示词。\n\n8. 仅描述有视觉证据支持的特征,不虚构不可见的细节。\n\n9. 将最有辨识度、最影响生成结果的特征放在提示词前部。\n\n10. 避免重复堆砌近义词,避免相互矛盾的描述,避免无意义的质量标签。\n\n11. 允许详细、丰富、具体的提示词,但每个短语都必须具有实际的视觉控制价值。\n\n12. 将所有分析结果重新组织为流畅、准确、信息密度高的英文生图提示词,而不是逐条罗列分析项目。\n\n13. 优先使用适合 MOSS-Image 等自然语言生图模型的英文描述。若用户明确指定 Anima 或 Danbooru 标签格式,则转换为适合该模型的英文标签式提示词。\n\n14. 对无法确认的视觉细节不进行猜测,不为了增加长度而编造特征。\n\n【最终输出规范——最高优先级】\n\n最终输出必须严格遵守以下要求:\n\n只输出一段英文 Positive Prompt。\n\n不允许出现任何标题。\n不允许出现任何分析内容。\n不允许解释为什么选择这些词。\n不允许输出 Negative Prompt。\n不允许附加任何建议。\n不允许添加引号或代码块。\n不允许输出第二个版本。\n\n所有视觉分析仅在内部完成,最终仅保留经过筛选、整合和优化的高精度提示词。\n\n无论分析多么复杂,输出形式都不能改变。\n\n用户上传图像后,直接输出最终英文提示词,不要进行任何说明。\n"
1015
+ ],
1016
+ "widgets_values_named": {
1017
+ "value": "\n你是一名顶级视觉逆向分析专家、数字绘画研究员和 AI 图像生成提示词工程师。\n\n你的唯一任务是:对用户提供的参考图像进行极其细致的视觉分析,并将所有具有复现价值的视觉特征整合为一段高精度、可直接用于 AI 图像生成模型的英文提示词。\n\n【核心要求】\n\n你必须在内部完成充分、深入的图像分析,但绝对不能向用户展示分析过程、分析结果、推理步骤、分类标题或任何解释。\n\n最终回答只能包含英文 Positive Prompt 本身。\n\n禁止输出:\n- 任何中文内容\n- 开场白或结束语\n- 标题、编号、分类标签\n- Markdown 格式或代码块\n- 分析过程和解释\n- Negative Prompt\n- 参数建议\n- 额外说明\n- 与提示词无关的任何文字\n\n【分析深度】\n\n在内部对图像进行像素级视觉特征观察,尽可能细致地识别:\n\n1. 整体艺术风格、媒介特征、风格融合方式、二维或三维表现倾向。\n\n2. 线稿系统:外轮廓粗细、内部线条粗细、线宽变化、线条收尖、线条转折、曲率、闭合程度、线条颜色、线条疏密、边缘平滑度。\n\n3. 形状语言:几何概括方式、曲线与直线比例、轮廓简化程度、局部夸张、形状节奏、视觉体积的建立方式。\n\n4. 上色方法:平涂、赛璐璐、厚涂、渐变、色块分层、色彩边界、硬边与软边过渡、局部混合渲染。\n\n5. 阴影结构:阴影边界锐利度、阴影形状、阴影色相、明暗层级、结构性阴影、接触阴影、装饰性阴影、阴影与形体的关系。\n\n6. 高光系统:高光形状、大小、分布、方向、边缘硬度、色彩、亮度、反射效果、块状高光、条状高光、局部辉光。\n\n7. 配色逻辑:主色调、辅助色、强调色、色彩饱和度、明度分布、冷暖对比、亮暗部色相偏移、色彩统一性。\n\n8. 光照处理:主光方向、环境光、轮廓光、光照对比度、局部泛光、光影风格化程度。\n\n9. 渲染质感:数字绘画感、笔触可见度、纹理密度、边缘锐度、画面平滑度、颗粒感、印刷感、类似矢量图的视觉质感。\n\n10. 细节组织:视觉焦点、细节密度、局部简化、主次关系、大中小形状的组合、画面精致程度。\n\n11. 构图与画面语言:景别、视角、透视、裁切、空间层次、主体位置、视觉动线。\n\n12. 画面内容:主体外观、服装、表情、姿态、道具、背景和场景元素。\n\n【反推规则】\n\n1. 默认采用纯画风反推模式:只提取能够迁移到其他主体的绘画风格,不描述具体人物、体型、性别、肌肉、服装、动作或场景。\n\n2. 如果用户明确要求完整反推,则同时还原主体、场景、构图和绘画风格。\n\n3. 如果提供多张参考图,提取所有图像共同、稳定且具有辨识度的视觉特征,不把单张图片独有的元素误认为共同画风。\n\n4. 不允许只用 anime style、beautiful illustration、detailed shading、high quality 等笼统词汇代替具体的视觉描述。\n\n5. 优先使用能够控制画面视觉效果的具体英文短语,准确描述线条、阴影、高光、色块、色彩、边缘和渲染方式。\n\n6. 区分硬边阴影与柔和渐变、结构性高光与装饰性高光、真实材质质感与绘画渲染质感。\n\n7. 不得凭空猜测原作者、绘画软件、具体模型、LoRA 或原始提示词。\n\n8. 仅描述有视觉证据支持的特征,不虚构不可见的细节。\n\n9. 将最有辨识度、最影响生成结果的特征放在提示词前部。\n\n10. 避免重复堆砌近义词,避免相互矛盾的描述,避免无意义的质量标签。\n\n11. 允许详细、丰富、具体的提示词,但每个短语都必须具有实际的视觉控制价值。\n\n12. 将所有分析结果重新组织为流畅、准确、信息密度高的英文生图提示词,而不是逐条罗列分析项目。\n\n13. 优先使用适合 MOSS-Image 等自然语言生图模型的英文描述。若用户明确指定 Anima 或 Danbooru 标签格式,则转换为适合该模型的英文标签式提示词。\n\n14. 对无法确认的视觉细节不进行猜测,不为了增加长度而编造特征。\n\n【最终输出规范——最高优先级】\n\n最终输出必须严格遵守以下要求:\n\n只输出一段英文 Positive Prompt。\n\n不允许出现任何标题。\n不允许出现任何分析内容。\n不允许解释为什么选择这些词。\n不允许输出 Negative Prompt。\n不允许附加任何建议。\n不允许添加引号或代码块。\n不允��输出第二个版本。\n\n所有视觉分析仅在内部完成,最终仅保留经过筛选、整合和优化的高精度提示词。\n\n无论分析多么复杂,输出形式都不能改变。\n\n用户上传图像后,直接输出最终英文提示词,不要进行任何说明。\n"
1018
+ }
1019
+ },
1020
+ {
1021
+ "id": 92,
1022
+ "type": "TextGenerate",
1023
+ "pos": [
1024
+ 1171.0822008806106,
1025
+ -2365.5035628809155
1026
+ ],
1027
+ "size": [
1028
+ 400,
1029
+ 550
1030
+ ],
1031
+ "flags": {},
1032
+ "order": 14,
1033
+ "mode": 4,
1034
+ "inputs": [
1035
+ {
1036
+ "name": "clip",
1037
+ "type": "CLIP",
1038
+ "link": 165
1039
+ },
1040
+ {
1041
+ "name": "image",
1042
+ "shape": 7,
1043
+ "type": "IMAGE",
1044
+ "link": 248
1045
+ },
1046
+ {
1047
+ "name": "video",
1048
+ "shape": 7,
1049
+ "type": "IMAGE",
1050
+ "link": null
1051
+ },
1052
+ {
1053
+ "name": "audio",
1054
+ "shape": 7,
1055
+ "type": "AUDIO",
1056
+ "link": null
1057
+ },
1058
+ {
1059
+ "name": "system_prompt",
1060
+ "shape": 7,
1061
+ "type": "STRING",
1062
+ "link": 163
1063
+ }
1064
+ ],
1065
+ "outputs": [
1066
+ {
1067
+ "name": "generated_text",
1068
+ "type": "STRING",
1069
+ "links": [
1070
+ 166
1071
+ ]
1072
+ },
1073
+ {
1074
+ "name": "thinking",
1075
+ "type": "STRING",
1076
+ "links": []
1077
+ }
1078
+ ],
1079
+ "properties": {
1080
+ "cnr_id": "comfy-core",
1081
+ "ver": "0.42.0",
1082
+ "Node name for S&R": "TextGenerate"
1083
+ },
1084
+ "widgets_values": [
1085
+ "",
1086
+ 4096,
1087
+ "on",
1088
+ 0.7,
1089
+ 64,
1090
+ 0.95,
1091
+ 0.05,
1092
+ 1,
1093
+ 0,
1094
+ 0,
1095
+ true,
1096
+ true,
1097
+ "auto"
1098
+ ],
1099
+ "widgets_values_named": {
1100
+ "prompt": "",
1101
+ "max_length": 4096,
1102
+ "sampling_mode": "on",
1103
+ "sampling_mode.temperature": 0.7,
1104
+ "sampling_mode.top_k": 64,
1105
+ "sampling_mode.top_p": 0.95,
1106
+ "sampling_mode.min_p": 0.05,
1107
+ "sampling_mode.repetition_penalty": 1,
1108
+ "sampling_mode.seed": 0,
1109
+ "sampling_mode.presence_penalty": 0,
1110
+ "thinking": true,
1111
+ "use_default_template": true,
1112
+ "mtp": "auto"
1113
+ }
1114
+ },
1115
+ {
1116
+ "id": 93,
1117
+ "type": "CLIPLoader",
1118
+ "pos": [
1119
+ 468.0082400074318,
1120
+ -2370.6899919511707
1121
+ ],
1122
+ "size": [
1123
+ 270,
1124
+ 106
1125
+ ],
1126
+ "flags": {},
1127
+ "order": 8,
1128
+ "mode": 4,
1129
+ "inputs": [],
1130
+ "outputs": [
1131
+ {
1132
+ "name": "CLIP",
1133
+ "type": "CLIP",
1134
+ "links": [
1135
+ 165
1136
+ ]
1137
+ }
1138
+ ],
1139
+ "properties": {
1140
+ "cnr_id": "comfy-core",
1141
+ "ver": "0.44.0",
1142
+ "Node name for S&R": "CLIPLoader"
1143
+ },
1144
+ "widgets_values": [
1145
+ "qwen3.5-9b-nsfw-captioning-v5.safetensors",
1146
+ "moss_image",
1147
+ "default"
1148
+ ],
1149
+ "widgets_values_named": {
1150
+ "clip_name": "qwen3.5-9b-nsfw-captioning-v5.safetensors",
1151
+ "type": "moss_image",
1152
+ "device": "default"
1153
+ }
1154
+ },
1155
+ {
1156
+ "id": 94,
1157
+ "type": "PreviewAny",
1158
+ "pos": [
1159
+ 1713.012250094057,
1160
+ -2364.526445608352
1161
+ ],
1162
+ "size": [
1163
+ 470.96127217369775,
1164
+ 425.59604314573403
1165
+ ],
1166
+ "flags": {},
1167
+ "order": 17,
1168
+ "mode": 4,
1169
+ "inputs": [
1170
+ {
1171
+ "name": "source",
1172
+ "type": "*",
1173
+ "link": 166
1174
+ }
1175
+ ],
1176
+ "outputs": [
1177
+ {
1178
+ "name": "STRING",
1179
+ "type": "STRING",
1180
+ "links": []
1181
+ }
1182
+ ],
1183
+ "properties": {
1184
+ "cnr_id": "comfy-core",
1185
+ "ver": "0.44.0",
1186
+ "Node name for S&R": "PreviewAny"
1187
+ },
1188
+ "widgets_values": [],
1189
+ "widgets_values_named": {}
1190
+ },
1191
+ {
1192
+ "id": 124,
1193
+ "type": "VAEEncode",
1194
+ "pos": [
1195
+ 1149.2279511667382,
1196
+ 242.2625195138657
1197
+ ],
1198
+ "size": [
1199
+ 140,
1200
+ 46
1201
+ ],
1202
+ "flags": {},
1203
+ "order": 20,
1204
+ "mode": 0,
1205
+ "inputs": [
1206
+ {
1207
+ "name": "pixels",
1208
+ "type": "IMAGE",
1209
+ "link": 265
1210
+ },
1211
+ {
1212
+ "name": "vae",
1213
+ "type": "VAE",
1214
+ "link": 246
1215
+ }
1216
+ ],
