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
183 stars 3 forks source link

Question about the introduction video #8

Open emjay73 opened 3 weeks ago

emjay73 commented 3 weeks ago

Thank you for sharing the code.

From the introduction video, I noticed a clip featuring a cook, which doesn't seem to be part of the Kubric or Parallel datasets. Is that correct?

Is this video actually used for GCD training, or is it just for concept visualization?

image

basilevh commented 4 days ago

The example in the figure is from the Ego-Exo4D dataset by Meta. We did not use this dataset in our paper so it is merely for visualization. Since it contains many real-world multi-view video scenes, it hints at what future work could possibly and hopefully achieve ;-)