Endless MiniMax H3: open ComfyUI workflow stitches unlimited-length lip-synced video on a 3060
embryo10 · reddit · 2026-09-29
A new v1.5 of the open-source 'Endless MiniMax H3' ComfyUI workflow combines ComfyUI-H3-Motion Context with a custom Clip Stitcher node that saves latents from each generation and stitches clips into unlimited-length videos with no visible seams. The author produced a 1.5-minute lip-synced video in 2 hours (with retries) on an RTX 3060 12GB. v1.5 adds project directories, a Purge-all-clips group, a fifth reference image, a simplified Audio 1 Forced Multi, and a refiner that fixes overexposed clips on long steady-camera videos. Supports T2V/I2V (FL2AV) and Ref2AV modes; per-clip retry and redo are supported. Both repos are on GitHub.
Related event: ComfyUI Workflows Enable Seamless Long Video Extension with MiniMax-H3(4 posts)→
More from Multimodal
- NVIDIA's LongLive-Plug: Distill Once, Deploy Training-Free Across 54 Downstream Video Models — nvidia · 2026-09-30
- Adobe Research Shows Adversarial Post-Training Restores Missing High-Frequency Detail in Pixel Diffusion — adobe-research · 2026-09-30
- UCSD's LIFT Lets You Control Future Video Layouts via On-Policy Self-Distillation — UCSanDiego · 2026-09-30
- MiniMax-H3 RefMod Upgrade: Packing JPEGs Into Safetensors Cuts Encoding Time and Kills Identity Bleed — acedelgado · 2026-09-30
- Creator Says Opus 5.5 Renders in 45 Mins What Used to Need Four-Figure Plugins — justin_hart · 2026-09-30
- Casey: AI video will inevitably get so good that critics must admit great works — johncoogan · 2026-09-30