1217
+ "outputs": [
1218
+ {
1219
+ "name": "LATENT",
1220
+ "type": "LATENT",
1221
+ "links": [
1222
+ 247
1223
+ ]
1224
+ }
1225
+ ],
1226
+ "properties": {
1227
+ "cnr_id": "comfy-core",
1228
+ "ver": "0.44.0",
1229
+ "Node name for S&R": "VAEEncode"
1230
+ }
1231
+ },
1232
+ {
1233
+ "id": 125,
1234
+ "type": "MossTileControlLoader",
1235
+ "pos": [
1236
+ 758.3282271676215,
1237
+ -61.473909743927265
1238
+ ],
1239
+ "size": [
1240
+ 270,
1241
+ 58
1242
+ ],
1243
+ "flags": {},
1244
+ "order": 9,
1245
+ "mode": 0,
1246
+ "inputs": [],
1247
+ "outputs": [
1248
+ {
1249
+ "name": "MOSS_TILE_CONTROL",
1250
+ "type": "MOSS_TILE_CONTROL",
1251
+ "links": [
1252
+ 249
1253
+ ]
1254
+ }
1255
+ ],
1256
+ "properties": {
1257
+ "Node name for S&R": "MossTileControlLoader"
1258
+ },
1259
+ "widgets_values": [
1260
+ "moss-tile-control/beta3/epoch2/adapter_model.safetensors"
1261
+ ],
1262
+ "widgets_values_named": {
1263
+ "name": "moss-tile-control/beta3/epoch2/adapter_model.safetensors"
1264
+ }
1265
+ },
1266
+ {
1267
+ "id": 126,
1268
+ "type": "MossTileControlApply",
1269
+ "pos": [
1270
+ 759.0227970689675,
1271
+ 111.48495135711164
1272
+ ],
1273
+ "size": [
1274
+ 270,
1275
+ 166
1276
+ ],
1277
+ "flags": {},
1278
+ "order": 22,
1279
+ "mode": 0,
1280
+ "inputs": [
1281
+ {
1282
+ "name": "model",
1283
+ "type": "MODEL",
1284
+ "link": 255
1285
+ },
1286
+ {
1287
+ "name": "control",
1288
+ "type": "MOSS_TILE_CONTROL",
1289
+ "link": 249
1290
+ },
1291
+ {
1292
+ "name": "vae",
1293
+ "type": "VAE",
1294
+ "link": 256
1295
+ },
1296
+ {
1297
+ "name": "image",
1298
+ "type": "IMAGE",
1299
+ "link": 257
1300
+ }
1301
+ ],
1302
+ "outputs": [
1303
+ {
1304
+ "name": "MODEL",
1305
+ "type": "MODEL",
1306
+ "links": [
1307
+ 253,
1308
+ 254
1309
+ ]
1310
+ }
1311
+ ],
1312
+ "properties": {
1313
+ "Node name for S&R": "MossTileControlApply"
1314
+ },
1315
+ "widgets_values": [
1316
+ 0.5,
1317
+ 0,
1318
+ 1
1319
+ ],
1320
+ "widgets_values_named": {
1321
+ "strength": 0.5,
1322
+ "start_percent": 0,
1323
+ "end_percent": 1
1324
+ }
1325
+ }
1326
+ ],
1327
+ "links": [
1328
+ [
1329
+ 3,
1330
+ 3,
1331
+ 0,
1332
+ 5,
1333
+ 0,
1334
+ "CLIP"
1335
+ ],
1336
+ [
1337
+ 4,
1338
+ 3,
1339
+ 0,
1340
+ 6,
1341
+ 0,
1342
+ "CLIP"
1343
+ ],
1344
+ [
1345
+ 9,
1346
+ 4,
1347
+ 0,
1348
+ 9,
1349
+ 1,
1350
+ "VAE"
1351
+ ],
1352
+ [
1353
+ 25,
1354
+ 20,
1355
+ 0,
1356
+ 9,
1357
+ 0,
1358
+ "LATENT"
1359
+ ],
1360
+ [
1361
+ 27,
1362
+ 22,
1363
+ 0,
1364
+ 20,
1365
+ 4,
1366
+ "SIGMAS"
1367
+ ],
1368
+ [
1369
+ 69,
1370
+ 36,
1371
+ 0,
1372
+ 2,
1373
+ 0,
1374
+ "MODEL"
1375
+ ],
1376
+ [
1377
+ 91,
1378
+ 1,
1379
+ 0,
1380
+ 49,
1381
+ 0,
1382
+ "MODEL"
1383
+ ],
1384
+ [
1385
+ 149,
1386
+ 42,
1387
+ 0,
1388
+ 80,
1389
+ 1,
1390
+ "IMAGE"
1391
+ ],
1392
+ [
1393
+ 150,
1394
+ 79,
1395
+ 0,
1396
+ 80,
1397
+ 0,
1398
+ "IMAGE"
1399
+ ],
1400
+ [
1401
+ 154,
1402
+ 86,
1403
+ 0,
1404
+ 20,
1405
+ 3,
1406
+ "SAMPLER"
1407
+ ],
1408
+ [
1409
+ 155,
1410
+ 55,
1411
+ 0,
1412
+ 87,
1413
+ 0,
1414
+ "IMAGE"
1415
+ ],
1416
+ [
1417
+ 157,
1418
+ 49,
1419
+ 0,
1420
+ 88,
1421
+ 0,
1422
+ "MODEL"
1423
+ ],
1424
+ [
1425
+ 161,
1426
+ 87,
1427
+ 0,
1428
+ 90,
1429
+ 0,
1430
+ "IMAGE"
1431
+ ],
1432
+ [
1433
+ 166,
1434
+ 92,
1435
+ 0,
1436
+ 94,
1437
+ 0,
1438
+ "STRING"
1439
+ ],
1440
+ [
1441
+ 228,
1442
+ 42,
1443
+ 0,
1444
+ 85,
1445
+ 0,
1446
+ "IMAGE"
1447
+ ],
1448
+ [
1449
+ 243,
1450
+ 9,
1451
+ 0,
1452
+ 42,
1453
+ 0,
1454
+ "IMAGE"
1455
+ ],
1456
+ [
1457
+ 246,
1458
+ 4,
1459
+ 0,
1460
+ 124,
1461
+ 1,
1462
+ "VAE"
1463
+ ],
1464
+ [
1465
+ 247,
1466
+ 124,
1467
+ 0,
1468
+ 20,
1469
+ 5,
1470
+ "LATENT"
1471
+ ],
1472
+ [
1473
+ 249,
1474
+ 125,
1475
+ 0,
1476
+ 126,
1477
+ 1,
1478
+ "MOSS_TILE_CONTROL"
1479
+ ],
1480
+ [
1481
+ 250,
1482
+ 5,
1483
+ 0,
1484
+ 20,
1485
+ 1,
1486
+ "CONDITIONING"
1487
+ ],
1488
+ [
1489
+ 251,
1490
+ 6,
1491
+ 0,
1492
+ 20,
1493
+ 2,
1494
+ "CONDITIONING"
1495
+ ],
1496
+ [
1497
+ 253,
1498
+ 126,
1499
+ 0,
1500
+ 20,
1501
+ 0,
1502
+ "MODEL"
1503
+ ],
1504
+ [
1505
+ 254,
1506
+ 126,
1507
+ 0,
1508
+ 22,
1509
+ 0,
1510
+ "MODEL"
1511
+ ],
1512
+ [
1513
+ 165,
1514
+ 93,
1515
+ 0,
1516
+ 92,
1517
+ 0,
1518
+ "CLIP"
1519
+ ],
1520
+ [
1521
+ 248,
1522
+ 55,
1523
+ 0,
1524
+ 92,
1525
+ 1,
1526
+ "IMAGE"
1527
+ ],
1528
+ [
1529
+ 163,
1530
+ 91,
1531
+ 0,
1532
+ 92,
1533
+ 4,
1534
+ "STRING"
1535
+ ],
1536
+ [
1537
+ 255,
1538
+ 2,
1539
+ 0,
1540
+ 126,
1541
+ 0,
1542
+ "MODEL"
1543
+ ],
1544
+ [
1545
+ 256,
1546
+ 4,
1547
+ 0,
1548
+ 126,
1549
+ 2,
1550
+ "VAE"
1551
+ ],
1552
+ [
1553
+ 257,
1554
+ 79,
1555
+ 0,
1556
+ 126,
1557
+ 3,
1558
+ "IMAGE"
1559
+ ],
1560
+ [
1561
+ 260,
1562
+ 88,
1563
+ 0,
1564
+ 36,
1565
+ 0,
1566
+ "MODEL"
1567
+ ],
1568
+ [
1569
+ 264,
1570
+ 90,
1571
+ 0,
1572
+ 79,
1573
+ 0,
1574
+ "IMAGE"
1575
+ ],
1576
+ [
1577
+ 265,
1578
+ 90,
1579
+ 0,
1580
+ 124,
1581
+ 0,
1582
+ "IMAGE"
1583
+ ]
1584
+ ],
1585
+ "groups": [
1586
+ {
1587
+ "id": 1,
1588
+ "title": "Models",
1589
+ "bounding": [
1590
+ -460,
1591
+ -570,
1592
+ 550,
1593
+ 950
1594
+ ],
1595
+ "color": "#3f789e",
1596
+ "flags": {}
1597
+ },
1598
+ {
1599
+ "id": 2,
1600
+ "title": "Conditioning",
1601
+ "bounding": [
1602
+ 120,
1603
+ -140,
1604
+ 970,
1605
+ 520
1606
+ ],
1607
+ "color": "#A88",
1608
+ "flags": {}
1609
+ },
1610
+ {
1611
+ "id": 3,
1612
+ "title": "Group",
1613
+ "bounding": [
1614
+ 120,
1615
+ -570,
1616
+ 970,
1617
+ 400
1618
+ ],
1619
+ "color": "#3f789e",
1620
+ "flags": {}
1621
+ },
1622
+ {
1623
+ "id": 4,
1624
+ "title": "Sampling",
1625
+ "bounding": [
1626
+ 1106.0160595404368,
1627
+ -500.08001764249127,
1628
+ 740,
1629
+ 950
1630
+ ],
1631
+ "color": "#3f789e",
1632
+ "flags": {}
1633
+ },
1634
+ {
1635
+ "id": 5,
1636
+ "title": "Control Images",
1637
+ "bounding": [
1638
+ -460,
1639
+ 410,
1640
+ 1550,
1641
+ 640
1642
+ ],
1643
+ "color": "#A88",
1644
+ "flags": {}
1645
+ }
1646
+ ],
1647
+ "config": {},
1648
+ "extra": {
1649
+ "ds": {
1650
+ "scale": 0.7627768444386022,
1651
+ "offset": [
1652
+ 1735.1036226056763,
1653
+ 1127.6731023127873
1654
+ ]
1655
+ },
1656
+ "frontendVersion": "1.58.5",
1657
+ "VHS_latentpreview": false,
1658
+ "VHS_latentpreviewrate": 0,
1659
+ "VHS_MetadataImage": true,
1660
+ "VHS_KeepIntermediate": true
1661
+ },
1662
+ "version": 0.4,
1663
+ "seed_widgets": {
1664
+ "20": 1
1665
+ }
1666
+ }
workflow/moss_image_tile-test-beta2-color-match.json ADDED
