Nolasaurus commented 1 year ago

I'm trying to run Real-ESRGAN with Python 3.10.8 on Ubuntu 22.04.1 through WSL2.

I followed the instructions found here:

When I run the code snippets wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.0/RealESRGAN_x4plus.pth -P weights

python inference_realesrgan.py -n RealESRGAN_x4plus -i inputs --face_enhance

I get the following error:

(test_esrgan) nolan@NolanDesktop:~/Python/Real-ESRGAN$ python inference_realesrgan.py -n RealESRGAN_x4plus -i inputs --face_enhance
/home/nolan/miniconda3/envs/test_esrgan/lib/python3.10/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/home/nolan/miniconda3/envs/test_esrgan/lib/python3.10/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=None`.
  warnings.warn(msg)
Testing 0 00003
Testing 1 00017_gray
Traceback (most recent call last):
  File "/home/nolan/Python/Real-ESRGAN/inference_realesrgan.py", line 166, in <module>
    main()
  File "/home/nolan/Python/Real-ESRGAN/inference_realesrgan.py", line 145, in main
    _, _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True)
  File "/home/nolan/miniconda3/envs/test_esrgan/lib/python3.10/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "/home/nolan/miniconda3/envs/test_esrgan/lib/python3.10/site-packages/gfpgan/utils.py", line 145, in enhance
    restored_img = self.face_helper.paste_faces_to_input_image(upsample_img=bg_img)
  File "/home/nolan/miniconda3/envs/test_esrgan/lib/python3.10/site-packages/facexlib/utils/face_restoration_helper.py", line 355, in paste_faces_to_input_image
    upsample_img = inv_soft_mask * pasted_face + (1 - inv_soft_mask) * upsample_img
ValueError: operands could not be broadcast together with shapes (1404,2000,1) (1404,2000)

error message.txt

Nolasaurus commented 1 year ago

When I follow the code a little further down the page ...

wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.2.4/RealESRGAN_x4plus_anime_6B.pth -P weights

python inference_realesrgan.py -n RealESRGAN_x4plus_anime_6B -i inputs

I get this error instead:

python inference_realesrgan.py -n RealESRGAN_x4plus_anime_6B -i inputs
Testing 0 00003
Testing 1 00017_gray
Testing 2 0014
Testing 3 0030
Testing 4 ADE_val_00000114
Testing 5 OST_009
Testing 6 children-alpha
Testing 7 tree_alpha_16bit
        Input is a 16-bit image
Testing 8 video
Traceback (most recent call last):
  File "/home/nolan/Python/Real-ESRGAN/inference_realesrgan.py", line 166, in <module>
    main()
  File "/home/nolan/Python/Real-ESRGAN/inference_realesrgan.py", line 138, in main
    if len(img.shape) == 3 and img.shape[2] == 4:
AttributeError: 'NoneType' object has no attribute 'shape'

Nolasaurus commented 1 year ago

I don't know what the original issue was, but I reinstalled from scratch and it is now working.

Nolasaurus commented 1 year ago

Never mind, ran into the same error

Nolasaurus commented 1 year ago

I did not have the .pth files in the \Real-ESRGAN directory and now I no longer get the first error, only the one ending in AttributeError: 'NoneType' object has no attribute 'shape'

abhijiticelero commented 1 year ago

I got this error with 00017_gray.png (https://github.com/xinntao/Real-ESRGAN/blob/master/inputs/00017_gray.png)

$ python3 inference_realesrgan.py -n RealESRGAN_x4plus -i inputs --face_enhance

Traceback (most recent call last): File "/home/ubuntu/Real-ESRGAN/inference_realesrgan.py", line 166, in main() File "/home/ubuntu/Real-ESRGAN/inferencerealesrgan.py", line 145, in main , _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True) File "/home/ubuntu/.local/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context return func(*args, *kwargs) File "/home/ubuntu/.local/lib/python3.10/site-packages/gfpgan/utils.py", line 145, in enhance restored_img = self.face_helper.paste_faces_to_input_image(upsample_img=bg_img) File "/home/ubuntu/.local/lib/python3.10/site-packages/facexlib/utils/face_restoration_helper.py", line 355, in paste_faces_to_input_image upsample_img = inv_soft_mask pasted_face + (1 - inv_soft_mask) * upsample_img ValueError: operands could not be broadcast together with shapes (351,500,1) (351,500)

Eliya-Shalom commented 1 year ago

I'm getting the same error with the example input file 00017_gray.png. It only occurs with png file extension, and face_enhance arg.

Python 3.11 Windows OS I've tried with other models, but failed with the same error RealESRGAN_x4plus | RealESRNet_x4plus | RealESRGAN_x2plus | realesr-general-x4v3

Working well with jpg file extension.

`Testing 0 00017_gray Traceback (most recent call last): File "C:\AI\Real-ESRGAN\inference_realesrgan.py", line 167, in main() File "C:\AI\Real-ESRGAN\inferencerealesrgan.py", line 146, in main , _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\The Beast\anaconda3\Lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context return func(*args, *kwargs) ^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\xxx\anaconda3\Lib\site-packages\gfpgan\utils.py", line 145, in enhance restored_img = self.face_helper.paste_faces_to_input_image(upsample_img=bg_img) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\xxx\anaconda3\Lib\site-packages\facexlib\utils\face_restoration_helper.py", line 355, in paste_faces_to_input_image upsample_img = inv_soft_mask pasted_face + (1 - inv_soft_mask) * upsample_img


ValueError: operands could not be broadcast together with shapes (1404,2000,1) (1404,2000)`

AlionSSS commented 1 year ago

Here is a solution

Analyze

I've compared the code for projects GFPGAN 1.3.8 and Real-ESRGAN v0.3.0
I found that the way to read the image is different in the code.
- GFPGAN 1.3.8: inference_gfpgan.py 137
```
  img = cv2.imread(path, cv2.IMREAD_COLOR)
```
- Real-ESRGAN v0.3.0: inference_realesrgan.py 131
```
  input_img = cv2.imread(img_path, cv2.IMREAD_UNCHANGED)
```
  Reason
- If you use images in PNG format, IMREAD_UNCHANGED parameter will retain the alpha channel.
- Therefore, this will result in a mismatch of the channel numbers in the image.
  Solution
One way to do this is to modify the format of your image to JPG.

Another way

We can change the code(Real-ESRGAN) to this:

    input_img = cv2.imread(img_path, cv2.IMREAD_COLOR)

Now, you can use the command with --face_enhance. For example:

python inference_realesrgan.py -n RealESRGAN_x4plus -i D:\code\Real-ESRGAN\inputs\00017_gray.png -o D:\code\Real-ESRGAN\inputs\ --face_enhance

Example

00017_gray_ffixed

AlionSSS commented 1 year ago

I'm getting the same error with the example input file 00017_gray.png. It only occurs with png file extension, and face_enhance arg.

Python 3.11 Windows OS I've tried with other models, but failed with the same error RealESRGAN_x4plus | RealESRNet_x4plus | RealESRGAN_x2plus | realesr-general-x4v3

Working well with jpg file extension.

Testing 0 00017_gray Traceback (most recent call last): File "C:\AI\Real-ESRGAN\inference_realesrgan.py", line 167, in <module> main() File "C:\AI\Real-ESRGAN\inference_realesrgan.py", line 146, in main _, _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\The Beast\anaconda3\Lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context return func(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\xxx\anaconda3\Lib\site-packages\gfpgan\utils.py", line 145, in enhance restored_img = self.face_helper.paste_faces_to_input_image(upsample_img=bg_img) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\xxx\anaconda3\Lib\site-packages\facexlib\utils\face_restoration_helper.py", line 355, in paste_faces_to_input_image upsample_img = inv_soft_mask * pasted_face + (1 - inv_soft_mask) * upsample_img ~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~ ValueError: operands could not be broadcast together with shapes (1404,2000,1) (1404,2000)

I'm getting the same error with the example input file 00017_gray.png. It only occurs with png file extension, and face_enhance arg.

Python 3.11 Windows OS I've tried with other models, but failed with the same error RealESRGAN_x4plus | RealESRNet_x4plus | RealESRGAN_x2plus | realesr-general-x4v3

Working well with jpg file extension.

Testing 0 00017_gray Traceback (most recent call last): File "C:\AI\Real-ESRGAN\inference_realesrgan.py", line 167, in <module> main() File "C:\AI\Real-ESRGAN\inference_realesrgan.py", line 146, in main _, _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\The Beast\anaconda3\Lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context return func(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\xxx\anaconda3\Lib\site-packages\gfpgan\utils.py", line 145, in enhance restored_img = self.face_helper.paste_faces_to_input_image(upsample_img=bg_img) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\xxx\anaconda3\Lib\site-packages\facexlib\utils\face_restoration_helper.py", line 355, in paste_faces_to_input_image upsample_img = inv_soft_mask * pasted_face + (1 - inv_soft_mask) * upsample_img ~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~ ValueError: operands could not be broadcast together with shapes (1404,2000,1) (1404,2000)

See here https://github.com/xinntao/Real-ESRGAN/issues/555#issuecomment-1705470424

RichardZhangHi commented 8 months ago

I did not have the .pth files in the \Real-ESRGAN directory and now I no longer get the first error, only the one ending in AttributeError: 'NoneType' object has no attribute 'shape'

Hi,do you find the solution about this issue?

WangJunjie97 commented 7 months ago

When I follow the code a little further down the page ...

wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.2.4/RealESRGAN_x4plus_anime_6B.pth -P weights

python inference_realesrgan.py -n RealESRGAN_x4plus_anime_6B -i inputs

I get this error instead:
python inference_realesrgan.py -n RealESRGAN_x4plus_anime_6B -i inputs
Testing 0 00003
Testing 1 00017_gray
Testing 2 0014
Testing 3 0030
Testing 4 ADE_val_00000114
Testing 5 OST_009
Testing 6 children-alpha
Testing 7 tree_alpha_16bit
        Input is a 16-bit image
Testing 8 video
Traceback (most recent call last):
  File "/home/nolan/Python/Real-ESRGAN/inference_realesrgan.py", line 166, in <module>
    main()
  File "/home/nolan/Python/Real-ESRGAN/inference_realesrgan.py", line 138, in main
    if len(img.shape) == 3 and img.shape[2] == 4:
AttributeError: 'NoneType' object has no attribute 'shape'

Testing 8 video video is a folder name, not an image file

xinntao / Real-ESRGAN

ValueError: operands could not be broadcast together with shapes (1404,2000,1) (1404,2000) #555

Here is a solution

Analyze

Reason

Solution

Example