caramel-jiao

I think this can be found in the Supplementary: https://openaccess.thecvf.com/content/CVPR2021/supplemental/Yu_pixelNeRF_Neural_Radiance_CVPR_2021_supplemental.pdf

https://github.com/facebookresearch/co3d/pull/61 You can refer to this issue, it may be cased by changes of different numpy versions.

No I haven't 😥. But I have tried to directly adopt my own rendered images with such textures for some pretraining, and found it also helpful for my tasks.

Perhaps using the data obtained by re-rendering according to the existing annotation can help alleviate this problem for you. I may not have any other better ideas either.😣

Sorry maybe I didn't explain clearly. It can be understood as following the guidelines of the current repo (https://github.com/hassony2/obman_render?tab=readme-ov-file#launch-) to generate obman dataset from scratch according to the original workflow....