@@ -0,0 +1,1721 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "id": "5140b129-e927-49a6-816e-668c69d43008",
3
+ "revision": 0,
4
+ "last_node_id": 130,
5
+ "last_link_id": 276,
6
+ "nodes": [
7
+ {
8
+ "id": 1,
9
+ "type": "UNETLoader",
10
+ "pos": [
11
+ -423.26391629976365,
12
+ -689.4382881990866
13
+ ],
14
+ "size": [
15
+ 500,
16
+ 90
17
+ ],
18
+ "flags": {},
19
+ "order": 0,
20
+ "mode": 0,
21
+ "inputs": [],
22
+ "outputs": [
23
+ {
24
+ "name": "MODEL",
25
+ "type": "MODEL",
26
+ "slot_index": 0,
27
+ "links": [
28
+ 91
29
+ ]
30
+ }
31
+ ],
32
+ "properties": {
33
+ "cnr_id": "comfy-core",
34
+ "ver": "0.44.0",
35
+ "Node name for S&R": "UNETLoader",
36
+ "models": [
37
+ {
38
+ "name": "moss_image_sft_int8_convrot.safetensors",
39
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors",
40
+ "directory": "diffusion_models"
41
+ }
42
+ ]
43
+ },
44
+ "widgets_values": [
45
+ "moss-image/moss_image_sft_int8_convrot.safetensors",
46
+ "default"
47
+ ],
48
+ "widgets_values_named": {
49
+ "unet_name": "moss-image/moss_image_sft_int8_convrot.safetensors",
50
+ "weight_dtype": "default"
51
+ }
52
+ },
53
+ {
54
+ "id": 2,
55
+ "type": "SkipLayerGuidanceDiTSimple",
56
+ "pos": [
57
+ 620,
58
+ -460
59
+ ],
60
+ "size": [
61
+ 400,
62
+ 130
63
+ ],
64
+ "flags": {},
65
+ "order": 20,
66
+ "mode": 4,
67
+ "showAdvanced": true,
68
+ "inputs": [
69
+ {
70
+ "name": "model",
71
+ "type": "MODEL",
72
+ "link": 69
73
+ }
74
+ ],
75
+ "outputs": [
76
+ {
77
+ "name": "MODEL",
78
+ "type": "MODEL",
79
+ "slot_index": 0,
80
+ "links": [
81
+ 270,
82
+ 271
83
+ ]
84
+ }
85
+ ],
86
+ "title": "SkipLayerGuidanceDiTSimple (uncond skips block 22)",
87
+ "properties": {
88
+ "cnr_id": "comfy-core",
89
+ "ver": "0.44.0",
90
+ "Node name for S&R": "SkipLayerGuidanceDiTSimple"
91
+ },
92
+ "widgets_values": [
93
+ "",
94
+ "24",
95
+ 0,
96
+ 1
97
+ ],
98
+ "widgets_values_named": {
99
+ "double_layers": "",
100
+ "single_layers": "24",
101
+ "start_percent": 0,
102
+ "end_percent": 1
103
+ }
104
+ },
105
+ {
106
+ "id": 3,
107
+ "type": "CLIPLoader",
108
+ "pos": [
109
+ -440,
110
+ -50
111
+ ],
112
+ "size": [
113
+ 500,
114
+ 110
115
+ ],
116
+ "flags": {},
117
+ "order": 1,
118
+ "mode": 0,
119
+ "inputs": [],
120
+ "outputs": [
121
+ {
122
+ "name": "CLIP",
123
+ "type": "CLIP",
124
+ "slot_index": 0,
125
+ "links": [
126
+ 3,
127
+ 4
128
+ ]
129
+ }
130
+ ],
131
+ "properties": {
132
+ "cnr_id": "comfy-core",
133
+ "ver": "0.44.0",
134
+ "Node name for S&R": "CLIPLoader",
135
+ "models": [
136
+ {
137
+ "name": "qwen3.5_9b_int8_convrot.safetensors",
138
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors",
139
+ "directory": "text_encoders"
140
+ }
141
+ ]
142
+ },
143
+ "widgets_values": [
144
+ "qwen3.5_9b_bf16.safetensors",
145
+ "moss_image",
146
+ "default"
147
+ ],
148
+ "widgets_values_named": {
149
+ "clip_name": "qwen3.5_9b_bf16.safetensors",
150
+ "type": "moss_image",
151
+ "device": "default"
152
+ }
153
+ },
154
+ {
155
+ "id": 4,
156
+ "type": "VAELoader",
157
+ "pos": [
158
+ -440,
159
+ 130
160
+ ],
161
+ "size": [
162
+ 500,
163
+ 60
164
+ ],
165
+ "flags": {},
166
+ "order": 2,
167
+ "mode": 0,
168
+ "inputs": [],
169
+ "outputs": [
170
+ {
171
+ "name": "VAE",
172
+ "type": "VAE",
173
+ "slot_index": 0,
174
+ "links": [
175
+ 9,
176
+ 246,
177
+ 272
178
+ ]
179
+ }
180
+ ],
181
+ "properties": {
182
+ "cnr_id": "comfy-core",
183
+ "ver": "0.44.0",
184
+ "Node name for S&R": "VAELoader",
185
+ "models": [
186
+ {
187
+ "name": "flux2-vae_small.safetensors",
188
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors",
189
+ "directory": "vae"
190
+ }
191
+ ]
192
+ },
193
+ "widgets_values": [
194
+ "flux2-vae.safetensors"
195
+ ],
196
+ "widgets_values_named": {
197
+ "vae_name": "flux2-vae.safetensors"
198
+ }
199
+ },
200
+ {
201
+ "id": 5,
202
+ "type": "CLIPTextEncode",
203
+ "pos": [
204
+ 185.95627952212953,
205
+ -36.734098034606404
206
+ ],
207
+ "size": [
208
+ 470,
209
+ 240
210
+ ],
211
+ "flags": {},
212
+ "order": 10,
213
+ "mode": 0,
214
+ "inputs": [
215
+ {
216
+ "name": "clip",
217
+ "type": "CLIP",
218
+ "link": 3
219
+ }
220
+ ],
221
+ "outputs": [
222
+ {
223
+ "name": "CONDITIONING",
224
+ "type": "CONDITIONING",
225
+ "slot_index": 0,
226
+ "links": [
227
+ 266
228
+ ]
229
+ }
230
+ ],
231
+ "title": "Positive Prompt",
232
+ "properties": {
233
+ "cnr_id": "comfy-core",
234
+ "ver": "0.44.0",
235
+ "Node name for S&R": "CLIPTextEncode"
236
+ },
237
+ "widgets_values": [
238
+ "A highly realistic and detailed image that blends realistic photography with digital painting, featuring a grey tabby cat as the main subject. The cat is in a spread-eagle pose, flying through the air, with its front paws and hind legs stretched out. The cat is strapped into a flying drone. There are four propellers, two attached to the front paws and two to the hind legs, featuring silver hubs, black shafts, and bright orange blades. The drone frame is made of black metal rods, and red straps hold the cat securely in place. The cat's head is tilted upward, with wide eyes, small pupils, and a slightly open mouth, looking shocked or surprised. The fur is highly detailed, rendered with a painterly style that combines realism with digital art. The background is a soft, out-of-focus bokeh of green and white, suggesting a bright outdoor setting. The lighting is bright and natural, casting soft shadows that give the scene a solid, 3D illustration quality."
239
+ ],
240
+ "widgets_values_named": {
241
+ "text": "A highly realistic and detailed image that blends realistic photography with digital painting, featuring a grey tabby cat as the main subject. The cat is in a spread-eagle pose, flying through the air, with its front paws and hind legs stretched out. The cat is strapped into a flying drone. There are four propellers, two attached to the front paws and two to the hind legs, featuring silver hubs, black shafts, and bright orange blades. The drone frame is made of black metal rods, and red straps hold the cat securely in place. The cat's head is tilted upward, with wide eyes, small pupils, and a slightly open mouth, looking shocked or surprised. The fur is highly detailed, rendered with a painterly style that combines realism with digital art. The background is a soft, out-of-focus bokeh of green and white, suggesting a bright outdoor setting. The lighting is bright and natural, casting soft shadows that give the scene a solid, 3D illustration quality."
