New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
implicitron:"ValidationError: Unexpected object type: _empty"---video synthesis #1314
Comments
I think we need to add |
There are a few fixes required to visualize_reconstruction and I think you have already made some. |
Does it help to change visdom_server to |
That did work. Thank you!
That seems complex to fix.
Yes, there is a few bugs in this py file. But I have made only one or two fixes in |
I am working to fix this all soon. For that error, you need to replace |
Yes!I forgot that!The name of this args has the same error that I mentioned in #1304、in configs and your markdown file. Thanks! (^ω^) |
Hello. I have run your fixed files. They killed many bugs, thanks. |
This may be a simple code problem. Have you updated with the fixes from 6e25fe8 (from a few hours ago)? In that I changed |
Great!I solved that error with your help. Thank you.
I added this line in Then it seems worse:
|
Have you got the whole of the changes in 6e25fe8? The SingleSceneDataset change in that commit should be a proper way to avoid the first error. What version of OmegaConf do you have by the way? If you /do/ need to diagnose the second error, you can add |
Thats my fault. I updated the whole changes. The error disappeared. But video still can not be written in
The second error :(I am searching for reason now)
This is my omegaconf version: |
Your version of ffmpeg may be an odd one. If you are running in a conda environment, I think you can get a local one in your environment with |
Ok, I will try it. By the way, is the new video synthesis module just fit for JsonindexData? I can not get good render images as I got in train_loop. |
What do you mean by the "new video synthesis module"? Do you mean evaluate_new_view_synthesis? If so, no, that should be useful for evaluating any implicitron model. If it is giving poor results with a certain dataset, that should mean that the training did not do well. |
No, I refer to |
I think I have got my answer to these questions. I will close this issue. |
🐛 Hello, there is something wrong with video making in implicitron.
INPUT
python pytorch3d/implicitron_trainer/visualize_reconstruction.py exp_dir='pytorch3d/projects/implicitron_trainer/checkpoint' \ visdom_show_preds=True visdom_server=True n_eval_cameras=40 render_size="[64,64]" video_size="[256,256]"
OUTPUT ERROR
raise ValidationError(f"Unexpected object type: {type_str(type_)}") omegaconf.errors.ValidationError: Unexpected object type: _empty full_key: visdom_server object_type=_export_scenes_default_args_
I read many relative py files.
visdom_server
is defined asstr
. I have no idea how the error comes.THANK U
The text was updated successfully, but these errors were encountered: