ComfyUI inpaint workflow | Updated for anima — 1.5 V3.05
Added ability to use segmentation
Profil d'exécution
- Accès
- Flux de travail externe
- Statut
- Présenté
- Version
- PotatCat_ComfyUi_Inpaint_V3.05.json
- Source
- Civitai
- Mis à jour
- 18 oct. 2026
Description de la source
Added ability to batch
Added ability to use segmentation
What you need
Hardware and download-size estimates are not available yet.
Exigences
Modèles
3control_v11f1e_sd15_tile.pth
Non résoluControlNet · Unknown
lazylatte_v1.safetensors
Non résoluCheckpoint · Unknown
vae-ft-mse-840000-ema-pruned.vae.pt
Non résoluVAE · Unknown
Commentaire généré par l’IA
Explication générée par l’IA à partir des détails de la source et de la configuration. Les suggestions sont clairement signalées.
Objectif
This ComfyUI inpainting workflow is for editing images with an input image, mask, and text prompt; it is described as supporting full-image inpainting and segmentation, with preview and saved image outputs.
Flux
The named components cover image loading, image-to-mask conversion, text encoding, sampling, previewing, and image saving.
Entrées et sorties
The edit mode is described for simple changes such as changing clothes or adding items; the LoRA limits the edit, while other parts generally stay the same.
PreviewImage and SaveImage are named for previewing and saving image results.
Installation
The listed release is version 1.5 V3.05 for SD 1.5, published by PotatCat.
This item is a workflow rather than a Pipe, so a Pipe version does not apply.
Dépendances
The named model files are control_v11f1e_sd15_tile.pth, lazylatte_v1.safetensors, and vae-ft-mse-840000-ema-pruned.vae.pt.
The workflow names these components: LoadImage, ImageToMask, CLIPTextEncode, ScheduleToCond, ScheduleToModel, CheckpointLoaderSimple, CLIPSetLastLayer, VAELoader, VAEEncode, SetLatentNoiseMask, and LatentDuplicator.
It also names GroundingDinoModelLoader (segment anything), GroundingDinoSAMSegment (segment anything), SAMModelLoader (segment anything), ScaledSoftControlNetWeights, GrowMaskWithBlur, Inpaint Segments, MaskToImage, ControlNetLoaderAdvanced, Combine and Paste, JinjaRender, Prompt
Configuration
You can adjust mask opacity and set colormatch to 1; this setting is intended to try to match image color when a VAE, model, or LoRA changes it.
Suggestion · non vérifié
For edit mode, use an edit LoRA with Cosmos Reference; the node and LoRA are described as necessary, and the example uses Denoise: 1.
Suggestion · non vérifié
If shading or a shadow from the original image appears, raise “Expand mask area.”
Suggestion · non vérifié
Précautions
The guide is outdated, the anima release is labeled beta, and bypasser/muter does not work with subgraph.
An older 1.5 path is described as untested, and the named components have no confirmed connections; verify the version, model files, and ImageToMask before use.
Suggestion · non vérifié
Faut-il modifier cela ?
Connectez-vous pour envoyer une demande de modification.
Sources
1 source
Sources
1 sourceExtraits de sources
3 extraits
Extraits de sources
3 extraitsSource context: 22672 downloads · Type Workflows · Base model SD 1.5
Source context: 21675 downloads · Type Workflows · Base model SD 1.5
Source context: 20416 downloads · Type Workflows · Base model SD 1.5
