Wan-Dancer’s ComfyUI Setup Needs Global and Local Weights Plus Four Extra Support Files
Thick_Impression_307 · reddit · 2026-07-26
A Reddit post maps out the full ComfyUI setup for Wan-Dancer, because the official repos do not contain everything you need.
What the author found
- The official Wan-AI repo is not a ComfyUI package; it uses DiffSynth and shell scripts instead.
- For the model weights, you need both the global and local files — either FP8 or GGUF — because Wan-Dancer is a two-pass system where the global model plans keyframes and the local model refines them.
- The missing support files come from multiple repos: text encoder, VAE, CLIP vision, and a LoRA from a lightx2v repo.
Workflow notes
- The working JSON workflow lives in the GGUF repo.
- It depends on Rebels Audio Nodes for audio preprocessing, so those custom nodes must be installed first.
- Early GGUF quants were reportedly corrupt, so the author advises pulling the files again.
- The source repo naming is messy, with the live one resolving as Wan-AI/Wan-Dancer-14B rather than the README’s stale pointer.
Why it’s useful
The post is basically an installation and file-path audit, not a benchmark claim, and it avoids inventing VRAM numbers the authors never published.
More from coding & agent
- PinchTab ships a 12MB Go browser-control binary for AI agents, with HTTP API and token-saving diff mode — Shruti_0810 · 2026-07-26
- GRPO-style RL gains denser feedback from step-level credit assignment, with 54.7→60.7 pass@1 on Qwen3-0.6B — dejavucoder · 2026-07-26
- OpenCode shows a terminal-first AI coding workflow for fixing TODOs in a codebase — dSebastien · 2026-07-26
- Multi-model agent loops get cheaper when the planner is smart and the executor stays cheap — Fantastic-Act-8476 · 2026-07-26
- Open-source storyboard workflow uses Qwen 2511 Image Edit for consistent characters and scenes — wjc_5 · 2026-07-26
- Ethan Mollick’s new AI guide pushes users toward context, apps and agent loops — The AI Daily Brief · 2026-07-26