a16z Video Watch posted an update
How Real-Time AI Video Is Changing How Creators Work
Why it mattersa16z GP Jennifer Li talks with fal's Gorkem Yurtseven and Batuhan Taskaya about H3 Max, their post-trained version of MiniMax's open-weight video model, and how combining post-training with systems and hardware optimization cut generation time enough to run continuous video streams that remember earlier scenes and respond to new directions mid-stream. They argue the next bottleneck is less speed than control – camera movement, lighting, characters, motion and lip sync – before artists and studios get predictable professional tools.
Discuss: If real-time video models can remember previous scenes and take directions mid-stream, does control – camera angles, lighting, lip sync – become a harder problem than generation speed for professional studios?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.