Meituan open-sources LongCat-Video-Avatar 1.6 for 10+ minute talking-avatar videos
pmttyji · reddit · 2026-11-15
Meituan's LongCat team released LongCat-Video-Avatar 1.6, an upgrade to its open-source audio-driven human video generation framework.
Key improvements:
- More natural lip animation: reduces exaggerated mouth movements, overly red lip coloring, and temporal instability in the mouth region;
- Stable long-form generation: mitigates error accumulation in autoregressive video continuation, enabling videos over 10 minutes while preserving visual quality, identity consistency, and coherence;
- Faster inference: end-to-end pipeline and execution optimizations for more efficient deployment.
A full technical report is available; weights are open on Hugging Face (meituan-longcat/LongCat-Video-Avatar-1.6).
More from Multimodal
- Seedream 5.5 tested: fashion shots showcase impressive quality, prompts included — LudovicCreator · 2026-11-15
- Early Seedream 5.5 tests impress: fashion-grade portrait generations with shared prompts — LudovicCreator · 2026-11-15
- Kimodo runs text-to-3D-motion locally on iPad at 30fps, clips up to 10 seconds — Swaggsz · 2026-11-15
- speech.cpp v0.11.0 lands Haru-4B, a full-duplex voice chat model running locally in C++ — alexcovo_eth · 2026-11-15
- Everyone uses Fable 5.2 to code, almost nobody for UGC ads — _jaydeepkarale · 2026-11-15
- Change one thing at a time: why iterating AI images isn't design improvement — alifcoder · 2026-11-15