Open ietres opened 3 weeks ago
What matters is the number of frames of the resulting depth map.
I haven't tried, but if you set --target-fps -1
, you will get the same length of video.
parser.add_argument(
"--target-fps", type=int, default=15, help="Target FPS for the output video"
) # -1 for original fps
It's strange, the default configuration would subsample the input video at 15 FPS and process the video up to 195 frames, you can tune the "--process-length" arguments for the maximum length, but this also may lead to a longer process time
I think the real limitation might be your GPU video memory
I have a 27-second video, but the output gives me a 7-second depth map. What am I doing wrong? Is there any limitation on the input regarding the length of frames or seconds?