Prioritize regional subjects before environment

API.md

@@ -56,7 +56,7 @@ Content-Type: application/json
 }
 ```
 
-Returns a PNG with ComfyUI workflow and parameter metadata, scaled after generation by the decimal `return_scale` from `0.01` through `1`, with content type `image/png`. Regional prompts and soft mask-conditioning nodes are included in the metadata. Use `1` for full resolution. `regions` accepts up to four spatial prompts on a 5 by 5 grid. Global and regional prompts guide the full sampling process, with reduced global strength when regions are present. Each internal region edge feathers by half a grid cell, so overlap adjacent regions by one cell for a smooth crossfade. `area` accepts `full`, `tl`, `tc`, `tr`, `ml`, `mc`, `mr`, `bl`, `bc`, `br`, `th`, `mh`, `bh`, `lh`, `ch`, or `rh`. It also accepts any single cell from `a1` through `e5`, or a rectangular inclusive range such as `a1:c5`. Columns run left to right and rows run top to bottom. Region strength defaults to `1`. Use the global `prompt` for subject count, composition, interaction, background, lighting, and shared style. Omitting `regions` keeps whole-image generation unchanged. Set either scheduler to `AlignYourSteps` to use `AlignYourStepsScheduler` with fixed model type `SDXL` and the corresponding steps and denoise values. Only `prompt` is required for the built-in site defaults. LoRAs run in array order. Each request creates fresh sampler seeds. When `upscale` is true, the request performs latent upscaling followed by a second sampler. Leave `second_model` empty to reuse the first model and LoRA chain; otherwise `second_loras` applies to the selected second model. Width and height must be multiples of 8 from 64 through 2048; steps must be from 1 through 100.
+Returns a PNG with ComfyUI workflow and parameter metadata, scaled after generation by the decimal `return_scale` from `0.01` through `1`, with content type `image/png`. Regional prompts and soft mask-conditioning nodes are included in the metadata. Use `1` for full resolution. `regions` accepts up to four spatial prompts on a 5 by 5 grid. Regional prompts establish subjects from the first sampling step. The reduced-strength global prompt starts after the first fifth to complete the environment while regions remain active. Each internal region edge feathers by half a grid cell, so overlap adjacent regions by one cell for a smooth crossfade. `area` accepts `full`, `tl`, `tc`, `tr`, `ml`, `mc`, `mr`, `bl`, `bc`, `br`, `th`, `mh`, `bh`, `lh`, `ch`, or `rh`. It also accepts any single cell from `a1` through `e5`, or a rectangular inclusive range such as `a1:c5`. Columns run left to right and rows run top to bottom. Region strength defaults to `1`. Use the global `prompt` for environment, lighting, shared style, and composition outside regions. Omitting `regions` keeps whole-image generation unchanged. Set either scheduler to `AlignYourSteps` to use `AlignYourStepsScheduler` with fixed model type `SDXL` and the corresponding steps and denoise values. Only `prompt` is required for the built-in site defaults. LoRAs run in array order. Each request creates fresh sampler seeds. When `upscale` is true, the request performs latent upscaling followed by a second sampler. Leave `second_model` empty to reuse the first model and LoRA chain; otherwise `second_loras` applies to the selected second model. Width and height must be multiples of 8 from 64 through 2048; steps must be from 1 through 100.
 
 ## Archive
 

app.py

@@ -77,6 +77,7 @@ IMAGE_SUFFIXES = (".epng", ".png")
 PREVIEW_QUALITY = 75
 PREVIEW_SCALE = .25
 DEFAULT_RETURN_SCALE = 1
+ENVIRONMENT_START = .2
 REGIONAL_GLOBAL_STRENGTH = .6
 AREA_PRESETS = {
     "full": "a1:e5",
@@ -655,7 +656,15 @@ def encode_positive(clip, prompt, regions, image_width, image_height):
     positive = state["encode"].encode(clip=clip, text=prompt)[0]
     if regions:
         positive = [
-            [item[0], {**item[1], "strength": REGIONAL_GLOBAL_STRENGTH}]
+            [
+                item[0],
+                {
+                    **item[1],
+                    "start_percent": ENVIRONMENT_START,
+                    "end_percent": 1,
+                    "strength": REGIONAL_GLOBAL_STRENGTH,
+                },
+            ]
             for item in positive
         ]
     for region in regions:
@@ -1151,6 +1160,7 @@ def generate_image(request, from_api=True):
                 [first_seed, second_seed],
                 [first_vae, second_vae],
                 regions,
+                ENVIRONMENT_START,
                 REGIONAL_GLOBAL_STRENGTH,
             )
         )

workflow_api.py

@@ -635,6 +635,7 @@ def image_metadata(
     seeds,
     vaes,
     regions,
+    environment_start,
     global_strength,
 ):
     api = {}
@@ -731,6 +732,11 @@ def image_metadata(
                 "conditioning": positive,
                 "strength": global_strength,
             }), 0]
+            positive = [add("ConditioningSetTimestepRange", {
+                "conditioning": positive,
+                "start": environment_start,
+                "end": 1,
+            }), 0]
         for prompt, x, y, width, height, strength in regions:
             conditioning = [add("CLIPTextEncode", {
                 "text": prompt, "clip": clip,