Optimized video workflow: 10s at 1MP in ~125s on an RTX 5090, with custom audio driving
Tokyo_Jab · reddit · 2026-09-11
A Reddit user shared their heavily optimized video generation workflow (downloadable via Google Drive), adding a custom audio option.
- Custom audio driving: external audio drives the generation, working well for lipsync and music videos.
- Performance: 125 seconds to generate 10 seconds of video at 1.0 megapixel on an RTX 5090; bumping steps from 6 to 7+ improves quality further.
- Options: the VSA part can be bypassed on quality issues at 15% extra generation time; block swapping supported for low-VRAM setups. The workflow has been cleaned up with notes for easier onboarding.
More from Multimodal
- I tested 11 video upscaling methods for AI video: local, Topaz and cloud compared — lopas99 · 2026-10-28
- AI Horde adds Moss-Image for free, no-account image generation on volunteer GPUs — Wanabeelee2 · 2026-10-28
- AI band Techhalla reimagines Slipknot's 'Wait and Bleed' as a 1971 soul performance — Mother-Pear-2392 · 2026-10-28
- Inside the Hailuo H3 Max retro-MV workflow: 15-second chunks, one reference image, multi-shot prompts — techhalla · 2026-10-28
- Open-source pipeline turns Seedance AI video into real PBR textures on your 3D mesh — OdinLovis · 2026-10-28
- AI-generated cyberpunk short 'VHS Sirens' shows off consistent characters and neon city detail — brythehuman · 2026-10-28