Unable to find error_log.txt

[galaxyentity904]

Describe the bug A clear and concise description of what the bug is. When running any inference from GIMP-ML, it fails. But I cannot find a error_log.txt even after doing find / error_log.txt | grep error_log.txt I've gotten this error on both my laptop and desktop To Reproduce Steps to reproduce the behavior:

1. Go to '...'

2. Click on '....'

3. Scroll down to '....'

4. See error

5. Git clone of gimp-ml and switch branch to GIMP3-ML

6. Install gimp development version commit bf1b405

7. Run install.bat with no errors

8. Add folder to plugins

9. Attempt to do something using GIMP-ML on an image See error

Screenshots If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS Version: [e.g. Ubuntu] Xubuntu 21.04
• GIMP Version: [e.g. 2.10] 2.99.6 Commit bf1b405

[kritiksoman]

what files are present in the GIMP-ML folder in your home directory?

[galaxyentity904]

Immediately after running, the following files: weights (directory) gimp_ml_run.pkl cache.png cache0.png

[kritiksoman]

are you able to run kmeans or open color palette?

[galaxyentity904]

Kmeans gives me the same error

But Color Palette gives me this image

[xjdeng]

I'm able to run kmeans and color palette but superresolution isn't working for me. Neither is colorize. Both times, I'm getting the "inference not successful" message.

I checked my weights directory and found that some of the models weren't downloaded (folders were missing.) I tried downloading those folders manually, at least for superresolution and colorize but it still fails to run.

Does anyone know of a commit where at least superresolution is working? I'm also on bf1b405 for GIMP.

[kritiksoman]

are you able to run the demo jupyter notebook using the environment created by the setup?

[Begalov]

I have cloned GIMP3-ML branch Run "install.bat gpu", but to complete installation I've manually (using pip) added this packages to 'B:\DEV\GIMP-ML\gimpenv3' in the order in which the script swore at their absence: opencv-python, timm, gdown, scikit-image, pretrainedmodels

weights were added to the 'C:\Users\RUben\GIMP-ML' folder via a simlink

win10 gimp2.99.8 python3.8

[AlSchemist]

Hi Xjdeng,

gimp-2.99.10 Windows, CPU no CUDA:

on 14 Sep 2021 Xjdeng wrote:

[...] superresolution isn't working for me. Neither is colorize. Both times, I'm getting the "inference not successful" message.

I checked my weights directory and found that some of the models weren't downloaded (folders were missing.) I tried downloading those folders manually, at least for superresolution and colorize but it still fails to run.

You are right. "inference not successful" could mean that GIMP3-ML does not find the corresponding: C:\Users\UserName\GIMP-ML\weights Even if you clicked "Download ALL" from Google Drive, some critical sub-folders could be not loaded.

GIMP-ML/gimpml/tools/superresolution.py> line 24: "superresolution" is written without separator between "super" and "resolution".

However in Google Drive super_resolution is written with the separator underscore "_".

After you unzipped the big weights from Google Drive, try to rename "super_resolution" to "superresolution" without separator: C:\Users\UserName\GIMP-ML\weights\super_resolution becomes: C:\Users\UserName\GIMP-ML\weights\superresolution where there is the .pth file: model_srresnet.pth 6257113 = 6,111 KB

@kritiksoman Thanks for your great GIMP3 ML plugins. SuperResolution is probably the most interesting plugin. However its weights are the smallest. 6 MB with regards to 1.5 GB. Could you improve this neural network in order to take in consideration the readability of AI enlarged fonts in a small screen capture?