ComfyUI Node
RedNode Paint Render
Renders ONLY what you painted on the Workspace's Paint tab: it crops to the painted region, samples that crop at the tab's mask size and denoise, and composites it back. The rest of the image is never re-rendered. Wire model, positive, negative, vae, and the image you want to paint on, then press Generate on the Paint tab. Leaving the image unwired makes it follow the tab instead. The result appears on this node and in the tab's result pane. Passes runs that same pass over its own result more than once, which is the low-denoise chain people run by hand to settle a shape; only the last picture comes back.
RedNode Paint Render
- model
- positive
- negative
- vae
- image
- clip
- positive_override
- negative_override
- image
◄seed0►
◄steps8►
◄cfg1.0►
◄sampler_nameeuler►
◄schedulersimple►
◄run_token►
◄passes1►
CategoryRedNode/Image
Inputs (15)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | — | |
| positive | CONDITIONING | — | |
| negative | CONDITIONING | — | |
| vae | VAE | — | |
| seed | INT | 00–18446744073709550000 | — |
| steps | INT | 81–200 | — |
| cfg | FLOAT | 1.00–30 | — |
| sampler_name | COMBO | euler | 44 options: euler, euler_cfg_pp, euler_ancestral, euler_ancestral_cfg_pp, heun, heunpp2, +38 |
| scheduler | COMBO | simple | 9 options: simple, sgm_uniform, karras, exponential, ddim_uniform, beta, +3 |
| run_token | STRING | leave this empty. The Paint tab's Generate button fills it in for its own runs; while it is empty this node sits out of normal queues instead of rendering a crop nobody asked for | |
| passes | INT | 11–10 | How many times to run the paint pass over its own result before handing it back. This is the low-denoise loop done for you: one Generate, one picture at the end, instead of dragging the result back onto the canvas and pressing Generate again. The Paint tab's own count wins when it has one. |
| imageopt | IMAGE | the picture to paint on. Wire it and there is no guessing about which image this is working from. Leave it empty and it falls back to whatever the Paint tab is showing | |
| clipopt | CLIP | wire this to type a prompt on the Paint tab. Without it the patch is rendered with whatever conditioning is on the positive and negative inputs, which is your whole-image prompt | |
| positive_overrideopt | CONDITIONING | conditioning for the painted patch instead of the main positive, for when you want a full grounded encode rather than a plain text one | |
| negative_overrideopt | CONDITIONING | the matching negative for the override |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | — |