autonomousvision / unisurf

[ICCV'21] UNISURF: Unifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstruction
MIT License
417 stars 35 forks source link

How to train on custom data? #5

Open AugustasMacijauskas opened 2 years ago

AugustasMacijauskas commented 2 years ago

Great work, thank you for sharing the code!

I would like to try it on custom data for which I acquired camera poses in the form of the poses_bounds.npy file by running img2poses.py from LLFF. How could I use your model with this data?

WillKen commented 2 years ago

@AugustasMacijauskas You can see the IDR project. The data in UNISURF is adapted from IDR. See this for more information.

silence401 commented 2 years ago

Do you find the solution of run on own data, the idr data format seems different from this

CozmoxJC commented 2 years ago

@AugustasMacijauskas Did you manage to use the model on custom data? I also acquired camera poses in the form of the poses_bounds.npy file by running img2poses.py from LLFF.