242
+ },
243
+ "color": "#232",
244
+ "bgcolor": "#353"
245
+ },
246
+ {
247
+ "id": 6,
248
+ "type": "CLIPTextEncode",
249
+ "pos": [
250
+ 150,
251
+ 250
252
+ ],
253
+ "size": [
254
+ 480,
255
+ 110
256
+ ],
257
+ "flags": {},
258
+ "order": 11,
259
+ "mode": 0,
260
+ "inputs": [
261
+ {
262
+ "name": "clip",
263
+ "type": "CLIP",
264
+ "link": 4
265
+ }
266
+ ],
267
+ "outputs": [
268
+ {
269
+ "name": "CONDITIONING",
270
+ "type": "CONDITIONING",
271
+ "slot_index": 0,
272
+ "links": [
273
+ 267
274
+ ]
275
+ }
276
+ ],
277
+ "title": "Negative Prompt",
278
+ "properties": {
279
+ "cnr_id": "comfy-core",
280
+ "ver": "0.44.0",
281
+ "Node name for S&R": "CLIPTextEncode"
282
+ },
283
+ "widgets_values": [
284
+ "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
285
+ ],
286
+ "widgets_values_named": {
287
+ "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact"
288
+ },
289
+ "color": "#223",
290
+ "bgcolor": "#335"
291
+ },
292
+ {
293
+ "id": 9,
294
+ "type": "VAEDecode",
295
+ "pos": [
296
+ 1910,
297
+ -520
298
+ ],
299
+ "size": [
300
+ 210,
301
+ 50
302
+ ],
303
+ "flags": {},
304
+ "order": 24,
305
+ "mode": 0,
306
+ "inputs": [
307
+ {
308
+ "name": "samples",
309
+ "type": "LATENT",
310
+ "link": 25
311
+ },
312
+ {
313
+ "name": "vae",
314
+ "type": "VAE",
315
+ "link": 9
316
+ }
317
+ ],
318
+ "outputs": [
319
+ {
320
+ "name": "IMAGE",
321
+ "type": "IMAGE",
322
+ "slot_index": 0,
323
+ "links": [
324
+ 274
325
+ ]
326
+ }
327
+ ],
328
+ "properties": {
329
+ "cnr_id": "comfy-core",
330
+ "ver": "0.44.0",
331
+ "Node name for S&R": "VAEDecode"
332
+ },
333
+ "color": "#322",
334
+ "bgcolor": "#533"
335
+ },
336
+ {
337
+ "id": 20,
338
+ "type": "SamplerCustom",
339
+ "pos": [
340
+ 1518.3539918681,
341
+ -304.93798438717295
342
+ ],
343
+ "size": [
344
+ 280,
345
+ 500
346
+ ],
347
+ "flags": {},
348
+ "order": 23,
349
+ "mode": 0,
350
+ "inputs": [
351
+ {
352
+ "name": "model",
353
+ "type": "MODEL",
354
+ "link": 270
355
+ },
356
+ {
357
+ "name": "positive",
358
+ "type": "CONDITIONING",
359
+ "link": 268
360
+ },
361
+ {
362
+ "name": "negative",
363
+ "type": "CONDITIONING",
364
+ "link": 269
365
+ },
366
+ {
367
+ "name": "sampler",
368
+ "type": "SAMPLER",
369
+ "link": 154
370
+ },
371
+ {
372
+ "name": "sigmas",
373
+ "type": "SIGMAS",
374
+ "link": 27
375
+ },
376
+ {
377
+ "name": "latent_image",
378
+ "type": "LATENT",
379
+ "link": 247
380
+ }
381
+ ],
382
+ "outputs": [
383
+ {
384
+ "name": "output",
385
+ "type": "LATENT",
386
+ "links": [
387
+ 25
388
+ ]
389
+ },
390
+ {
391
+ "name": "denoised_output",
392
+ "type": "LATENT",
393
+ "links": null
394
+ }
395
+ ],
396
+ "properties": {
397
+ "cnr_id": "comfy-core",
398
+ "ver": "0.44.0",
399
+ "Node name for S&R": "SamplerCustom"
400
+ },
401
+ "widgets_values": [
402
+ true,
403
+ 994529790995974,
404
+ "fixed",
405
+ 5
406
+ ],
407
+ "widgets_values_named": {
408
+ "add_noise": true,
409
+ "noise_seed": 994529790995974,
410
+ "control_after_generate": "fixed",
411
+ "cfg": 5
412
+ }
413
+ },
414
+ {
415
+ "id": 22,
416
+ "type": "BasicScheduler",
417
+ "pos": [
418
+ 1169.6140490851362,
419
+ 1.7699193319720672
420
+ ],
421
+ "size": [
422
+ 270,
423
+ 110
424
+ ],
425
+ "flags": {},
426
+ "order": 22,
427
+ "mode": 0,
428
+ "inputs": [
429
+ {
430
+ "name": "model",
431
+ "type": "MODEL",
432
+ "link": 271
433
+ }
434
+ ],
435
+ "outputs": [
436
+ {
437
+ "name": "SIGMAS",
438
+ "type": "SIGMAS",
439
+ "links": [
440
+ 27
441
+ ]
442
+ }
443
+ ],
444
+ "properties": {
445
+ "cnr_id": "comfy-core",
446
+ "ver": "0.44.0",
447
+ "Node name for S&R": "BasicScheduler"
448
+ },
449
+ "widgets_values": [
450
+ "sgm_uniform",
451
+ 15,
452
+ 1
453
+ ],
454
+ "widgets_values_named": {
455
+ "scheduler": "sgm_uniform",
456
+ "steps": 15,
457
+ "denoise": 1
458
+ }
459
+ },
460
+ {
461
+ "id": 36,
462
+ "type": "ModelSamplingAuraFlow",
463
+ "pos": [
464
+ 216.72263354772872,
465
+ -300
466
+ ],
467
+ "size": [
468
+ 400,
469
+ 90
470
+ ],
471
+ "flags": {},
472
+ "order": 17,
473
+ "mode": 0,
474
+ "inputs": [
475
+ {
476
+ "name": "model",
477
+ "type": "MODEL",
478
+ "link": 260
479
+ }
480
+ ],
481
+ "outputs": [
482
+ {
483
+ "name": "MODEL",
484
+ "type": "MODEL",
485
+ "links": [
486
+ 69
487
+ ]
488
+ }
489
+ ],
490
+ "properties": {
491
+ "cnr_id": "comfy-core",
492
+ "ver": "0.44.0",
493
+ "Node name for S&R": "ModelSamplingAuraFlow"
494
+ },
495
+ "widgets_values": [
496
+ 3,
497
+ "flow"
498
+ ],
499
+ "widgets_values_named": {
500
+ "shift": 3,
501
+ "sampling": "flow"
502
+ }
503
+ },
504
+ {
505
+ "id": 42,
506
+ "type": "PreviewImage",
507
+ "pos": [
508
+ 2721.7953298463226,
509
+ -238.68997116367268
510
+ ],
511
+ "size": [
512
+ 890,
513
+ 780
514
+ ],
515
+ "flags": {},
516
+ "order": 26,
517
+ "mode": 0,
518
+ "inputs": [
519
+ {
520
+ "name": "images",
521
+ "type": "IMAGE",
522
+ "link": 276
523
+ }
524
+ ],
525
+ "outputs": [
526
+ {
527
+ "name": "images",
528
+ "type": "IMAGE",
529
+ "links": [
530
+ 149,
531
+ 228
532
+ ]
533
+ }
534
+ ],
535
+ "properties": {
536
+ "cnr_id": "comfy-core",
537
+ "ver": "0.44.0",
538
+ "Node name for S&R": "PreviewImage"
539
+ }
540
+ },
541
+ {
542
+ "id": 49,
543
+ "type": "LoraLoaderModelOnly",
544
+ "pos": [
545
+ -423.26391629976365,
546
+ -439.4382881990866
547
+ ],
548
+ "size": [
549
+ 500,
550
+ 120
551
+ ],
552
+ "flags": {},
553
+ "order": 9,
554
+ "mode": 0,
555
+ "inputs": [
556
+ {
557
+ "name": "model",
558
+ "type": "MODEL",
559
+ "link": 91
560
+ }
561
+ ],
562
+ "outputs": [
563
+ {
564
+ "name": "MODEL",
565
+ "type": "MODEL",
566
+ "links": [
567
+ 157
568
+ ]
569
+ }
570
+ ],
571
+ "properties": {
572
+ "cnr_id": "comfy-core",
573
+ "ver": "0.44.0",
574
+ "Node name for S&R": "LoraLoaderModelOnly",
575
+ "models": [
576
+ {
577
+ "name": "moss_image_color_depth_control.safetensors",
578
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
579
+ "directory": "loras"
580
+ }
581
+ ]
582
+ },
583
+ "widgets_values": [
584
+ "moss/moss-tile-control/beta2/epoch1/adapter_model.safetensors",
585
+ 1
586
+ ],
587
+ "widgets_values_named": {
588
+ "lora_name": "moss/moss-tile-control/beta2/epoch1/adapter_model.safetensors",
589
+ "strength_model": 1
590
+ }
591
+ },
592
+ {
593
+ "id": 55,
594
+ "type": "LoadImage",
595
+ "pos": [
596
+ -400,
597
+ 630
598
+ ],
599
+ "size": [
600
+ 290,
601
+ 110
602
+ ],
603
+ "flags": {},
604
+ "order": 3,
605
+ "mode": 0,
606
+ "inputs": [],
607
+ "outputs": [
608
+ {
609
+ "name": "IMAGE",
610
+ "type": "IMAGE",
611
+ "links": [
612
+ 155,
613
+ 248
614
+ ]
615
+ },
616
+ {
617
+ "name": "MASK",
618
+ "type": "MASK",
619
+ "links": null
620
+ }
621
+ ],
622
+ "properties": {
623
+ "cnr_id": "comfy-core",
624
+ "ver": "0.44.0",
625
+ "Node name for S&R": "LoadImage"
626
+ },
627
+ "widgets_values": [
628
+ "3794962993127468234.jpg",
629
+ "image"
630
+ ],
631
+ "widgets_values_named": {
632
+ "image": "3794962993127468234.jpg",
633
+ "upload": "image"
634
+ },
635
+ "color": "#322",
636
+ "bgcolor": "#533"
637
+ },
638
+ {
639
+ "id": 79,
640
+ "type": "PreviewImage",
641
+ "pos": [
642
+ 733.854443722757,
643
+ 622.1325731950062
644
+ ],
645
+ "size": [
646
+ 140,
647
+ 30
648
+ ],
649
+ "flags": {},
650
+ "order": 18,
651
+ "mode": 0,
652
+ "inputs": [
653
+ {
654
+ "name": "images",
655
+ "type": "IMAGE",
656
+ "link": 264
657
+ }
658
+ ],
659
+ "outputs": [
660
+ {
661
+ "name": "images",
662
+ "type": "IMAGE",
663
+ "links": [
664
+ 150,
665
+ 273,
666
+ 275
667
+ ]
668
+ }
669
+ ],
670
+ "properties": {
671
+ "cnr_id": "comfy-core",
672
+ "ver": "0.44.0",
673
+ "Node name for S&R": "PreviewImage"
674
+ }
675
+ },
676
+ {
677
+ "id": 80,
678
+ "type": "ImageCompare",
679
+ "pos": [
680
+ 3646.411023362776,
681
+ -201.8754374257217
682
+ ],
683
+ "size": [
684
+ 930,
685
+ 780
686
+ ],
687
+ "flags": {},
688
