Hi,
when I try to train dtu data without prior, I found that the result of bakedangelo is bad, even worse than volsdf. I want to know whether this data (without prior) is not good enough, or there is something wrong with my training command.
My bakedangelo training command is ns-train bakedangelo --machine.num-gpus 1 --pipeline.model.level-init 8 --trainer.steps-per-eval-image 5000 --pipeline.datamanager.train-num-rays-per-batch 2048 --pipeline.datamanager.eval-num-rays-per-batch 512 --pipeline.model.sdf-field.use-appearance-embedding True --trainer.save-only-latest-checkpoint False --pipeline.model.background-color white --pipeline.model.sdf-field.bias 0.1 --pipeline.model.sdf-field.inside-outside True --pipeline.model.background-model grid --vis wandb --experiment-name scan55_bakedangelo sdfstudio-data --data data/dtu/scan55,
and the results is
Its depth and normal results are bad.
My volsdf training command is volsdf --machine.num-gpus 1 --pipeline.model.background-color white --pipeline.model.sdf-field.inside-outside True --trainer.load-dir outputs/scan55-volsdf/volsdf/2024-03-16_115658/sdfstudio_models --trainer.max-num-iterations 100001 --vis wandb --experiment-name scan55-volsdf sdfstudio-data --data data/dtu/scan55,
and the results is
BTW, the meta_data.json of this dtu data(scan55) is
Hi,
when I try to train dtu data without prior, I found that the result of bakedangelo is bad, even worse than volsdf. I want to know whether this data (without prior) is not good enough, or there is something wrong with my training command.
My bakedangelo training command is
ns-train bakedangelo --machine.num-gpus 1 --pipeline.model.level-init 8 --trainer.steps-per-eval-image 5000 --pipeline.datamanager.train-num-rays-per-batch 2048 --pipeline.datamanager.eval-num-rays-per-batch 512 --pipeline.model.sdf-field.use-appearance-embedding True --trainer.save-only-latest-checkpoint False --pipeline.model.background-color white --pipeline.model.sdf-field.bias 0.1 --pipeline.model.sdf-field.inside-outside True --pipeline.model.background-model grid --vis wandb --experiment-name scan55_bakedangelo sdfstudio-data --data data/dtu/scan55,and the results is
Its depth and normal results are bad.
My volsdf training command is
volsdf --machine.num-gpus 1 --pipeline.model.background-color white --pipeline.model.sdf-field.inside-outside True --trainer.load-dir outputs/scan55-volsdf/volsdf/2024-03-16_115658/sdfstudio_models --trainer.max-num-iterations 100001 --vis wandb --experiment-name scan55-volsdf sdfstudio-data --data data/dtu/scan55,and the results is
BTW, the meta_data.json of this dtu data(scan55) is
