search

basilevh / gcd

Generative Camera Dolly: Extreme Monocular Dynamic Novel View Synthesis (ECCV 2024 Oral) - Official Implementation
https://gcd.cs.columbia.edu/
GNU General Public License v3.0
161 stars 3 forks source link

Single image novel-view generation #5

Open anithselva opened 1 week ago

anithselva commented 1 week ago

Hi there,

Neat work!

I had a question if this codebase would support generating a novel view using a single image?

basilevh commented 5 days ago

While GCD is designed for dynamic rather than static scenes, yes it does! I think it works out of the box (the script should duplicate frames as needed to turn an input image into a video), but if not it should be relatively easy to repeat that preprocessing step yourself using ffmpeg for example. Hope this helps!