Locally hosted H3 video model renders 25 seconds of 1942 Shanghai
CurieuxExplorer · x · 2026-09-22
CurieuxExplorer shares a locally deployed H3 video model output: a 25-second cinematic clip of 1942 Shanghai, praised as resembling an Ang Lee-style film dialogue scene rather than AI content. The author notes 25 seconds is a conservative setting for quality, not the local generation limit, and has shared the prompt used.
More from Multimodal
- ElevenLabs launches Studio 4.0: AI-native video editor with frame-level control — nikola_mr64990 · 2026-09-22
- Pexo tested: direct, review and refine an entire AI video through conversation — nikola_mr64990 · 2026-09-22
- BytePlus adds Draft Mode to Dreamina Seedance 2.5: 480p drafts, 1080p finals — DavidmComfort · 2026-09-22
- Grok Imagine demo shows 30-second continuous AI video generation — tetsuoai · 2026-09-22
- Pexo's Mark to Fix lets you revise targeted parts of a video frame — alifcoder · 2026-09-22
- Wyrdoom: an AI-generated gothic horror short film — EphemeralReverie_ · 2026-09-22