Character Replacement Video

What you’ll build

Replace the character in an existing clip with a reference face using local Wan VACE. Every step runs locally on your own GPU. No subscription. No per-render meter. The chain is yours, the file is yours, the render lives on your drive forever.

The chain

One pipeline node:

1. 1
   Wan VACE face swap (video-to-video)Local~~

   Replace the face on an existing video clip with your reference identity. Local, free. WHAT: A source video + a reference face image in, the same video back with the face swapped out. Locks identity across every frame via PuLID conditioning on top of Wan 2.2 VACE. WHY: The technique behind viral 'insert yourself into Game of Thrones' or 'fake news chase' content. Free locally. No per-clip cloud bill, no faces uploaded to a SaaS. WHEN: Inserting yourself or any character into existing footage. Fake-news / fake-interview ad genre. Replacing a stunt double with the actor. Cameo-style brand spots. WHEN NOT: If you want to ANIMATE a still (motion from one starting image), use Wan 2.2 I2V instead. If you want to morph between two locked stills, use Wan 2.2 FLF2V. Face swap is for v2v specifically — existing video in, swapped video out. HOW: The reference face must be a clean, well-lit, neutral-expression headshot. Bad reference = bad identity lock. Source clips of 3-8 seconds work best — longer drifts. SETUP: Same install as Wan 2.2 FLF2V — kijai/WanVideoWrapper custom nodes + Wan 2.2 Q5_K_M GGUF (~12 GB) in ComfyUI/models/diffusion_models/. One setup unlocks both nodes.

Prerequisites

• An NVIDIA RTX GPU with 12 GB+ VRAM. 24 GB recommended for the full pipeline. Check system requirements →
• Hybrig desktop installed and your worker connected. Get the desktop app →
• Source video on hand. MP4, MOV, or WebM. Anything your camera shoots works.

Try it

The fastest way to learn this chain is to drop it onto the Studio canvas and run it on your own rig. The whole graph is pre-wired.

Watch