零存储危机隐私保护：您上传的证据影像仅在内存中进行即时神经分析，完成后立即彻底销毁。无日志、无云端备份。
Open-Weights Video Model威胁严重程度: 中度威胁
Mochi 1 Video AI Detector: Genmo Open Video Model Forensics
Forensic evaluation of Genmo’s open-weights 10B parameter diffusion video generation model.
使用 Sealed Rose 验证媒体真实性
实时神经网络检测 · 免费 · 无需注册 · 零存储 · 内存瞬时处理
Mochi 1 is an open-source video generation model from Genmo with 10 billion parameters. It produces high-motion video from text prompts, but exhibits distinctive fluid dynamics inconsistencies and unnatural physical inertia.
关键警讯与破绽特征
1Fluid Dynamics and Smoke Anomalies
Liquids, clouds, and hair morph smoothly but disobey real-world physical gravity.
Diffusion Video Architecture
- Asymmetric Diffusion Transformer (AsymmDiT): Emits subtle checkerboard artifacts in high-contrast textures.
防范措施与应对步骤
步骤 1
Verify Clip on Sealed Rose Video Scanner
Inspect frame-by-frame temporal consistency.
常见问题
Is Mochi 1 open source?▼
Yes. Mochi 1 is an open-weights model available for local deployment and fine-tuning.
支持 23 种语言：