3 ms·
One can imagine that resolving this will require a separate model to construct a timeline/keyframe series from a text description then interpolate between them.
by kitx 4y ago
One can imagine that resolving this will require a separate model to construct a timeline/keyframe series from a text description then interpolate between them.
Fascinating times, either way this is allegedly being open–sourced, so I'm hopeful others will be able to build on top.
- lairv 4y agoI wouldn't have thought we'd have these text-to-video models before the end of this year, so we might actually not be that far from minutes-long video generation