Published version

Version details

The selected output is a masked replacement composited over the working image. LoadImage MASK output 1 reaches mask expansion and blur, then latent noise masking, inpainting ControlNet and final compositing. The repaired blur controls independently bind to node 255's source-backed blur_radius and sigma inputs; packageSource.ready=true and qualification reports no issues. Image-only resizing has a documented alignment requirement and an existing bypass. Complete the guidance and expose the retained sampling-shift control. The sibling's node 264 binding mismatch belongs to its own subject review and does not invalidate these distinct family boundaries.

Add to Studio keeps this exact version. Updates won’t replace your saved copy.

PipePrésentéHébergé

Qwen Image masked inpainting

Replace a masked region of an existing image with prompted content using Qwen Image and inpainting ControlNet. The result is composited over the working image; this tool edits within the existing canvas.

Loading this version’s controls and setup…

Workflow details

FamilleQwen-ImageRésolution1024x1024Estimé VRAMNot yet characterizedÉditeurSweet Tea AutomationMis à jour2 nov.BClasse

Profil d'exécution

Éditeur: Sweet Tea Automation
Statut: Présenté
Version: v0.1.2
Mis à jour: 2 nov. 2026

Hover nodes for Pipe controls

Core

Primary Pipe surface

Source image

LoadImage

Input

Replacement prompt

CLIPTextEncode

Working image size

ImageScaleToMaxDimension

Sampling

KSampler

Extended

Additional surfaced controls

Negative prompt

CLIPTextEncode

Mask expansion

GrowMask

Mask softening

ImageBlur

Inpainting guidance

ControlNetInpaintingAliMamaApply

Lightning LoRA

LoraLoaderModelOnly

Sampling shift

ModelSamplingAuraFlow

What you need

Hardware and download-size estimates are not available yet.

How this recipe works

Original pixel preservation · Sweet Tea Preview Image

mask · aligned is true; mask · protected region zero is true; input · same representation is true; input · destination is original is true

Output

Sweet Tea Preview Image · image

Exigences

Modèles

qwen_2.5_vl_7b_fp8_scaled.safetensors
Non résolu
Text encoder · Hugging Face · Comfy-Org/Qwen-Image_ComfyUI
qwen_image_fp8_e4m3fn.safetensors
Non résolu
Diffusion model · Hugging Face · Comfy-Org/Qwen-Image_ComfyUI
qwen_image_vae.safetensors
Dépôt identifié
VAE · Hugging Face · Comfy-Org/Qwen-Image_ComfyUI
Qwen-Image-InstantX-ControlNet-Inpainting.safetensors
Non résolu
ControlNet · Hugging Face · Comfy-Org/Qwen-Image-InstantX-ControlNets
Qwen-Image-Lightning-4steps-V1.0.safetensors
Dépôt identifiéOptionnel
LoRA · Hugging Face · lightx2v/Qwen-Image-Lightning

Packs de nœuds

SweetTeaPreviewImage
Trouvé
Pack de nœud · Unknown
tea-time-labs/sweet-tea-nodes
Trouvé
Pack de nœud · GitHub · tea-time-labs/sweet-tea-nodes

Sources

1 source

Sweet TeaSource

Artefacts du modèle

1 artefact

88eaba0e-c98e-4868-ba58-bb4a97d50eba-comfy-image-qwen-image-instantx-inpainting-controlnet-qwen-image-masked-inpainti-0.1.2.tea

tea · 23,2 KB · SHA-256 75c622570d48…8fd0

Download file