facebookresearch / one_shot_3d_photography Goto Github PK

Hi,
This is definitely an awesome work and thank you for your initiative to share the source code. In the readme, you refer the impainting part to context-aware paper, but if I am not wrong, in the one shot 3d photography, the impainting is based on partial convolution. Are they the same when they are implemented in the facebook 3d images?

Videos like those showcased in the project page.

Hi! Thanks for your outstanding paper.
Do you have any plans to release the initial PyTorch code for the Tiefenrausch model in addition to the converted caffe2 weights?

I paste:

conda create -yn one_shot python=3.7
conda activate one_shot
conda install --file requirements.txt -c conda-forge -c pytorch

But it gives: /bin/bash: conda: command not found

How should i use it?

Hi,
I was trying to create the video file using the numpy output.
But, the shape of the output list is [1,384,288,1]

I am not able to understand why it contains only a single one channel image.

Would be really grateful if you could help. Thanks

Is their is a directandroid Implimintaion libirary or an API?
If not how to use it ?

Is there no way to inference in this code?

The inference I'm talking about will be provided in the form of a video after in-painting is completed.

Thank you for your work.