+ "order": 27,
689
+ "mode": 0,
690
+ "inputs": [
691
+ {
692
+ "name": "image_a",
693
+ "shape": 7,
694
+ "type": "IMAGE",
695
+ "link": 150
696
+ },
697
+ {
698
+ "name": "image_b",
699
+ "shape": 7,
700
+ "type": "IMAGE",
701
+ "link": 149
702
+ }
703
+ ],
704
+ "outputs": [],
705
+ "properties": {
706
+ "cnr_id": "comfy-core",
707
+ "ver": "0.44.0",
708
+ "Node name for S&R": "ImageCompare"
709
+ },
710
+ "widgets_values": [],
711
+ "widgets_values_named": {}
712
+ },
713
+ {
714
+ "id": 83,
715
+ "type": "MarkdownNote",
716
+ "pos": [
717
+ -1230,
718
+ -520
719
+ ],
720
+ "size": [
721
+ 710,
722
+ 890
723
+ ],
724
+ "flags": {},
725
+ "order": 4,
726
+ "mode": 0,
727
+ "inputs": [],
728
+ "outputs": [],
729
+ "title": "Note: About",
730
+ "properties": {},
731
+ "widgets_values": [
732
+ "## MOSS-Image: Color & Depth Control from a Photo\n\nDrop in a photo and the graph estimates its geometry, then generates a new image that keeps the same pose and framing. No map to prepare by hand.\n\n**How it works**\n\n1. **Load Image** (node 55) - the character reference. It is scaled to 1.0 MP, 16 px aligned (node 77), and drives the output size through **Get Image Size** (node 61) -> **Empty Latent** (node 54).\n2. **Geometry Estimation (MoGe 3)** subgraph - `Load MoGe Model` (`moge_3_vitg_fp16`) -> `Run MoGe Inference` -> `Render MoGe Geometry`. Its output is the `depth_map` of **MOSS-Image Color/Depth Control** (node 53).\n3. **Positive Prompt** (node 5) describes the character and the scene. Geometry comes from the render, everything else - identity, wardrobe, material, light, style - comes from the text.\n4. **MOSS-Image Color/Depth Control** (node 53) injects the map into the image tokens; the LoRA loaded just before it (`moss_image_color_depth_control.safetensors`, strength 1.0) is what makes the model read it. Without the LoRA the node is a no-op.\n5. **Save / Compare**: `VAE Decode` -> preview, plus **Compare Images** (node 80) holding the generated frame next to the rendered map.\n\n**Settings**\n\n`strength` 1.0 is the trained strength; `start_percent` / `end_percent` pick the part of the run that is controlled (0 to 0.6 here: the layout is fixed, the last steps render freely). Sampling follows the MOSS-Image control test graph: 25 steps, CFG 3, `er_sde` / `simple`, shift 5, skip layer guidance on layer 24, temporal score rescaling 0.95.\n\n**Useful knobs**\n\n- `Render MoGe Geometry` in the subgraph is set to `mask`: only the silhouette is used. Switch it to `depth` (`normalization = min_max`) when you want the face profile, shoulders and background falloff to be controlled too - the pose gets tighter, the prompt has less room.\n- `end_percent` 1.0 pins geometry to the last step; 0.5 frees the surface and lighting. For a mask-derived map 0.6 is a good middle.\n- The reference has **pointed elf ears**, and the geometry keeps them. Either keep that in the prompt (`pointed elf ears`) or hide them (`ears covered by hair`).\n- **Painter** (node 81) is in the graph for painting a map by hand: white near / black far for depth, one flat colour per region for colour. Connect its `IMAGE` to `depth_map` or `color_map` to override the estimate.\n- `color_map` is optional and currently unconnected. One flat colour per region, colours only act as region IDs, black means \"unspecified\"; it must match the output size (1024 x 1024 here).\n"
733
+ ],
734
+ "widgets_values_named": {
735
+ "text": "## MOSS-Image: Color & Depth Control from a Photo\n\nDrop in a photo and the graph estimates its geometry, then generates a new image that keeps the same pose and framing. No map to prepare by hand.\n\n**How it works**\n\n1. **Load Image** (node 55) - the character reference. It is scaled to 1.0 MP, 16 px aligned (node 77), and drives the output size through **Get Image Size** (node 61) -> **Empty Latent** (node 54).\n2. **Geometry Estimation (MoGe 3)** subgraph - `Load MoGe Model` (`moge_3_vitg_fp16`) -> `Run MoGe Inference` -> `Render MoGe Geometry`. Its output is the `depth_map` of **MOSS-Image Color/Depth Control** (node 53).\n3. **Positive Prompt** (node 5) describes the character and the scene. Geometry comes from the render, everything else - identity, wardrobe, material, light, style - comes from the text.\n4. **MOSS-Image Color/Depth Control** (node 53) injects the map into the image tokens; the LoRA loaded just before it (`moss_image_color_depth_control.safetensors`, strength 1.0) is what makes the model read it. Without the LoRA the node is a no-op.\n5. **Save / Compare**: `VAE Decode` -> preview, plus **Compare Images** (node 80) holding the generated frame next to the rendered map.\n\n**Settings**\n\n`strength` 1.0 is the trained strength; `start_percent` / `end_percent` pick the part of the run that is controlled (0 to 0.6 here: the layout is fixed, the last steps render freely). Sampling follows the MOSS-Image control test graph: 25 steps, CFG 3, `er_sde` / `simple`, shift 5, skip layer guidance on layer 24, temporal score rescaling 0.95.\n\n**Useful knobs**\n\n- `Render MoGe Geometry` in the subgraph is set to `mask`: only the silhouette is used. Switch it to `depth` (`normalization = min_max`) when you want the face profile, shoulders and background falloff to be controlled too - the pose gets tighter, the prompt has less room.\n- `end_percent` 1.0 pins geometry to the last step; 0.5 frees the surface and lighting. For a mask-derived map 0.6 is a good middle.\n- The reference has **pointed elf ears**, and the geometry keeps them. Either keep that in the prompt (`pointed elf ears`) or hide them (`ears covered by hair`).\n- **Painter** (node 81) is in the graph for painting a map by hand: white near / black far for depth, one flat colour per region for colour. Connect its `IMAGE` to `depth_map` or `color_map` to override the estimate.\n- `color_map` is optional and currently unconnected. One flat colour per region, colours only act as region IDs, black means \"unspecified\"; it must match the output size (1024 x 1024 here).\n"
736
+ },
737
+ "color": "#222",
738
+ "bgcolor": "#000"
739
+ },
740
+ {
741
+ "id": 84,
742
+ "type": "MarkdownNote",
743
+ "pos": [
744
+ -1970,
745
+ -520
746
+ ],
747
+ "size": [
748
+ 520,
749
+ 1250
750
+ ],
751
+ "flags": {},
752
+ "order": 5,
753
+ "mode": 0,
754
+ "inputs": [],
755
+ "outputs": [],
756
+ "title": "Note: Model Links",
757
+ "properties": {},
758
+ "widgets_values": [
759
+ "## Model Links\n\n- [Hugging Face: Comfy-Org/MOSS-Image](https://huggingface.co/Comfy-Org/MOSS-Image)\n- [Hugging Face: Comfy-Org/MoGe](https://huggingface.co/Comfy-Org/MoGe) (geometry estimation)\n- [Upstream control adapter: darask0/MOSS-Image-color-depth-control](https://huggingface.co/darask0/MOSS-Image-color-depth-control)\n\n**diffusion_models**\n\n- [moss_image_sft_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors) - 5.41 GB, the checkpoint this graph loads\n- [moss_image_sft_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_bf16.safetensors) - 10.16 GB, higher precision, slower\n- [moss_image_base_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_base_int8_convrot.safetensors) - 5.41 GB, the base checkpoint runs in the same graph\n\n**text_encoders**\n\n- [qwen3.5_9b_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors) - 9.75 GB, the encoder this graph loads\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) - 7.34 GB, drop-in swap, saves 2.4 GB\n- [qwen3.5_9b_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_bf16.safetensors) - 19.31 GB, full precision\n\n**vae**\n\n- [flux2-vae_small.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors) - 249 MB, the one this graph loads\n- [flux2-vae.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae.safetensors) - 336 MB, full size, pixel-level differences only\n\n**loras**\n\n- [moss_image_color_depth_control.