Open duckinchina opened 1 month ago
Is there a way to use 50 cameras to capture nerf data instead of having one camera update 50 times for each time step, which would cause the physical engine to exert greater force on the objects, especially on objects with insufficient friction like spheres?
Sorry for the late reply. I am also searching for an acceptable way to capture multi-view images...
"I found that in YARR.yarr.utils.video_utils.py, there is a scene.step() in NeRFTaskRocorder which updates the scene multiple times in a for loop. However, this causes the ball objects to roll on the table. I suspect that this is due to the high frequency of scene updates, amplifying the effect of certain random forces in the scene. However, if I comment out scene.step(), the images of the nerf camera view won't update, although the data can be generated without error. Is there any good solution to this?"