WorkflowPrésentéExterne

VACE Subject Replace! Replace Anything in Videos with Wan2.1+VACE — SubjectReplacerV1

Replace any subject in a video! This could also be enhanced by incorporating depth/openpose/inverted lineart in the workflow instead of just a gray background.

Sourcecivitai.com/models/1470557

TâcheMultipleCapacitésImage en image · Texte en imageFamilleWanVersionRelease 1663316 — WanWrapperVaceSubjectReplace.jsonEstimé VRAMNot yet characterizedUnratedClasse

Profil d'exécution

Accès: Flux de travail externe
Statut: Présenté
Version: WanWrapperVaceSubjectReplace.json
Source: Civitai
Mis à jour: 3 nov. 2026

Description de la source

No release description captured.

What you need

Model files: 1.5 GiB known subtotal

Remaining hardware or download-size estimates are not available yet.

Exigences

Modèles

depth_anything_v2_vitl.pth
Non résolu
Checkpoint · Unknown
dw-ll_ucoco_384_bs5.torchscript.pt
Trouvé
Checkpoint · 129 MB · PT · Unknown
flux/flux_vae.sft
Non résolu
VAE · Hugging Face · flux/flux_vae.sft
flux/flux1-dev.sft
Non résoluOptionnel
Diffusion model · Hugging Face · flux/flux1-dev.sft
instantX-Flux-Union.safetensors
Non résoluOptionnel
ControlNet · Unknown
juggernautXL_juggXIByRundiffusion.safetensors
Non résolu
Checkpoint · Unknown
sai_xl_depth_256lora.safetensors
Non résolu
ControlNet · Unknown
sam2.1_hiera_large.safetensors
Non résolu
Vision model · Unknown
Unresolved model selection: DualCLIPLoader
Non résoluOptionnel
Text encoder · Unknown
Wan2_1_VACE_1_3B_preview_bf16.safetensors
Fichier indiqué
Checkpoint · 1.37 GB · Hugging Face · Kijai/WanVideo_comfy
Wan2_1-T2V-1_3B_bf16.safetensors
Non résolu
Checkpoint · Unknown
wrapper_umt5-xxl-enc-bf16.safetensors
Non résolu
Text encoder · Unknown
wrapper_Wan2_1_VAE_fp32.safetensors
Non résolu
VAE · Unknown
yolox_l.onnx
Non résolu
Checkpoint · Unknown

Packs de nœuds

comfyroll
Non résolu
Pack de nœud · Registry
comfyui_controlnet_aux
Possible
Pack de nœud · Registry
comfyui_essentials
Trouvé
Pack de nœud · Registry
comfyui_layerstyle
Trouvé
Pack de nœud · Registry
comfyui-kjnodes
Trouvé
Pack de nœud · Registry
comfyui-videohelpersuite
Trouvé
Pack de nœud · Registry
ComfyUI-WanVideoWrapper
Trouvé
Pack de nœud · Registry
Fast Groups Bypasser (rgthree)
Non résolu
Pack de nœud · Unknown
GetNode
Non résolu
Pack de nœud · Unknown
segment-anything-2
Non résolu
Pack de nœud · Registry
SetNode
Non résolu
Pack de nœud · 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 workflow is for replacing a subject in a video: provide a video, and it can produce a video with that subject replaced using Wan2.1+VACE.

Entrées et sorties

Input: a video. Output: a video with the subject replaced.

Installation

It is listed as a workflow and uses Wan Video as its base model.

Dépendances

The listed node packs are comfyroll, comfyui_controlnet_aux, comfyui_essentials, comfyui_layerstyle, comfyui-kjnodes, comfyui-videohelpersuite, ComfyUI-WanVideoWrapper, and segment-anything-2.
The listed model files include flux/flux_vae.sft, flux/flux1-dev.sft, instantX-Flux-Union.safetensors, juggernautXL_juggXIByRundiffusion.safetensors, and sai_xl_depth_256lora.safetensors.
Other listed model files are sam2.1_hiera_large.safetensors, wan/Wan2_1_VACE_1_3B_preview_bf16.safetensors, wan/Wan2_1-T2V-1_3B_bf16.safetensors, and wan/wrapper_Wan2_1_VAE_fp32.safetensors.

Configuration

To move beyond the gray-background approach, consider adding depth, openpose, or inverted lineart.
Suggestion · non vérifié

Précautions

Before setup, verify which package provides comfyui_controlnet_aux.
Suggestion · non vérifié
Before use, check that every listed model file is available under its exact filename.
Suggestion · non vérifié

Faut-il modifier cela ?

Connectez-vous pour envoyer une demande de modification.

Sources

1 source

CivitaiPrincipalVersion 1663316

civitai.com/models/1470557

Vérifié 3 nov.

Workflow instructions

Author notes from WanWrapperVaceSubjectReplace.json · selected release

https://huggingface.co/Kijai/WanVideo_comfy/tree/main Download the Wan2_1_VACE_1_3B_preview_bf16.safetensors file and put it into ComfyUI>models>diffusion_models, then select it in the "WanVideo VACE Model Select" The rest of the models are the same as base Wan2.1. If you do not already have Wan2.1 Setup, GO TO THIS LINK AND WATCH THE VIDEO: https://youtu.be/D4VLXF4ZNq0?si=2c3ekxuDbKbc5m-K

Author notes from WanWrapperVaceSubjectReplace.json · selected release

<- To learn how to use this click the tool tip!

Author notes from WanWrapperVaceSubjectReplace.json · selected release

DO NOT USE MASK BLURRING. VACE doesn't handle the blurring properly and you'll be left with artifacts

Author notes from WanWrapperVaceSubjectReplace.json · selected release

You can use any pre-processor you'd like, I'm using Depth here. DWPose, Canny, Lineart, all work as expected.

Author notes from WanWrapperVaceSubjectReplace.json · selected release

The flux group takes a lot of VRAM, if you are strugging with that, you can use SD15 or SDXL instead Flux model: https://huggingface.co/black-forest-labs/FLUX.1-dev Flux Clips (Text Encoders): https://huggingface.co/comfyanonymous/flux_text_encoders/tree/main For flux, make sure to download flux union controlnet. https://huggingface.co/Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro/tree/main If you want to use the SDXL controlnet, just make sure to switch the "controlnet image" node SDXL Model: https://civitai.com/models/133005/juggernaut-xl SDXL Controlnets: https://huggingface.co/lllyasviel/sd_control_collection/tree/main Put the controlnets in the models > controlnet folder

Author notes from WanWrapperVaceSubjectReplace.json · selected release

You can use any pre-processor you'd like, I'm using lineart here. DWPose, Canny, Lineart, all work as expected.