comfymonkv1.1-guidemode-web.mp4
A Photoshop plugin that connects to a ComfyUI backend, enabling AI-powered image generation and inpainting directly inside Photoshop.
Built by the Monks creative technology team.
1. Download the latest .ccx from the Releases page and open it. Photoshop installs it automatically.
2. Open the plugin, click ⚙ Settings, and fill in:
| Field | Value |
|---|---|
| API URL | The shared endpoint — ask a CT for this |
| Gemini Key | Required for Guided (Nano) and QA Tool modes only |
Hit Save.
3. Click Generate. That's it.
The backend scales automatically — no setup, no Wake button. If it's been idle, the first generation takes a few minutes while the backend loads. After that it stays warm and responds instantly.
| Mode | Description |
|---|---|
| Text to Image | Generate an image from a prompt, placed as a new layer |
| Inpaint | Make a selection, describe the edit, get a result placed in place |
| Guided | Inpaint with one or two reference images |
| Guided (Nano) | Guided inpainting powered by Gemini |
| Upscale | Upscale the current canvas with SeedVR2 |
| Upscale (Nano) | Upscale via Imagen4 at 2×/3×/4× |
| QA Tool | Compare ASIN/SKU, original, and regenerated canvas against client feedback |
Generation controls:
- Steps — diffusion steps (default: 4)
- Batch — sequential generations
- Mask Res — inpaint crop resolution (higher = more detail, slower)
- Seed — set manually or lock to repeat a result
The status bar shows whether the backend is warm (🟢 Ready) or cold (⚫ Cold). After each generation a 5-minute countdown runs — if it hits zero before your next generation, expect a short wait on the next one.
300 DPI canvases — the plugin only supports 72 DPI documents. At 300 DPI the result will be placed at the wrong scale.
Workaround: Image → Image Size → uncheck Resample → set Resolution to 72.
Ping #comfy-monk-plugin on Slack.