long-video 开源项目推荐（32 个）

SkyworkAI /Matrix-Game

Matrix-Game 3.0：具有长期记忆的实时和流式交互世界模型

j955229 /ComfyUI-MiniMax-H3-Motion-Director

Independent multi-segment MiniMax H3 Motion Director for ComfyUI

mll-lab-nu /TStar

用于长视频理解的 T* 关键帧搜索（CVPR 2025）+ LV-Haystack 时序搜索基准代码

Linketic /TC-Light

[NeurIPS`25] TC-Light: Temporally Coherent Generative Rendering for Realistic World Transfer

ZenHG /ComfyUI-H3-Latent-Relay

ComfyUI nodes for MiniMax-H3 multi-segment video: seamless continuation via a latent bridge (previous segment's AV latent pinned into the next segment's conditioning) — zero re-encoding. | MiniMax-H3 多段续接 latent 桥：上一段 AV latent 钉进本段 conditioning，零重编码，长视频不漂。

yeahdongcn /nautilus-studio

AI-native long-form video studio with hidden storyboard orchestration and pluggable generation backends.

1113800942 /comfy-h3md-director

comfyUI nodes for long MiniMax H3 video: 2+ minutes at 1080p in one run, multi-shot Director, motion copy, re-render only the bad segments, no long-video color drift.

ooizj /ComfyUI-H3-Long-Video

Generate long MiniMax H3 videos in ComfyUI from one prompt and a total duration: segments continue motion and audio across joins, with optional LLM prompt splitting and a Ref2VA prompt builder.

hany-liu /comfy-director

ComfyUI 本地长视频导演台：一句话故事 → 逐段生成审核 → 拼接成片（MiniMax H3）

lichen1015 /tethermem

Official implementation and project page for TetherMem: training-free query-aware memory routing for long-horizon autoregressive video generation.

NitcanKen /Codex-native-long-video-memory

Local-first Codex plugin that indexes long videos with OCR/ASR and answers with timestamped, SHA-verified visual evidence.

isno0907 /SOSampler

Official code for "Scalable Frame Sampling for Video Classification: A Semi-Optimal Policy Approach with Reduced Search Space"

FATFATHAO /MOSS-ChunkTranscribe

MOSS长视频切块转写 + Web说话人对轴编辑器 | Chunked long-video transcription (mtd-chunk) + web speaker-alignment editor for MOSS-Transcribe-Diarize

Shidu-Ren /ICM-Bench

Official repository for ICM-Bench: identity-centric reasoning over long video memories. Benchmark, video generation, and evaluation code.

mjzcng /ComfyUI-MiniMaxH3-Disk-Relay

Prompt-boundary disk relay, review, resume, and assembly for memory-constrained MiniMax H3 long-video chains in ComfyUI

xt-mahh /h3chain

基于 MiniMax H3 的真正连贯的无限长视频生成——不是把一堆片段拼起来，而是每一段都从前一段的运动上下文 latent 中延续出来：人物、动作、镜头、光影自然延续，看不到接缝。

minar09 /steady-forcing

Steady-Forcing: Balancing Spatial Persistence and Motion Continuity in Long-Horizon Nature Video Diffusion

a86582751 /doubao-seedance-video-skill

Natural-language long-video creation for Codex: plan, generate, review, regenerate, create audio, edit with FFmpeg, and cost-report Doubao Seedance videos.

FURUYAN1234 /comfyui-long-video-upscaler

Memory-safe sequential long-video upscaling with audio preservation for ComfyUI / 長尺動画を逐次処理・音声維持で高解像度化

Chengzhi-W /LongVideo

Long-history autoregressive video generation with Memory-DMD and UltraVideo-Long AR-teacher fine-tuning