What's the render_poses meaning?
wtishere opened this issue · 3 comments
wtishere commented
Thanks for your great work! When I ran training.py by using the example_sequence data, I found the loaded data are from precomputed.json, which you provide. I'd like to train other datasets, so could you tell how to compute the precomputed.json? By the way, from precomputed.json, what's the render_poses meaning? And what's the difference between poses and render_poses? Thank you so much!
edgar-tr commented
Hi,please follow the optional installation instructions about replacing the included llff files with the original ones: https://github.com/facebookresearch/nonrigid_nerf#installationAfter that, preprocessing estimates the camera parameters: https://github.com/facebookresearch/nonrigid_nerf#preprocessrender_poses are novel poses for test-time rendering, while poses are the poses for the training images.
edgar-tr commented
Hi, please follow the optional installation instructions about replacing the included llff files with the original ones: https://github.com/facebookresearch/nonrigid_nerf#installation After that, preprocessing estimates the camera parameters: https://github.com/facebookresearch/nonrigid_nerf#preprocess render_poses are novel poses for test-time rendering, while poses are the poses for the training images.
wtishere commented
Thank you so much.