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors) - 313 MB, `LoraLoaderModelOnly` at strength 1.0\n\n**geometry_estimation**\n\n- [moge_3_vitg_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitg_fp16.safetensors) - 2.50 GB, the depth estimator used by the subgraph\n- [moge_3_vitl_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitl_fp16.safetensors) - 741 MB, the lighter ViT-L variant\n\n**Disk footprint**: the graph as shipped needs **18.23 GB** (18,226,010,396 bytes) (5.41 + 9.75 + 0.25 + 0.31 + 2.50). Swapping the text encoder to `w4a8_gptq` brings it to **15.82 GB**; a bf16 checkpoint and encoder instead push it past 32 GB.\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 diffusion_models/\n│ │ └── moss_image_sft_int8_convrot.safetensors 5.41 GB\n│ ├── 📂 text_encoders/\n│ │ └── qwen3.5_9b_int8_convrot.safetensors 9.75 GB\n│ ├── 📂 vae/\n│ │ └── flux2-vae_small.safetensors 249 MB\n│ ├── 📂 loras/\n│ │ └── moss_image_color_depth_control.safetensors 313 MB\n│ └── 📂 geometry_estimation/\n│ └── moge_3_vitg_fp16.safetensors 2.50 GB\n```\n\nCore support (`MOSSImageControl`, MoGe nodes) ships from ComfyUI **0.43.0**.\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
760
+ ],
761
+ "widgets_values_named": {
762
+ "text": "## Model Links\n\n- [Hugging Face: Comfy-Org/MOSS-Image](https://huggingface.co/Comfy-Org/MOSS-Image)\n- [Hugging Face: Comfy-Org/MoGe](https://huggingface.co/Comfy-Org/MoGe) (geometry estimation)\n- [Upstream control adapter: darask0/MOSS-Image-color-depth-control](https://huggingface.co/darask0/MOSS-Image-color-depth-control)\n\n**diffusion_models**\n\n- [moss_image_sft_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors) - 5.41 GB, the checkpoint this graph loads\n- [moss_image_sft_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_bf16.safetensors) - 10.16 GB, higher precision, slower\n- [moss_image_base_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_base_int8_convrot.safetensors) - 5.41 GB, the base checkpoint runs in the same graph\n\n**text_encoders**\n\n- [qwen3.5_9b_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors) - 9.75 GB, the encoder this graph loads\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) - 7.34 GB, drop-in swap, saves 2.4 GB\n- [qwen3.5_9b_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_bf16.safetensors) - 19.31 GB, full precision\n\n**vae**\n\n- [flux2-vae_small.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors) - 249 MB, the one this graph loads\n- [flux2-vae.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae.safetensors) - 336 MB, full size, pixel-level differences only\n\n**loras**\n\n- [moss_image_color_depth_control.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors) - 313 MB, `LoraLoaderModelOnly` at strength 1.0\n\n**geometry_estimation**\n\n- [moge_3_vitg_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitg_fp16.safetensors) - 2.50 GB, the depth estimator used by the subgraph\n- [moge_3_vitl_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitl_fp16.safetensors) - 741 MB, the lighter ViT-L variant\n\n**Disk footprint**: the graph as shipped needs **18.23 GB** (18,226,010,396 bytes) (5.41 + 9.75 + 0.25 + 0.31 + 2.50). Swapping the text encoder to `w4a8_gptq` brings it to **15.82 GB**; a bf16 checkpoint and encoder instead push it past 32 GB.\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 diffusion_models/\n│ │ └── moss_image_sft_int8_convrot.safetensors 5.41 GB\n│ ├── 📂 text_encoders/\n│ │ └── qwen3.5_9b_int8_convrot.safetensors 9.75 GB\n│ ├── 📂 vae/\n│ │ └── flux2-vae_small.safetensors 249 MB\n│ ├── 📂 loras/\n│ │ └── moss_image_color_depth_control.safetensors 313 MB\n│ └── 📂 geometry_estimation/\n│ └── moge_3_vitg_fp16.safetensors 2.50 GB\n```\n\nCore support (`MOSSImageControl`, MoGe nodes) ships from ComfyUI **0.43.0**.\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
763
+ },
764
+ "color": "#222",
765
+ "bgcolor": "#000"
766
+ },
767
+ {
768
+ "id": 85,
769
+ "type": "SaveImage",
770
+ "pos": [
771
+ 3041.0621042800776,
772
+ 678.5850493672125
773
+ ],
774
+ "size": [
775
+ 420,
776
+ 470
777
+ ],
778
+ "flags": {},
779
+ "order": 28,
780
+ "mode": 0,
781
+ "inputs": [
782
+ {
783
+ "name": "images",
784
+ "type": "IMAGE",
785
+ "link": 228
786
+ }
787
+ ],
788
+ "outputs": [
789
+ {
790
+ "name": "images",
791
+ "type": "IMAGE",
792
+ "links": null
793
+ }
794
+ ],
795
+ "properties": {
796
+ "cnr_id": "comfy-core",
797
+ "ver": "0.44.0",
798
+ "Node name for S&R": "SaveImage"
799
+ },
800
+ "widgets_values": [
801
+ "image_moss_image_color_depth_control"
802
+ ],
803
+ "widgets_values_named": {
804
+ "filename_prefix": "image_moss_image_color_depth_control"
805
+ }
806
+ },
807
+ {
808
+ "id": 86,
809
+ "type": "KSamplerSelect",
810
+ "pos": [
811
+ 1170.4423572560272,
812
+ -260.582916386391
813
+ ],
814
+ "size": [
815
+ 270,
816
+ 58
817
+ ],
818
+ "flags": {},
819
+ "order": 6,
820
+ "mode": 0,
821
+ "inputs": [],
822
+ "outputs": [
823
+ {
824
+ "name": "SAMPLER",
825
+ "type": "SAMPLER",
826
+ "links": [
827
+ 154
828
+ ]
829
+ }
830
+ ],
831
+ "properties": {
832
+ "cnr_id": "comfy-core",
833
+ "ver": "0.44.0",
834
+ "Node name for S&R": "KSamplerSelect"
835
+ },
836
+ "widgets_values": [
837
+ "euler"
838
+ ],
839
+ "widgets_values_named": {
840
+ "sampler_name": "euler"
841
+ }
842
+ },
843
+ {
844
+ "id": 87,
845
+ "type": "ImageScaleBy",
846
+ "pos": [
847
+ -7.178545924871845,
848
+ 681.7385897176745
849
+ ],
850
+ "size": [
851
+ 270,
852
+ 82
853
+ ],
854
+ "flags": {},
855
+ "order": 12,
856
+ "mode": 4,
857
+ "inputs": [
858
+ {
859
+ "name": "image",
860
+ "type": "IMAGE",
861
+ "link": 155
862
+ }
863
+ ],
864
+ "outputs": [
865
+ {
866
+ "name": "IMAGE",
867
+ "type": "IMAGE",
868
+ "links": [
869
+ 161
870
+ ]
871
+ }
872
+ ],
873
+ "properties": {
874
+ "cnr_id": "comfy-core",
875
+ "ver": "0.44.0",
876
+ "Node name for S&R": "ImageScaleBy"
877
+ },
878
+ "widgets_values": [
879
+ "lanczos",
880
+ 0.5
881
+ ],
882
+ "widgets_values_named": {
883
+ "upscale_method": "lanczos",
884
+ "scale_by": 0.5
885
+ }
886
+ },
887
+ {
888
+ "id": 88,
889
+ "type": "LoraLoaderModelOnly",
890
+ "pos": [
891
+ -447.13559989481814,
892
+ -263.5408892596166
893
+ ],
894
+ "size": [
895
+ 500,
896
+ 120
897
+ ],
898
+ "flags": {},
899
+ "order": 14,
900
+ "mode": 0,
901
+ "inputs": [
902
+ {
903
+ "name": "model",
904
+ "type": "MODEL",
905
+ "link": 157
906
+ }
907
+ ],
908
+ "outputs": [
909
+ {
910
+ "name": "MODEL",
911
+ "type": "MODEL",
912
+ "links": [
913
+ 260
914
+ ]
915
+ }
916
+ ],
917
+ "properties": {
918
+ "cnr_id": "comfy-core",
919
+ "ver": "0.44.0",
920
+ "Node name for S&R": "LoraLoaderModelOnly",
921
+ "models": [
922
+ {
923
+ "name": "moss_image_color_depth_control.safetensors",
924
+ "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors",
925
+ "directory": "loras"
926
+ }
927
+ ]
928
+ },
929
+ "widgets_values": [
930
+ "moss/download/Coffmeteor-sft-15step-cfg3.safetensors",
931
+ 1
932
+ ],
933
+ "widgets_values_named": {
934
+ "lora_name": "moss/download/Coffmeteor-sft-15step-cfg3.safetensors",
935
+ "strength_model": 1
936
+ }
937
+ },
938
+ {
939
+ "id": 90,
940
+ "type": "ImageScaleBy",
941
+ "pos": [
942
+ 321.18555329294736,
943
+ 683.4828563588695
944
+ ],
945
+ "size": [
946
+ 270,
947
+ 82
948
+ ],
949
+ "flags": {},
950
+ "order": 15,
951
+ "mode": 0,
952
+ "inputs": [
953
+ {
954
+ "name": "image",
955
+ "type": "IMAGE",
956
+ "link": 161
957
+ }
958
+ ],
959
+ "outputs": [
960
+ {
961
+ "name": "IMAGE",
962
+ "type": "IMAGE",
963
+ "links": [
964
+ 264,
965
+ 265
966
+ ]
967
+ }
968
+ ],
969
+ "properties": {
970
+ "cnr_id": "comfy-core",
971
+ "ver": "0.44.0",
972
+ "Node name for S&R": "ImageScaleBy"
973
+ },
974
+ "widgets_values": [
975
+ "lanczos",
976
+ 4
977
+ ],
978
+ "widgets_values_named": {
979
+ "upscale_method": "lanczos",
