Examples don`t work correctly

[Kupchanski]

I tried to test inversion.py example, but got this mistake.

It seems like it just need to be corrected a little.
TY for pytorch realization btw! Love it

[ShenYujun]

Please try

cd idinvert_pytorch
python invert.py

We use relative path to find the images.

[Kupchanski]

Please try

cd idinvert_pytorch
python invert.py

We use relative path to find the images.

Thanks, It works now!

Another question. If I need this alg for 1024 res, I need to train encoder from scratch?

[ShenYujun]

Yes.

[ak9250]

@ShenYujun i am getting same issue when doing diffusion in colab
Context ID: 67% 2/3 [00:12<00:06, 6.47s/it]Traceback (most recent call last):
File "diffuse.py", line 146, in
main()
File "diffuse.py", line 131, in main
num_viz=args.num_results)
File "/content/idinvert_pytorch/utils/inverter.py", line 313, in easy_diffuse
self.preprocess(context),
File "/content/idinvert_pytorch/utils/inverter.py", line 105, in preprocess
raise ValueError(f'Input image should be with type numpy.ndarray!')
ValueError: Input image should be with type numpy.ndarray!

[ShenYujun]

How do you load the images and how do you feed them to the function easy_diffuse()?