how to get each picture's latent code?

[alip7]

I set the 'return_latents=True' in psp.py , but get TypeError ：

【
Traceback (most recent call last):
File "/home/zhz1/anaconda3/envs/env_36/lib/python3.7/site-packages/PIL/Image.py", line 2751, in fromarray
mode, rawmode = _fromarray_typemap[typekey]
KeyError: ((1, 1, 1, 256), '|u1')

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "scripts/inference.py", line 136, in
run()
File "scripts/inference.py", line 82, in run
result = tensor2im(result_batch[i])
File "./utils/common.py", line 23, in tensor2im
return Image.fromarray(var.astype('uint8'))
File "/home/zhz1/anaconda3/envs/env_36/lib/python3.7/site-packages/PIL/Image.py", line 2753, in fromarray
raise TypeError("Cannot handle this data type: %s, %s" % typekey) from e
TypeError: Cannot handle this data type: (1, 1, 1, 256), |u1
】

could you please tell me how to output each picture's latent code?

[yuval-alaluf]

Did you only make a change in psp.py?
You most likely got the error because you did not make all the necessary changes needed.

You should make the changes in inference.py instead. You should change the call

pixel2style2pixel/scripts/inference.py

Line 111 in 751fcbd

result_batch = net(inputs, randomize_noise=False, resize=opts.resize_outputs)

in inference.py to return a tuple as such:

images, result_latents = net(...,return_latents=True) 

Then, in inference.py in line 129 you should change

pixel2style2pixel/scripts/inference.py

Line 129 in 751fcbd

return result_batch

to

return images, result_latents

Finally, in line 73, you should change the line

pixel2style2pixel/scripts/inference.py

Line 73 in 751fcbd

result_batch = run_on_batch(input_cuda, net, opts)

to something like

result_batch, result_latents = run_on_batch(input_cuda, net, opts)

Now you have the latents of the batch and can save them if you wish.
In short, returning latents is very much possible with some minimal changes to the inference.py file.
I may have missed a small step, but the above changes should assist you in getting started. If you still encounter issues feel free to reply.

[alip7]

Did you only make a change in psp.py?
You most likely got the error because you did not make all the necessary changes needed.

You should make the changes in inference.py instead. You should change the call

pixel2style2pixel/scripts/inference.py

Line 111 in 751fcbd

result_batch = net(inputs, randomize_noise=False, resize=opts.resize_outputs)

in inference.py to return a tuple as such:

images, result_latents = net(...,return_latents=True) 

Then, in inference.py in line 129 you should change

pixel2style2pixel/scripts/inference.py

Line 129 in 751fcbd

return result_batch

to

return images, result_latents

Finally, in line 73, you should change the line

pixel2style2pixel/scripts/inference.py

Line 73 in 751fcbd

result_batch = run_on_batch(input_cuda, net, opts)

to something like

result_batch, result_latents = run_on_batch(input_cuda, net, opts)

Now you have the latents of the batch and can save them if you wish.
In short, returning latents is very much possible with some minimal changes to the inference.py file.
I may have missed a small step, but the above changes should assist you in getting started. If you still encounter issues feel free to reply.

Thanks for your instruction and your great work! It helps a lot.
now I get the (18,512) of latent code for each picture.

[yuval-alaluf]

Awesome!