My mid_results images are all black
zhenqis123 opened this issue · comments
After running this command "python dynamic_offsets_runner.py --root_dir $data_dir --name $subject_name --device_id $device_id", some predicted images are generated in the folder "HF-Avatar/results/diff_optiming/Body2D_2040_499_mid_results". But all the image_xxxx_pred.png and sil_xxxx_pred.png are all black. Did anybody encounter the same problem?
After running this command "python dynamic_offsets_runner.py --root_dir $data_dir --name $subject_name --device_id $device_id", some predicted images are generated in the folder "HF-Avatar/results/diff_optiming/Body2D_2040_499_mid_results". But all the image_xxxx_pred.png and sil_xxxx_pred.png are all black. Did anybody encounter the same problem?
I've solved the problem. The 'in_ndc' parameter should be set to 'False' when defining the camera. Because in the new version of pytorch3d, the 'in_ndc' parameter is set to be 'True' by default. So I guess it's because of the version change of pytorch3d.