File size: 5,361 Bytes
717fadd | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 | {
"id": "be1e85b1-4ede-4c33-a207-02c277979872",
"revision": 0,
"last_node_id": 4,
"last_link_id": 2,
"nodes": [
{
"id": 1,
"type": "MiniMaxH3RefModFolderLoader",
"pos": [
40,
160
],
"size": [
520,
360
],
"flags": {},
"order": 0,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "refs",
"type": "H3_REF_LIST",
"links": [
1
],
"slot_index": 0
},
{
"name": "count",
"type": "INT",
"links": [],
"slot_index": 1
}
],
"properties": {
"Node name for S&R": "MiniMaxH3RefModFolderLoader"
},
"widgets_values": [
"/workspace/datasets/CHANGE_TO_YOUR_FOLDER",
8,
16,
1280
],
"title": "1 | UPLOAD CHARACTER PHOTOS"
},
{
"id": 2,
"type": "VAELoader",
"pos": [
40,
600
],
"size": [
520,
100
],
"flags": {},
"order": 1,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "VAE",
"type": "VAE",
"links": [
2
],
"slot_index": 0
}
],
"properties": {
"Node name for S&R": "VAELoader"
},
"widgets_values": [
"minimax_h3_video_vae_fp16.safetensors"
],
"title": "H3 video VAE - leave selected"
},
{
"id": 3,
"type": "MiniMaxH3RefModExtract",
"pos": [
680,
160
],
"size": [
560,
850
],
"flags": {},
"order": 2,
"mode": 0,
"inputs": [
{
"name": "refs_bundle",
"type": "H3_REF_LIST",
"link": 1
},
{
"name": "vae",
"type": "VAE",
"link": 2
}
],
"outputs": [
{
"name": "mods",
"type": "H3_REF_MODS",
"links": [],
"slot_index": 0
},
{
"name": "details",
"type": "STRING",
"links": [],
"slot_index": 1
}
],
"properties": {
"Node name for S&R": "MiniMaxH3RefModExtract"
},
"widgets_values": [
"my_character_v1",
"Full Reference",
"identity",
0.0,
768,
16,
16,
16,
0,
false,
false,
1,
0,
"Reference photos of one character.",
true,
"manual",
"",
"error"
],
"title": "2 | CHANGE NAME, THEN RUN"
},
{
"id": 4,
"type": "Note",
"pos": [
40,
790
],
"size": [
560,
730
],
"flags": {},
"order": 3,
"mode": 0,
"properties": {},
"widgets_values": [
"CHARACTER REFMOD CREATOR\n\n1. Choose 5-8 clear photos of ONE character. One image works, but multiple views are useful. Include front, three-quarter, profile and a wider shot. Avoid blur, heavy filters, sunglasses and near-duplicates. Captions are not needed.\n2. Click Upload photos, or drop 1\u20138 selected photos onto the first node's upload area. Wait for photos ready. PNG, JPG, WebP or BMP; up to 20 MB each. The folder and count are set automatically.\n3. Set a unique name such as alex_h3_v1 in the Create node. Existing files with the same name are replaced when saved.\n4. Leave the other settings as supplied and click Run. This VAE-only creation workflow does not load H3's video generation model or train a LoRA.\n5. The saved file is /workspace/models/refmods/<name>.safetensors. In your generation workflow, refresh RefMods and select it in Load H3 RefMods.\n\nFull Reference preserves detail. Refinement Steps, pooling and merge are unused in this mode. concept_type=identity is descriptive metadata; it does not train facial recognition. max_tokens=0 prevents dropping reference views to fit a cap, so more photos increase generation cost.\n\nCROP NOTE: this simple creator stacks multiple photos into ONE video-kind reference, cropping them to the first photo's aspect ratio. Use matching aspect ratios and keep faces well inside the frame. With a single photo it creates an image-kind reference.\n\nPROMPT LABELS: with H3 RefMod Text Encode, a multi-photo file made here normally appears as <Video 1>, not eight separate <Picture> labels. Use the encoder's reference_map output to check. Separate-image bundles use Picture labels for each view. Always update the character-reference lines when replacing files.\n\n\nUPLOAD SETUP: Install the RefMod Pilot repository under ComfyUI/custom_nodes, restart ComfyUI, then refresh the page. This is a ComfyUI extension, not a browser add-on. Uploading does not run the workflow.\n"
],
"title": "READ ME / four simple steps",
"color": "#432",
"bgcolor": "#653"
}
],
"links": [
[
1,
1,
0,
3,
0,
"H3_REF_LIST"
],
[
2,
2,
0,
3,
1,
"VAE"
]
],
"groups": [],
"config": {},
"extra": {
"ds": {
"scale": 0.62,
"offset": [
30,
20
]
}
},
"version": 0.4
}
|