Zum Inhalt springen
WorkflowAufgelistetExtern
Herunterladen

Video Subject Inpaint (WAN2.2 KJversion+Fun) — v1.0

Fed up? Models too big filling up your storage, nodes a nightmare to install with constant errors, GPU RAM always running out and crashing halfway? RunningHub solves it all! With RTX 4090 24G VRAM in the cloud, all...

Quellecivitai.com/models/1971587
VersionRelease 2231602 — Video Subject Inpaint (WAN2.2 KJversion+Fun).jsonGeschätzte VRAMNot yet characterizedUnratedBewertung

Laufzeitprofil

Zugriff
Externer Arbeitsablauf
Status
Aufgelistet
Version
Video Subject Inpaint (WAN2.2 KJversion+Fun).json
Quelle
Civitai
Aktualisiert
4. Nov. 2026

Quellenbeschreibung

No release description captured.

What you need

Model files
17.3 GiB known subtotal

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

Anforderungen

Modelle

8

Knotenpakete

11

  • BiRefNetRMBG

    Gefunden

    Knotenpaket · Unknown

  • comfyui-kjnodes

    Gefunden

    Knotenpaket · Registry

  • Gefunden

    Knotenpaket · Registry

  • ComfyUI-WanVideoWrapper

    Gefunden

    Knotenpaket · Registry

  • CR Prompt Text

    Gefunden

    Knotenpaket · Unknown

  • Float

    Möglich

    Knotenpaket · Unknown

  • ImageResizeKJv2

    Möglich

    Knotenpaket · Unknown

  • LayerUtility: ImageScaleByAspectRatio V2

    Möglich

    Knotenpaket · Unknown

  • LayerUtility: PurgeVRAM

    Gefunden

    Knotenpaket · Unknown

  • WanVideoScheduler

    Möglich

    Knotenpaket · Unknown

  • WanVideoSigmaToStep

    Möglich

    Knotenpaket · Unknown

Quellen

1 Quelle
CivitaiPrimärVersion 2231602
civitai.com/models/1971587

Verifiziert 4. Nov.

Workflow instructions

Author notes from Video Subject Inpaint (WAN2.2 KJversion+Fun).json · selected release

longest side

Author notes from Video Subject Inpaint (WAN2.2 KJversion+Fun).json · selected release

total frame

Author notes from Video Subject Inpaint (WAN2.2 KJversion+Fun).json · selected release

If you just want to replace the subject without using a reference, you can disconnect the ref_image input and rely solely on text to control the generation results.

Author notes from Video Subject Inpaint (WAN2.2 KJversion+Fun).json · selected release

Replace the subject with one that closely matches the original shot, either as a half-body or full-body portrait, and ensure the image maintains the same vertical or horizontal orientation. For better results, you can use the above-mentioned disabled node to extract the first frame, then process it with another image inpainting workflow to achieve improved effects. You can use this Qwen Image workflow to create a reference image. Link: https://www.runninghub.ai/workflow/1968978526189862914?inviteCode=rh-v1153

Author notes from Video Subject Inpaint (WAN2.2 KJversion+Fun).json · selected release

WAN2.2 FUN VACE KJ Version: https://huggingface.co/Kijai/WanVideo_comfy_fp8_scaled/tree/main/VACE WAN2.2 T2V KJ Version: https://huggingface.co/Kijai/WanVideo_comfy_fp8_scaled/tree/main/T2V WAN2.2 FUN Reward: https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan22_FunReward/Wan2.2-Fun-A14B-InP-LOW-HPS2.1_resized_dynamic_avg_rank_15_bf16.safetensors (not necessary) WAN2.1 Lihgtx2v: https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Lightx2v/lightx2v_I2V_14B_480p_cfg_step_distill_rank64_bf16.safetensors

Download files

Modellartefakte

2 Artefakte

videoSubjectInpaintWAN2_v10.zip

zip · 12,1 KB · SHA-256 e55fda6f9375…4c18 · Civitai

Download file

Video Subject Inpaint (WAN2.2 KJversion+Fun).json

json · 100 KB · SHA-256 bd92ddf7a373…af84

Download file

Quellenauszüge

2 Auszüge
Note4. Nov.civitai.com/models/1971587

Source context: 285 downloads · Type Workflows · Base model Wan Video 2.2 T2V-A14B

Note16. Okt.civitai.com/models/1971587

Source context: 279 downloads · Type Workflows · Base model Wan Video 2.2 T2V-A14B