ComfyUI fails to detect Wan2.2 text encoders despite correct folders
dreamyrhodes · reddit · 2026-09-28
A Reddit user downloaded all Wan2.2 models from ComfyUI's HF page, placed the VAE, checkpoints and umt5xxL text encoder in the right folders, and symlinked diffusionmodels for Forge Neo. But in the Wan2.2 template ComfyUI only finds the VAE — the CLIP node shows no text encoder dropdown at all, with 15 encoders installed and all node field types tried. The user is asking for help.
More from Multimodal
- Engram is an offline sampler that turns AI hallucinations into music — The Verge AI · 2026-09-28
- Turning a Dune sandworm into a robot concept using Claude Opus 5.5 — nptacek · 2026-09-28
- Slopus: free open-source desktop app generates and edits AI video locally with Minimax H3 — BittiAI · 2026-09-28
- MiniMax H3 shows severe banding in dark scenes, creator reports — itchplease · 2026-09-28
- Redditor builds a 5-part Anima guide after finding all existing tutorials "genuinely sucked" — shapic · 2026-09-28
- 10 AI films screened in cinema leave audience rethinking the 'AI slop' narrative — RachelVT42 · 2026-09-28