Hi there,
Thank for your amazing work!
I tried to reproduce the result on broom case with default setting on 8 gpus.
trained with
"python train.py --jax_data_dir=$PATH_TO_DATA/broom --jax_base_folder=$WORK_DIR --jax_gin_configs=$PATH_TO_PROJ/nerfies/configs/gpu_vrig_paper.gin"
& evaled with
"python eval.py --jax_data_dir=$PATH_TO_DATA/broom --jax_base_folder=$WORK_DIR --jax_gin_configs=$PATH_TO_PROJ/nerfies/configs/gpu_vrig_paper.gin"
The avg psnr is about ~ 19 - 20 which is similar to the one in paper.
But got undesirable outputs like:
which is quite different with GT/paper result:
Could you help me to figure out the possible reasons? Many thanks!
Hi there, Thank for your amazing work! I tried to reproduce the result on broom case with default setting on 8 gpus. trained with "python train.py --jax_data_dir=$PATH_TO_DATA/broom --jax_base_folder=$WORK_DIR --jax_gin_configs=$PATH_TO_PROJ/nerfies/configs/gpu_vrig_paper.gin" & evaled with "python eval.py --jax_data_dir=$PATH_TO_DATA/broom --jax_base_folder=$WORK_DIR --jax_gin_configs=$PATH_TO_PROJ/nerfies/configs/gpu_vrig_paper.gin" The avg psnr is about ~ 19 - 20 which is similar to the one in paper. But got undesirable outputs like:
which is quite different with GT/paper result:
Could you help me to figure out the possible reasons? Many thanks!