Dev builds automated v2v pipeline that applies chosen characters with zero prompting
mrjonfinger · x · 2026-10-02
Developer mrjonfinger shared an automated video-to-video system he built: a computer or phone automatically captures footage and applies his selected characters via v2v, with no prompting or re-rolling required. He reports it's working reliably, calling the approach "more play, less prompt." The setup uses DreamLab (DreamLabLA), with a demo video attached.
More from Multimodal
- AITuber without Live2D: Opus builds 3D stage, Minimax h3 and TTS fill the rest — Hailuo_AI · 2026-10-02
- ComfyUI demo shows MiniMax H3 character-swap LoRA holding up under camera moves — Hailuo_AI · 2026-10-02
- Creator Used Comfy Agent to Automate Repetitive Work in Making AI Artwork YUI — Merzmensch · 2026-10-02
- Prompt template: 'Temporal Velocity' images show time passing via blurred lines and distorted forms — LudovicCreator · 2026-10-02
- "Hey, Nugget!" — an AI animated short film shared on Reddit — Violinist_Mental · 2026-10-02
- Krea update reportedly speeds up workflows by ~70%, slashing VAE decode times — MannY_SJ · 2026-10-02