ComfyUI Adds Native Support for Wan Animate 2 with Real-Time Latency
petewoodbridge · x · 2026-08-11
ComfyUI announced native support for the Wan Animate 2 video generation model. This version feeds the driving video directly into the transformer without an intermediate motion extractor, fully preserving subtle expressions and finger positions.
The update improves motion fluidity and identity consistency while enabling text-driven camera viewpoint control decoupled from the original driving clip. Additionally, the Wan Animate 2 Lite version achieves real-time generation latency.
More from Multimodal
- PixVerse Generates Stunning Logo Animations from Prompt and Image — aziz4ai · 2026-08-11
- AI Music Enters New Phase: Artistic Research Now Trumps Technical Research — jordiponsdotme · 2026-08-11
- Gemini vs. OpenAI: Image Generation Now Accurately One-Shots Rabbit R1 — SimonBalmain · 2026-08-11
- AI Video Ad Demo: Promoting a Sports Watch That Doesn't Exist — DeskZenZone · 2026-08-11
- Minimax H3 Video Model Leaks: Showcasing ref2va Feature — beatlepol · 2026-08-11
- AI Video Model Battle: Seedance 2.5 Leads, 2.0 Fast Offers Best Value in Hands-On Test — 新智元 · 2026-08-11