How to build 40+ second continuous talking videos with LTX 2.5 in ComfyUI
Majestic-Regret-3030 · reddit · 2026-09-10
A user running ComfyUI locally (RTX 3060 12GB + 32GB RAM) asks how to produce 40-60 second fixed-camera talking-character videos with LTX 2.5.
Key questions:
- Can LTX 2.5 generate 40+ seconds continuously, or is it better to generate 8-10 second segments and extend from final frames
- How to keep face, clothes, background, camera and motion consistent across segments — interested in overlapping frames, first/last frame conditioning, video extension and reference frames to hide transitions
- Speech and lip sync: generate full Slovak speech first, then either run LTX 2.5 for motion plus MuseTalk/LatentSync afterward, or use an audio-driven LTX 2.5 workflow directly
- Vertical 9:16 output may need lower resolution, offloading, chunking and longer render times
He explicitly wants battle-tested workflows built around LTX 2.5 (nodes, segment length, overlap, speech/lip-sync tools), not a generic list of talking-head models.
More from coding & agent
- AI agents don't know their code makes codebases harder to work with — _Stocko_ · 2026-09-10
- Hy4 preview tested: playable 3D survival game from a single prompt in WorkBuddy — mhdfaran · 2026-09-10
- Astra xhigh's token efficiency may stem from its subagent instructions — Al_Grigor · 2026-09-10
- Paper unveils preliminary Agent Swarm RL recipe; V4.1 teams hit 30% on ProgramBench — inductionheads · 2026-09-10
- AI-generated code maintenance costs up 300% in 18 months as firms 'let the agents cook' — bendee983 · 2026-09-10
- DeepSeek V4.1 Flash vs GPT-6 Astra: same coding task, wildly different code structures — kevinkern · 2026-09-10