980
+ "scale_by": 4
981
+ }
982
+ },
983
+ {
984
+ "id": 91,
985
+ "type": "PrimitiveStringMultiline",
986
+ "pos": [
987
+ 375.1526388094519,
988
+ -2150.2072456067854
989
+ ],
990
+ "size": [
991
+ 470,
992
+ 270
993
+ ],
994
+ "flags": {
995
+ "collapsed": false
996
+ },
997
+ "order": 7,
998
+ "mode": 4,
999
+ "inputs": [],
1000
+ "outputs": [
1001
+ {
1002
+ "name": "STRING",
1003
+ "type": "STRING",
1004
+ "links": [
1005
+ 163
1006
+ ]
1007
+ }
1008
+ ],
1009
+ "title": "Text (System prompt)",
1010
+ "properties": {
1011
+ "cnr_id": "comfy-core",
1012
+ "ver": "0.42.0",
1013
+ "Node name for S&R": "PrimitiveStringMultiline"
1014
+ },
1015
+ "widgets_values": [
1016
+ "\n你是一名顶级视觉逆向分析专家、数字绘画研究员和 AI 图像生成提示词工程师。\n\n你的唯一任务是:对用户提供的参考图像进行极其细致的视觉分析,并将所有具有复现价值的视觉特征整合为一段高精度、可直接用于 AI 图像生成模型的英文提示词。\n\n【核心要求】\n\n你必须在内部完成充分、深入的图像分析,但绝对不能向用户展示分析过程、分析结果、推理步骤、分类标题或任何解释。\n\n最终回答只能包含英文 Positive Prompt 本身。\n\n禁止输出:\n- 任何中文内容\n- 开场白或结束语\n- 标题、编号、分类标签\n- Markdown 格式或代码块\n- 分析过程和解释\n- Negative Prompt\n- 参数建议\n- 额外说明\n- 与提示词无关的任何文字\n\n【分析深度】\n\n在内部对图像进行像素级视觉特征观察,尽可能细致地识别:\n\n1. 整体艺术风格、媒介特征、风格融合方式、二维或三维表现倾向。\n\n2. 线稿系统:外轮廓粗细、内部线条粗细、线宽变化、线条收尖、线条转折、曲率、闭合程度、线条颜色、线条疏密、边缘平滑度。\n\n3. 形状语言:几何概括方式、曲线与直线比例、轮廓简化程度、局部夸张、形状节奏、视觉体积的建立方式。\n\n4. 上色方法:平涂、赛璐璐、厚涂、渐变、色块分层、色彩边界、硬边与软边过渡、局部混合渲染。\n\n5. 阴影结构:阴影边界锐利度、阴影形状、阴影色相、明暗层级、结构性阴影、接触阴影、装饰性阴影、阴影与形体的关系。\n\n6. 高光系统:高光形状、大小、分布、方向、边缘硬度、色彩、亮度、反射效果、块状高光、条状高光、局部辉光。\n\n7. 配色逻辑:主色调、辅助色、强调色、色彩饱和度、明度分布、冷暖对比、亮暗部色相偏移、色彩统一性。\n\n8. 光照处理:主光方向、环境光、轮廓光、光照对比度、局部泛光、光影风格化程度。\n\n9. 渲染质感:数字绘画感、笔触可见度、纹理密度、边缘锐度、画面平滑度、颗粒感、印刷感、类似矢量图的视觉质感。\n\n10. 细节组织:视觉焦点、细节密度、局部简化、主次关系、大中小形状的组合、画面精致程度。\n\n11. 构图与画面语言:景别、视角、透视、裁切、空间层次、主体位置、视觉动线。\n\n12. 画面内容:主体外观、服装、表情、姿态、道具、背景和场景元素。\n\n【反推规则】\n\n1. 默认采用纯画风反推模式:只提取能够迁移到其他主体的绘画风格,不描述具体人物、体型、性别、肌肉、服装、动作或场景。\n\n2. 如果用户明确要求完整反推,则同时还原主体、场景、构图和绘画风格。\n\n3. 如果提供多张参考图,提取所有图像共同、稳定且具有辨识度的视觉特征,不把单张图片独有的元素误认为共同画风。\n\n4. 不允许只用 anime style、beautiful illustration、detailed shading、high quality 等笼统词汇代替具体的视觉描述。\n\n5. 优先使用能够控制画面视觉效果的具体英文短语,准确描述线条、阴影、高光、色块、色彩、边缘和渲染方式。\n\n6. 区分硬边阴影与柔和渐变、结构性高光与装饰性高光、真实材质质感与绘画渲染质感。\n\n7. 不得凭空猜测原作者、绘画软件、具体模型、LoRA 或原始提示词。\n\n8. 仅描述有视觉证据支持的特征,不虚构不可见的细节。\n\n9. 将最有辨识度、最影响生成结果的特征放在提示词前部。\n\n10. 避免重复堆砌近义词,避免相互矛盾的描述,避免无意义的质量标签。\n\n11. 允许详细、丰富、具体的提示词,但每个短语都必须具有实际的视觉控制价值。\n\n12. 将所有分析结果重新组织为流畅、准确、信息密度高的英文生图提示词,而不是逐条罗列分析项目。\n\n13. 优先使用适合 MOSS-Image 等自然语言生图模型的英文描述。若用户明确指定 Anima 或 Danbooru 标签格式,则转换为适合该模型的英文标签式提示词。\n\n14. 对无法确认的视觉细节不进行猜测,不为了增加长度而编造特征。\n\n【最终输出规范——最高优先级】\n\n最终输出必须严格遵守以下要求:\n\n只输出一段英文 Positive Prompt。\n\n不允许出现任何标题。\n不允许出现任何分析内容。\n不允许解释为什么选择这些词。\n不允许输出 Negative Prompt。\n不允许���加任何建议。\n不允许添加引号或代码块。\n不允许输出第二个版本。\n\n所有视觉分析仅在内部完成,最终仅保留经过筛选、整合和优化的高精度提示词。\n\n无论分析多么复杂,输出形式都不能改变。\n\n用户上传图像后,直接输出最终英文提示词,不要进行任何说明。\n"
1017
+ ],
1018
+ "widgets_values_named": {
1019
+ "value": "\n你是一名顶级视觉逆向分析专家、数字绘画研究员和 AI 图像生成提示词工程师。\n\n你的唯一任务是:对用户提供的参考图像进行极其细致的视觉分析,并将所有具有复现价值的视觉特征整合为一段高精度、可直接用于 AI 图像生成模型的英文提示词。\n\n【核心要求】\n\n你必须在内部完成充分、深入的图像分析,但绝对不能向用户展示分析过程、分析结果、推理步骤、分类标题或任何解释。\n\n最终回答只能包含英文 Positive Prompt 本身。\n\n禁止输出:\n- 任何中文内容\n- 开场白或结束语\n- 标题、编号、分类标签\n- Markdown 格式或代码块\n- 分析过程和解释\n- Negative Prompt\n- 参数建议\n- 额外说明\n- 与提示词无关的任何文字\n\n【分析深度】\n\n在内部对图像进行像素级视觉特征观察,尽可能细致地识别:\n\n1. 整体艺术风格、媒介特征、风格融合方式、二维或三维表现倾向。\n\n2. 线稿系统:外轮廓粗细、内部线条粗细、线宽变化、线条收尖、线条转折、曲率、闭合程度、线条颜色、线条疏密、边缘平滑度。\n\n3. 形状语言:几何概括方式、曲线与直线比例、轮廓简化程度、局部夸张、形状节奏、视觉体积的建立方式。\n\n4. 上色方法:平涂、赛璐璐、厚涂、渐变、色块分层、色彩边界、硬边与软边过渡、局部混合渲染。\n\n5. 阴影结构:阴影边界锐利度、阴影形状、阴影色相、明暗层级、结构性阴影、接触阴影、装饰性阴影、阴影与形体的关系。\n\n6. 高光系统:高光形状、大小、分布、方向、边缘硬度、色彩、亮度、反射效果、块状高光、条状高光、局部辉光。\n\n7. 配色逻辑:主色调、辅助色、强调色、色彩饱和度、明度分布、冷暖对比、亮暗部色相偏移、色彩统一性。\n\n8. 光照处理:主光方向、环境光、轮廓光、光照对比度、局部泛光、光影风格化程度。\n\n9. 渲染质感:数字绘画感、笔触可见度、纹理密度、边缘锐度、画面平滑度、颗粒感、印刷感、类似矢量图的视觉质感。\n\n10. 细节组织:视觉焦点、细节密度、局部简化、主次关系、大中小形状的组合、画面精致程度。\n\n11. 构图与画面语言:景别、视角、透视、裁切、空间层次、主体位置、视觉动线。\n\n12. 画面内容:主体外观、服装、表情、姿态、道具、背景和场景元素。\n\n【反推规则】\n\n1. 默认采用纯画风反推模式:只提取能够迁移到其他主体的绘画风格,不描述具体人物、体型、性别、肌肉、服装、动作或场景。\n\n2. 如果用户明确要求完整反推,则同时还原主体、场景、构图和绘画风格。\n\n3. 如果提供多张参考图,提取所有图像共同、稳定且具有辨识度的视觉特征,不把单张图片独有的元素误认为共同画风。\n\n4. 不允许只用 anime style、beautiful illustration、detailed shading、high quality 等笼统词汇代替具体的视觉描述。\n\n5. 优先使用能够控制画面视觉效果的具体英文短语,准确描述线条、阴影、高光、色块、色彩、边缘和渲染方式。\n\n6. 区分硬边阴影与柔和渐变、结构性高光与装饰性高光、真实材质质感与绘画渲染质感。\n\n7. 不得凭空猜测原作者、绘画软件、具体模型、LoRA 或原始提示词。\n\n8. 仅描述有视觉证据支持的特征,不虚构不可见的细节。\n\n9. 将最有辨识度、最影响生成结果的特征放在提示词前部。\n\n10. 避免重复堆砌近义词,避免相互矛盾的描述,避免无意义的质量标签。\n\n11. 允许详细、丰富、具体的提示词,但每个短语都必须具有实际的视觉控制价值。\n\n12. 将所有分析结果重新组织为流畅、准确、信息密度高的英文生图提示词,而不是逐条罗列分析项目。\n\n13. 优先使用适合 MOSS-Image 等自然语言生图模型的英文描述。若用户明确指定 Anima 或 Danbooru 标签格式,则转换为适合该模型的英文标签式提示词。\n\n14. 对无法确认的视觉细节不进行猜测,不为了增加长度而编造特征。\n\n【最终输出规范——最高优先级】\n\n最终输出必须严格遵守以下要求:\n\n只输出一段英文 Positive Prompt。\n\n不允许出现任何标题。\n不允许出现任何分析内容。\n不允许解释为什么选择这些词。\n不允许输出 Negative Prompt。\n不允许附加任何建议。\n不允许添加引号或代码块。\n不允许输出第二个版本。\n\n所有视觉分析仅在内部完成,最终仅保留经过筛选、整合和优化的高精度提示词。\n\n无论分析多么复杂,输出形式都不能改变。\n\n用户上传图像后,直接输出最终英文提示词,不要进行任何说明。\n"
1020
+ }
1021
+ },
1022
+ {
1023
+ "id": 92,
1024
+ "type": "TextGenerate",
1025
+ "pos": [
1026
+ 1171.0822008806106,
1027
+ -2365.5035628809155
1028
+ ],
1029
+ "size": [
1030
+ 400,
1031
+ 550
1032
+ ],
1033
+ "flags": {},
1034
+ "order": 13,
1035
+ "mode": 4,
1036
+ "inputs": [
1037
+ {
1038
+ "name": "clip",
1039
+ "type": "CLIP",
1040
+ "link": 165
1041
+ },
1042
+ {
1043
+ "name": "image",
1044
+ "shape": 7,
1045
+ "type": "IMAGE",
1046
+ "link": 248
1047
+ },
1048
+ {
1049
+ "name": "video",
1050
+ "shape": 7,
1051
+ "type": "IMAGE",
1052
+ "link": null
1053
+ },
1054
+ {
1055
+ "name": "audio",
1056
+ "shape": 7,
1057
+ "type": "AUDIO",
1058
+ "link": null
1059
+ },
1060
+ {
1061
+ "name": "system_prompt",
1062
+ "shape": 7,
1063
+ "type": "STRING",
1064
+ "link": 163
1065
+ }
1066
+ ],
1067
+ "outputs": [
1068
+ {
1069
+ "name": "generated_text",
1070
+ "type": "STRING",
1071
+ "links": [
1072
+ 166
1073
+ ]
1074
+ },
1075
+ {
1076
+ "name": "thinking",
1077
+ "type": "STRING",
1078
+ "links": []
1079
+ }
1080
+ ],
1081
+ "properties": {
1082
+ "cnr_id": "comfy-core",
1083
+ "ver": "0.42.0",
1084
+ "Node name for S&R": "TextGenerate"
1085
+ },
1086
+ "widgets_values": [
1087
+ "",
1088
+ 4096,
1089
+ "on",
1090
+ 0.7,
1091
+ 64,
1092
+ 0.95,
1093
+ 0.05,
1094
+ 1,
1095
+ 0,
1096
+ 0,
1097
+ true,
1098
+ true,
1099
+ "auto"
1100
+ ],
1101
+ "widgets_values_named": {
1102
+ "prompt": "",
1103
+ "max_length": 4096,
1104
+ "sampling_mode": "on",
1105
+ "sampling_mode.temperature": 0.7,
1106
+ "sampling_mode.top_k": 64,
1107
+ "sampling_mode.top_p": 0.95,
1108
+ "sampling_mode.min_p": 0.05,
1109
+ "sampling_mode.repetition_penalty": 1,
1110
+ "sampling_mode.seed": 0,
1111
+ "sampling_mode.presence_penalty": 0,
1112
+ "thinking": true,
1113
+ "use_default_template": true,
1114
+ "mtp": "auto"
1115
+ }
1116
+ },
1117
+ {
1118
+ "id": 93,
1119
+ "type": "CLIPLoader",
1120
+ "pos": [
1121
+ 468.0082400074318,
1122
+ -2370.6899919511707
1123
+ ],
1124
+ "size": [
1125
+ 270,
1126
+ 106
1127
+ ],
1128
+ "flags": {},
1129
+ "order": 8,
1130
+ "mode": 4,
1131
+ "inputs": [],
1132
+ "outputs": [
1133
+ {
1134
+ "name": "CLIP",
1135
+ "type": "CLIP",
1136
+ "links": [
1137
+ 165
1138
+ ]
1139
+ }
1140
+ ],
1141
+ "properties": {
1142
+ "cnr_id": "comfy-core",
1143
+ "ver": "0.44.0",
1144
+ "Node name for S&R": "CLIPLoader"
1145
+ },
1146
+ "widgets_values": [
1147
+ "qwen3.5-9b-nsfw-captioning-v5.safetensors",
1148
+ "moss_image",
1149
+ "default"
1150
+ ],
1151
+ "widgets_values_named": {
1152
+ "clip_name": "qwen3.5-9b-nsfw-captioning-v5.safetensors",
1153
+ "type": "moss_image",
1154
+ "device": "default"
1155
+ }
1156
+ },
1157
+ {
1158
+ "id": 94,
1159
+ "type": "PreviewAny",
1160
+ "pos": [
1161
+ 1713.012250094057,
1162
+ -2364.526445608352
1163
+ ],
1164
+ "size": [
1165
+ 470.96127217369775,
1166
+ 425.59604314573403
1167
+ ],
1168
+ "flags": {},
1169
+ "order": 16,
1170
+ "mode": 4,
1171
+ "inputs": [
1172
+ {
1173
+ "name": "source",
1174
+ "type": "*",
1175
+ "link": 166
1176
+ }
1177
+ ],
1178
+ "outputs": [
1179
+ {
1180
+ "name": "STRING",
1181
