NVlabs / neuralangelo

Official implementation of "Neuralangelo: High-Fidelity Neural Surface Reconstruction" (CVPR 2023)
https://research.nvidia.com/labs/dir/neuralangelo/
Other
4.33k stars 387 forks source link

Images instead of videos #100

Open mowangmodi opened 1 year ago

mowangmodi commented 1 year ago

Can I use the already captured images and obtain JSON files without using videos

sunhoro commented 1 year ago

Also waiting for this. Using images instead of vidoes

xuduo18311199384 commented 1 year ago

Also waiting for this. If i can take 100 pictures of an object from different angles to reconstruct?

ninjamode commented 1 year ago

Also interested, but note that you can create a temporary video out of your images without encoding overhead, and then match the downsample rate when processing (set to 1 for the example below).

ffmpeg -framerate 1 -i 'myimages\frame_%05d.png' -c:v copy temp_video.mkv

Note that this requires the images to be named in the scheme frame_<5 digit number>.png, but can be adjusted.

bcherb2 commented 12 months ago

You can already use images, just run the preprocessing steps individually and skip the ffmpeg step.