Windows support?

[lucasjinreal]

Hello, tinycudann are really hard to install on windows, at least I failed on cuda11.8 + windows. Would consider add vanilla pytorch in modeling so that users can use without tinycudann?

[ken-ouyang]

Hi. You may disable the option deform_hash and vid_hash for training if you do not have tiny-cuda. We implement a version with vanilla pytorch using positional encoding. But the training time will greatly increase and thus it is not recommended.

[lucasjinreal]

inference also not support

[ken-ouyang]

In that case, for inference, you may also disable these two options. It should work.

[lucasjinreal]

disable deform_hash and vid_hash will effect result performance?

[ken-ouyang]

Certainly, the type of motion present within a video sequence can impact the performance of the model to varying degrees. Specifically, if the video primarily contains rigid motion, you can expect the performance to remain relatively consistent.

[justinjohn0306]

How do I switch to gloo backend?

[justinjohn0306]

Done.

[ataraxyz]

How did you switch? Can you post a how to?

[yuanjiacheng]

How did you switch? Can you post a how to?

replace code strategy="ddp_find_unused_parameters_true" in train.py line 547 by from pytorch_lightning.strategies import DDPStrategy ddp_gloo = DDPStrategy(process_group_backend="gloo",find_unused_parameters=True) trainer = Trainer(max_steps=hparams.num_steps, precision=16 if hparams.vid_hash == True else 32, callbacks=[checkpoint_callback], logger=logger, accelerator='gpu', devices=hparams.gpus, num_sanity_val_steps=1, benchmark=True, profiler="simple" if len(hparams.gpus) == 1 else None, val_check_interval=hparams.valid_iters, limit_val_batches=hparams.valid_batches, strategy=ddp_gloo)