+ "type": "STRING",
1182
+ "links": []
1183
+ }
1184
+ ],
1185
+ "properties": {
1186
+ "cnr_id": "comfy-core",
1187
+ "ver": "0.44.0",
1188
+ "Node name for S&R": "PreviewAny"
1189
+ },
1190
+ "widgets_values": [],
1191
+ "widgets_values_named": {}
1192
+ },
1193
+ {
1194
+ "id": 124,
1195
+ "type": "VAEEncode",
1196
+ "pos": [
1197
+ 1149.2279511667382,
1198
+ 242.2625195138657
1199
+ ],
1200
+ "size": [
1201
+ 140,
1202
+ 46
1203
+ ],
1204
+ "flags": {},
1205
+ "order": 19,
1206
+ "mode": 0,
1207
+ "inputs": [
1208
+ {
1209
+ "name": "pixels",
1210
+ "type": "IMAGE",
1211
+ "link": 265
1212
+ },
1213
+ {
1214
+ "name": "vae",
1215
+ "type": "VAE",
1216
+ "link": 246
1217
+ }
1218
+ ],
1219
+ "outputs": [
1220
+ {
1221
+ "name": "LATENT",
1222
+ "type": "LATENT",
1223
+ "links": [
1224
+ 247
1225
+ ]
1226
+ }
1227
+ ],
1228
+ "properties": {
1229
+ "cnr_id": "comfy-core",
1230
+ "ver": "0.44.0",
1231
+ "Node name for S&R": "VAEEncode"
1232
+ }
1233
+ },
1234
+ {
1235
+ "id": 129,
1236
+ "type": "MOSSImageControl",
1237
+ "pos": [
1238
+ 768.8538371422002,
1239
+ 39.715516256027
1240
+ ],
1241
+ "size": [
1242
+ 281.27109375,
1243
+ 186
1244
+ ],
1245
+ "flags": {},
1246
+ "order": 21,
1247
+ "mode": 0,
1248
+ "inputs": [
1249
+ {
1250
+ "name": "positive",
1251
+ "type": "CONDITIONING",
1252
+ "link": 266
1253
+ },
1254
+ {
1255
+ "name": "negative",
1256
+ "type": "CONDITIONING",
1257
+ "link": 267
1258
+ },
1259
+ {
1260
+ "name": "vae",
1261
+ "type": "VAE",
1262
+ "link": 272
1263
+ },
1264
+ {
1265
+ "name": "color_map",
1266
+ "shape": 7,
1267
+ "type": "IMAGE",
1268
+ "link": 273
1269
+ },
1270
+ {
1271
+ "name": "depth_map",
1272
+ "shape": 7,
1273
+ "type": "IMAGE",
1274
+ "link": null
1275
+ }
1276
+ ],
1277
+ "outputs": [
1278
+ {
1279
+ "name": "positive",
1280
+ "type": "CONDITIONING",
1281
+ "links": [
1282
+ 268
1283
+ ]
1284
+ },
1285
+ {
1286
+ "name": "negative",
1287
+ "type": "CONDITIONING",
1288
+ "links": [
1289
+ 269
1290
+ ]
1291
+ }
1292
+ ],
1293
+ "properties": {
1294
+ "cnr_id": "comfy-core",
1295
+ "ver": "0.44.0",
1296
+ "Node name for S&R": "MOSSImageControl"
1297
+ },
1298
+ "widgets_values": [
1299
+ 1,
1300
+ 0,
1301
+ 1
1302
+ ],
1303
+ "widgets_values_named": {
1304
+ "strength": 1,
1305
+ "start_percent": 0,
1306
+ "end_percent": 1
1307
+ }
1308
+ },
1309
+ {
1310
+ "id": 130,
1311
+ "type": "ColorMatchV2",
1312
+ "pos": [
1313
+ 2068.637924376414,
1314
+ -120.42971148724214
1315
+ ],
1316
+ "size": [
1317
+ 270,
1318
+ 126
1319
+ ],
1320
+ "flags": {},
1321
+ "order": 25,
1322
+ "mode": 0,
1323
+ "inputs": [
1324
+ {
1325
+ "name": "image_target",
1326
+ "type": "IMAGE",
1327
+ "link": 274
1328
+ },
1329
+ {
1330
+ "name": "image_ref",
1331
+ "type": "IMAGE",
1332
+ "link": 275
1333
+ }
1334
+ ],
1335
+ "outputs": [
1336
+ {
1337
+ "name": "image",
1338
+ "type": "IMAGE",
1339
+ "links": [
1340
+ 276
1341
+ ]
1342
+ }
1343
+ ],
1344
+ "properties": {
1345
+ "cnr_id": "comfyui-kjnodes",
1346
+ "ver": "37ca7aff7ed14dde6ba3c0b25713e21c18b28dcb",
1347
+ "Node name for S&R": "ColorMatchV2"
1348
+ },
1349
+ "widgets_values": [
1350
+ "mkl",
1351
+ 1,
1352
+ true
1353
+ ],
1354
+ "widgets_values_named": {
1355
+ "method": "mkl",
1356
+ "strength": 1,
1357
+ "multithread": true
1358
+ }
1359
+ }
1360
+ ],
1361
+ "links": [
1362
+ [
1363
+ 3,
1364
+ 3,
1365
+ 0,
1366
+ 5,
1367
+ 0,
1368
+ "CLIP"
1369
+ ],
1370
+ [
1371
+ 4,
1372
+ 3,
1373
+ 0,
1374
+ 6,
1375
+ 0,
1376
+ "CLIP"
1377
+ ],
1378
+ [
1379
+ 9,
1380
+ 4,
1381
+ 0,
1382
+ 9,
1383
+ 1,
1384
+ "VAE"
1385
+ ],
1386
+ [
1387
+ 25,
1388
+ 20,
1389
+ 0,
1390
+ 9,
1391
+ 0,
1392
+ "LATENT"
1393
+ ],
1394
+ [
1395
+ 27,
1396
+ 22,
1397
+ 0,
1398
+ 20,
1399
+ 4,
1400
+ "SIGMAS"
1401
+ ],
1402
+ [
1403
+ 69,
1404
+ 36,
1405
+ 0,
1406
+ 2,
1407
+ 0,
1408
+ "MODEL"
1409
+ ],
1410
+ [
1411
+ 91,
1412
+ 1,
1413
+ 0,
1414
+ 49,
1415
+ 0,
1416
+ "MODEL"
1417
+ ],
1418
+ [
1419
+ 149,
1420
+ 42,
1421
+ 0,
1422
+ 80,
1423
+ 1,
1424
+ "IMAGE"
1425
+ ],
1426
+ [
1427
+ 150,
1428
+ 79,
1429
+ 0,
1430
+ 80,
1431
+ 0,
1432
+ "IMAGE"
1433
+ ],
1434
+ [
1435
+ 154,
1436
+ 86,
1437
+ 0,
1438
+ 20,
1439
+ 3,
1440
+ "SAMPLER"
1441
+ ],
1442
+ [
1443
+ 155,
1444
+ 55,
1445
+ 0,
1446
+ 87,
1447
+ 0,
1448
+ "IMAGE"
1449
+ ],
1450
+ [
1451
+ 157,
1452
+ 49,
1453
+ 0,
1454
+ 88,
1455
+ 0,
1456
+ "MODEL"
1457
+ ],
1458
+ [
1459
+ 161,
1460
+ 87,
1461
+ 0,
1462
+ 90,
1463
+ 0,
1464
+ "IMAGE"
1465
+ ],
1466
+ [
1467
+ 166,
1468
+ 92,
1469
+ 0,
1470
+ 94,
1471
+ 0,
1472
+ "STRING"
1473
+ ],
1474
+ [
1475
+ 228,
1476
+ 42,
1477
+ 0,
1478
+ 85,
1479
+ 0,
1480
+ "IMAGE"
1481
+ ],
1482
+ [
1483
+ 246,
1484
+ 4,
1485
+ 0,
1486
+ 124,
1487
+ 1,
1488
+ "VAE"
1489
+ ],
1490
+ [
1491
+ 247,
1492
+ 124,
1493
+ 0,
1494
+ 20,
1495
+ 5,
1496
+ "LATENT"
1497
+ ],
1498
+ [
1499
+ 260,
1500
+ 88,
1501
+ 0,
1502
+ 36,
1503
+ 0,
1504
+ "MODEL"
1505
+ ],
1506
+ [
1507
+ 264,
1508
+ 90,
1509
+ 0,
1510
+ 79,
1511
+ 0,
1512
+ "IMAGE"
1513
+ ],
1514
+ [
1515
+ 265,
1516
+ 90,
1517
+ 0,
1518
+ 124,
1519
+ 0,
1520
+ "IMAGE"
1521
+ ],
1522
+ [
1523
+ 266,
1524
+ 5,
1525
+ 0,
1526
+ 129,
1527
+ 0,
1528
+ "CONDITIONING"
1529
+ ],
1530
+ [
1531
+ 267,
1532
+ 6,
1533
+ 0,
1534
+ 129,
1535
+ 1,
1536
+ "CONDITIONING"
1537
+ ],
1538
+ [
1539
+ 268,
1540
+ 129,
1541
+ 0,
1542
+ 20,
1543
+ 1,
1544
+ "CONDITIONING"
1545
+ ],
1546
+ [
1547
+ 269,
1548
+ 129,
1549
+ 1,
1550
+ 20,
1551
+ 2,
1552
+ "CONDITIONING"
1553
+ ],
1554
+ [
1555
+ 270,
1556
+ 2,
1557
+ 0,
1558
+ 20,
1559
+ 0,
1560
+ "MODEL"
1561
+ ],
1562
+ [
1563
+ 271,
1564
+ 2,
1565
+ 0,
1566
+ 22,
1567
+ 0,
1568
+ "MODEL"
1569
+ ],
1570
+ [
1571
+ 272,
1572
+ 4,
1573
+ 0,
1574
+ 129,
1575
+ 2,
1576
+ "VAE"
1577
+ ],
1578
+ [
1579
+ 273,
1580
+ 79,
1581
+ 0,
1582
+ 129,
1583
+ 3,
1584
+ "IMAGE"
1585
+ ],
1586
+ [
1587
+ 165,
1588
+ 93,
1589
+ 0,
1590
+ 92,
1591
+ 0,
1592
+ "CLIP"
1593
+ ],
1594
+ [
1595
+ 248,
1596
+ 55,
1597
+ 0,
1598
+ 92,
1599
+ 1,
1600
+ "IMAGE"
1601
+ ],
1602
+ [
1603
+ 163,
1604
+ 91,
1605
+ 0,
1606
+ 92,
1607
+ 4,
1608
+ "STRING"
1609
+ ],
1610
+ [
1611
+ 274,
1612
+ 9,
1613
+ 0,
1614
+ 130,
1615
+ 0,
1616
+ "IMAGE"
1617
+ ],
1618
+ [
1619
+ 275,
1620
+ 79,
1621
+ 0,
1622
+ 130,
1623
+ 1,
1624
+ "IMAGE"
1625
+ ],
1626
+ [
1627
+ 276,
1628
+ 130,
1629
+ 0,
1630
+ 42,
1631
+ 0,
1632
+ "IMAGE"
1633
+ ]
1634
+ ],
1635
+ "groups": [
1636
+ {
1637
+ "id": 1,
1638
+ "title": "Models",
1639
+ "bounding": [
1640
+ -460,
1641
+ -570,
1642
+ 550,
1643
+ 950
1644
+ ],
1645
+ "color": "#3f789e",
1646
+ "flags": {}
1647
+ },
1648
+ {
1649
+ "id": 2,
1650
+ "title": "Conditioning",
1651
+ "bounding": [
1652
+ 120,
1653
+ -140,
1654
+ 970,
1655
+ 520
1656
+ ],
1657
+ "color": "#A88",
1658
+ "flags": {}
1659
+ },
1660
+ {
1661
+ "id": 3,
1662
+ "title": "Group",
1663
+ "bounding": [
1664
+ 120,
1665
+ -570,
1666
+ 970,
1667
+ 400
1668
+ ],
1669
+ "color": "#3f789e",
1670
+ "flags": {}
1671
+ },
1672
+ {
1673
+ "id": 4,
1674
+ "title": "Sampling",
1675
+ "bounding": [
1676
+ 1106.0160595404368,
1677
+ -500.08001764249127,
1678
+ 740,
1679
+ 950
1680
+ ],
1681
+ "color": "#3f789e",
1682
+ "flags": {}
1683
+ },
1684
+ {
1685
+ "id": 5,
1686
+ "title": "Control Images",
1687
+ "bounding": [
1688
+ -460,
1689
+ 410,
1690
+ 1550,
1691
+ 640
1692
+ ],
1693
+ "color": "#A88",
1694
+ "flags": {}
1695
+ }
1696
+ ],
1697
+ "config": {},
1698
+ "extra": {
1699
+ "ds": {
1700
+ "scale": 0.5730855330117229,
1701
+ "offset": [
1702
+ -132.86812653908146,
1703
+ 966.2211335185046
1704
+ ]
1705
+ },
1706
+ "frontendVersion": "1.58.5",
1707
+ "VHS_latentpreview": false,
1708
+ "VHS_latentpreviewrate": 0,
1709
+ "VHS_MetadataImage": true,
1710
+ "VHS_KeepIntermediate": true
1711
+ },
1712
+ "extensions": {
1713
+ "seed_widgets": {
1714
+ "20": 1
1715
+ }
1716
+ },
1717
+ "version": 0.4,
1718
+ "seed_widgets": {
1719
+ "20": 1
1720
+ }
1721
+ }