Legacy2_ClownInpaintSimple
The no-fuss RES4LYF inpainting node
- model
- positive_inpaint
- negative
- latent_image
- mask
- model
- positive
- negative
- latent
- guides
This is the stripped-down sibling of ClownInpaint. Same job - set up an inpaint guide for a RES4LYF sampler - but without the separate background prompt and its own weight and schedule. You give it one prompt for the masked region, a mask, and the usual model/latent/negative, and it builds the GUIDES bundle. If you don't need to describe the background separately, this is the one to grab; it's less wiring and fewer knobs to second-guess.
RES4LYF's guides are a nice way to steer a generation without a ControlNet, and pointed at inpainting they let you regenerate a masked area under prompt control. This is the Legacy version (display name Legacy2_ClownInpaintSimple) - kept around for older graphs, still works fine.
How it works
You supply positive_inpaint, a mask, and the model, latent, and negative. The node guides the masked region toward your inpaint prompt on a single weight and schedule, passes the model/positive/negative/latent through, and emits the guides bundle for the sampler's guides input. Because there's no background conditioning, everything outside the mask is left to the model's default handling rather than steered by a second prompt - which is exactly the simplification that makes this node "simple."
The inputs that matter
positive_inpaint+mask- the prompt for the area you're regenerating, and where it applies. This is the whole setup.guide_weight(default0.1) - how hard the masked region is pushed toward the prompt. Low by default (gentle inpaint); raise it if the region isn't following your prompt.guide_end_step(default15) - the step where the guide releases and lets the model finish on its own. Extend it if the inpaint needs to be held longer.
guide_weight_scheduler (default beta57) shapes how the weight moves over steps. The single meaningful outputs are the passthroughs (model, positive, negative, latent) and guides (GUIDES), which is the one you feed to the sampler.
Installing RES4LYF
ComfyUI Manager: search RES4LYF, install, restart. Or:
cd ComfyUI/custom_nodes
git clone https://github.com/ClownsharkBatwing/RES4LYF
pip install -r RES4LYF/requirements.txt
Portable users: embedded pip under python_embedded/Scripts/. Restart, hard-refresh with F5. No model downloads.
Common issues
Inpaint too weak or too strong. guide_weight is your dial - 0.1 is gentle. Up for a firmer follow of the prompt, down for a lighter touch. Pair with guide_end_step (hold longer vs release earlier).
The background you wanted to describe isn't described. That's by design - this node has no background prompt. If you need to steer the area outside the mask with its own prompt, use the full ClownInpaint instead.
Feed the guides to the sampler. The guides output is what drives the inpaint; wire it into a RES4LYF sampler's guides input. Stock samplers won't consume it.
Nothing masked, nothing happens. No mask means no region to inpaint. Make sure a mask is connected and actually covers the area you intend to regenerate.
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| guide_weight | FLOAT | 0.10-100–100 | Set the strength of the guide. |
| guide_weight_scheduler | COMBO | beta57 | 12 options: constant, simple, sgm_uniform, karras, exponential, ddim_uniform, +6 |
| guide_end_step | INT | 151–10000 | — |
| modelopt | MODEL | — | |
| positive_inpaintopt | CONDITIONING | — | |
| negativeopt | CONDITIONING | — | |
| latent_imageopt | LATENT | — | |
| maskopt | MASK | — |
Outputs (5)
| Name | Type | Description |
|---|---|---|
| model | MODEL | — |
| positive | CONDITIONING | — |
| negative | CONDITIONING | — |
| latent | LATENT | — |
| guides | GUIDES | — |