Adaptation to Photoshop color scheme
Plugin element colors change together with Photoshop color scheme changes
Prompt and Negative Prompt
Text areas for your AI prompts. Sent directly to ComfyUI via prompt.txt and negative.txt files.
Fixed Seed / Random seed
Seed behavior control:
- Fixed: Use specific seed value for reproducible results
- Random: ComfyUI generates new seed each run
- Actual used seed displayed after generation completes
Additional image support
Upload up to 3 additional images for:
- Inpainting workflows
- Image-to-image generation
- Multi-input image scenarios
Images saved as extra1.png, extra2.png, extra3.png to exchange folder. By default active mode is resize and convert when exceeding 1 megapixel (button "↓1Mp" - green). If button pressed ("↓1Mp" - red), additional images will be sent to workflow unchanged.
Output Mode Selection (Where generation goes)
Under the Run button, there is an output mode selector with three options, allowing you to configure where generation results are stored:
- Photoshop: The result is returned directly to the Photoshop canvas as a new layer (convenient for fast editing and compositing).
- Folder: Images are saved as files into the local
Output directory on your drive. - All: The result is both returned as a new layer in Photoshop and saved as a file in the
Output directory.
Next to the selector is the FOLDER button, which instantly opens the target output folder on your computer. This is extremely useful for quickly browsing generation results after a batch run to find the best option.
The filenames of saved results contain full generation parameter details, for example, F2-Klein 5-2 s-4035056786.png:
F2-Klein — name of the ComfyUI workflow used.5-2 — slot color set (5) and slot number (2) from which the generation was triggered.s-4035056786 — seed value used for generation.
Prompt Slots (Slots, Import & AUTO Mode)
5 color sections, 6 numbered slots for parameter presets, with fast file loading, loop mode, and batch generation options:
- Loading mode (default): Loads saved set with all parameters (prompts, seed, etc.)
- Saving mode: Saves current settings to slot (buttons highlight red)
- Import Prompts: Fast loading of prompts from a
.txt file: - Button 6▼ — imports up to 6 prompts into the active color set (slots 1–6).
- Button 30▼ — sequentially populates all 30 slots across all 5 color sets (colors 1-5, slots 1-6).
- If the file has fewer prompts, loading stops, leaving the remaining slots untouched.
- If a prompt line is followed by a line starting with
N: text, it will be loaded into the negative prompt field. If followed by D: text, it will be loaded into the slot description field. - Lines starting with
# are treated as comments and skipped.
- AUTO Mode: Toggles batch generation through all slots. When enabled, it sets the Batch count to 6, and during batch generation sequentially runs through prompts from slots 1 to 6 for each iteration. If you manually set a smaller batch count (e.g. 3), the batch will start from the currently selected prompt slot and iterate sequentially (e.g. slots 3 and 4, if slot 3 is selected).
- Loop Mode (
∞): Located to the right of the AUTO button. When enabled, if generation is started at a slot (e.g. slot 5 in the pink set) and the total generation count (batch size) exceeds the remaining slots in that set, the generation seamlessly transitions to the next color set (e.g. green slot 1, 2, 3, etc.) instead of stopping or wrapping around within the same set. - Automatically restored when plugin launches
- Customizable colors in settings - PROMPT_SET_COLORS=#ffb0b0,#388e3c,#1565c0,#f9a825,#7b1fa2
Snap to Grid
Automatic size alignment for different models:
- FLUX x64: Sizes rounded to 64px (for FLUX models, can change in settings - sm_flux = 64)
- Qwen x112: Sizes rounded to 112px (for Qwen models, can change in settings - sm_qwen = 112)
- Personal x8: Sizes rounded to 8px (configurable any size - sm_personal = 8)
- Auto-selected based on workflow name for Flux and Qwen
- Selection brush tool: When selecting with this tool, context padding is automatically added on export to reach a size that is a multiple of the value specified in SNAP.
Workflows
When plugin loads, folder <ComfyUI>\user\<User>\workflows\Photoshop is scanned. All nested folders at 1 level are added to plugin interface. If you change workflows during work, add new ones or modify and save existing - you can re-scan the folder by pressing "↻ Update Workflows". Several workflows supplied with plugin:
- Base - all Workflow for working with selected area. If no selection, you will see error when launching. Number of steps in name - more = better quality. For Flux 2 klein usually 8 steps is enough. For Qwen 2511 - 4 and 8 steps using Qwen-Image-Edit-2511-Lightning-4 (8)steps-V1.0-bf16.safetensors, usually this is enough. If you need to increase quality - choose CUSTOM step / CFG and configure your parameters.
- Canvas - workflows for working with entire artboard. Entire artboard sent to workflow. Numbers mean what is transferred, e.g. "F2-Klein 1+1" - 1+1 - for processing artboard + can connect 1 additional image. "Qwen 0+3" - artboard not transferred, work only with 3 loaded images (can load fewer - 1 or 2). "F2-Klein Restore&Colorise" - prompt needed here as it built into workflow. "Qwen 1+2" - artboard transferred + up to 2 loaded images
- Txt-Img - Image generation from text prompt.
- Zoom - Image enlargement. Connected scheme with fast upscale model. By default configured for minimum configuration. If you have 10-12 Vram or modern card (tested on 5060 / VRAM) - try changing SeedVR2_up_to 4096.json in GUI ComfyUI and select bigger model (e.g. seedvr2_ema_7b_fp8_e4m3fn_mixed_block35_fp16.safetensors). How enlargement works: to enlarge document size along with selected size press "Expand" button. If not pressed, process completes and image is enlarged but fits into current document. Zoom in by double-click to see real size. If your scheme has enlargement - e.g. as in "Qwen 2512 + Upscale", for Photoshop document enlargement need to press "Expand" button in "Zoom" block.
- Note #1 - If your scheme has enlargement - e.g. as in "Qwen 2512 + Upscale", for Photoshop document enlargement need to press "Expand" button in "Zoom" block.
- Note #2 - Button "↑1 Mp" in "Zoom" block - enlarges selection to 1 megapixel. Needed when selection is very small and some models start distorting image heavily. Enlargement can be built into workflow or press this button.
- Note #3 When exceeding configured image size (MAX_IMAGE_MP=4), warning appears during generation send. If agree - image sent unchanged to ComfyUI, and for sizes above 4 megapixels possible slowdown and sometimes hanging. This block offers reducing image to 2/3/4 megapixels. Green buttons - with step from Snap menu (64 for Flux, 112 for Qwen). Blue - without step binding.
Guide
Adds guides with selected step from selection. Parameter configurable: GUIDE_PRESETS=100,200,300,500
Add to content
Block for expanding selection area before sending to ComfyUI. Adds padding around selection — useful for inpainting tasks when model needs additional context around edited area.
Input field - padding values in pixels in `top/right/bottom/left` format (or shortened: one value for all sides, two — top-bottom and left-right). Preset buttons (default: 64, 128, 256, 112, 224, 336 px) — press sets equal padding from all sides. Custom Step & CFG
Send custom parameters to workflow nodes:
- Custom Step: Integer (up to 4 digits)
- CFG: Guidance scale (up to 2 decimal places)
Tools Tab
Language
Interface translation into 8 languages: English, Spanish, Japanese, German, Chinese, Portuguese, French, Russian.
- Save — save and apply the language selection.
Workflow folder
Displays the current path to the JSON workflow folder (configured in config.txt). The path can be edited directly in the input field — the list of workflows reloads automatically upon change. Use the ↻ Update Workflows button to refresh the list manually.
ComfyUI Server URL
Field for entering the ComfyUI server address (default is http://localhost:8188). Buttons:
- TEST — checks server availability (5-second timeout).
- SAVE — saves the URL to
ui_prefs.json to restore the connection configuration on the next launch.
Saved prompts history
Each time a prompt is saved to a slot, a log entry is added to prompts_old.txt. Buttons:
- Open .txt — opens the log file in your system text editor.
- Open .html — generates and opens a readable, well-formatted HTML version of the history log.
The ON / OFF status shows whether prompt logging is enabled (managed by the SAVE_PROMPTS_HISTORY parameter in config.txt). File opening behavior is controlled by OPEN_LOGS_METHOD (options: auto, mac, or pc).
All prompts history
Every generation run is automatically logged in all_prompts.txt. The same Open .txt and Open .html buttons are available for viewing. History recording can be turned ON / OFF using the SAVE_ALL_PROMPTS parameter in config.txt.
Clear all prompts
The 🗑 Clear all prompts button completely clears both history files (all_prompts.txt and prompts_old.txt) and resets all 30 prompt slots in the plugin. Confirmation is required before execution.
View images (Canvas / Mask)
Displays thumbnails of the most recently sent files from the active exchange directory:
- Canvas — thumbnail of
canvas.png (the canvas image sent to ComfyUI on the last run). - Mask — thumbnail of
mask.png (the selection mask, if it was used).
Actual dimensions in pixels and size in megapixels (Mp) are displayed below each thumbnail. The ↻ button refreshes the preview manually.
Add bleed
Expands the document canvas size uniformly on all sides by the specified value.
- Unit Selection: You can select the active unit system — Millimeters, Inches, or Pixels — directly in the UI. Switching units automatically recalculates dimensions and updates preset button values.
- Active Document Dimensions Button: Displays the current width and height of the active document. If document dimensions change but do not refresh automatically in the plugin panel, click this button to force-read and update the canvas size (flashes green upon success).
- Preset Buttons: Clicking a preset button instantly applies the canvas expansion.
The default unit of measurement and presets are defined in config.txt via the BLEED_UNIT (values: mm / inches / pixel) and BLEED_PRESETS parameters (e.g. 2,3,5,-2,-3,-5, where negative values shrink/crop the canvas).
Crop & Uncrop
Allows you to precisely crop (shrink) or uncrop (expand) the document canvas along selected edges.
- Unit Selection: You can switch active units between Millimeters, Inches, and Pixels directly within the block control.
- Resize Directions: Four direction buttons (Up, Down, Left, Right arrows) let you specify which edges of the document canvas should be cropped or expanded. You can select multiple directions at once.
- Uncrop: Expands the canvas in the selected directions by the value specified in the input field.
- Crop: Shrinks the canvas (crops the image) from the selected sides by the specified value.
Default settings are configured in config.txt via the UNCROP_CROP_UNIT parameter (falls back to BLEED_UNIT if not defined).
Guides for bleed
Automatically places helper guides relative to document edges to assist with layout alignment, trim margins, and safe areas.
- Unit Selection: Switch active offset units between Millimeters, Inches, and Pixels dynamically in the UI.
- Preset Buttons: Each button sets a pair of guides
inner+outer (e.g. 3+7 draws an inner guide at 3 units and an outer guide at 7 units from the document edge). If the outer value is 0, only one guide is created per side. - Center canvas — places vertical and horizontal guides directly through the center axes of the canvas.
- Center select — places guides crossing exactly through the center of the active selection area.
- Clear — instantly clears all guides in the current document.
The default unit of measurement and presets are configured in config.txt via the ADD_GUIDE_UNIT and ADD_GUIDE_PRESETS parameters (by default: mm, presets 3+7 / 2+5 / 2+0).