search

JiahuiLei / MoSca

MIT License
116 stars 0 forks source link

videos with fixed viewing angle #5

Open DingChunQ opened 2 months ago

DingChunQ commented 2 months ago

Great work!

Is your method applicable to monocular videos with fixed viewpoint? Thanks.

JiahuiLei commented 2 months ago

thanks for your interest.

Definitely yes, although we don't show it in the paper, we haved tried many static camera videos (the camera does not move, only scene content moves), and Mosca also works for these kinds of videos.

DingChunQ commented 2 months ago

thanks for your interest.

Definitely yes, although we don't show it in the paper, we haved tried many static camera videos (the camera does not move, only scene content moves), and Mosca also works for these kinds of videos.

Thank you very much for your reply! Some previous works seem to have poor results for video datasets where the camera does not move. I don't know much about nerf and 3dgs. Could you explain in an intuitive and profound way why it is more difficult to process static camera videos than videos with moving cameras for this task of obtaining dynamic 3d representations from monocular videos? In addition, why can Mosca be applied to static camera videos (compared to previous work)? What modules or steps are mainly at work?