It either says Error: image failed to be upscaled or Error: a segment failed to be created... in the powershell im not seeing any errors i can copy and paste it if you want... i have a gtx 980 ti 16gigs of ddr4 ram and 6600k processor... both cpu and gpu are overclocked. I can add you on discord or steam if that would be easier to help me out?
Custom Texture Tool PS v52.5
|
If a segment fails to be created, there's a 99% chance the reason is because SFTGAN is failing and not the script, aside from everything failing after that since there's no segment. I can upload something to test, but if you want to make a test case yourself, edit the script on line 2567 and 2573, removing 2>&1>$null from the ends of the lines.
Spoiler: So in short, edit those lines, run SFTGAN, see if anything interesting pops up in the console. If you want me to edit and upload a version instead let me know. Edit: Also, what install method are you using? Did you install the models? I have the programs pack I uploaded (comes with models), the script has the built-in installer (models page pops up, manual download and install), and then of course you could be manually installing (find models yourself, manual download and install). 07-18-2019, 11:12 PM
(07-18-2019, 09:17 PM)Bighead Wrote: If a segment fails to be created, there's a 99% chance the reason is because SFTGAN is failing and not the script, aside from everything failing after that since there's no segment. I can upload something to test, but if you want to make a test case yourself, edit the script on line 2567 and 2573, removing 2>&1>$null from the ends of the lines.Hey ill edit these soon and report back any errors I'm receiving. As for my install method I used the program pack originally. At the time i already had python installed so only needed to install cuda manually which i did... (I didn't forget to use the update .bat afterwards for pip etc) that didn't work so i uninstalled both python and cuda and used the built in installer to see if that would make a difference still nothing... i did test with both of the 2 models included for SFTGAN. (07-18-2019, 11:12 PM)opengrip Wrote: Hey ill edit these soon and report back any errors I'm receiving. As for my install method I used the program pack originally. At the time i already had python installed so only needed to install cuda manually which i did... (I didn't forget to use the update .bat afterwards for pip etc) that didn't work so i uninstalled both python and cuda and used the built in installer to see if that would make a difference still nothing... i did test with both of the 2 models included for SFTGAN.Hey sorry for the delay I can't actually find the script to edit that you showed above... in my SFTGAN folder I have py_torch test folder... and in that I have architectures.py block.py test_segmentation.py test_sftgan.py and util.py and from what I could tell none of those had the code that you showed above... Am i missing something? actually envermind i found it now lol reporting error soon SFTGAN: Upscaling image... --- Job Loop Started (SFTGAN) << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> Traceback (most recent call last): + CategoryInfo : NotSpecified: (Traceback (most recent call last)::String) [], RemoteException + FullyQualifiedErrorId : NativeCommandError + PSComputerName : localhost File "test_segmentation.py", line 11, in <module> import torchvision.utils ModuleNotFoundError: No module named 'torchvision' << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> << Job Loop Iteration >> Traceback (most recent call last): + CategoryInfo : NotSpecified: (Traceback (most recent call last)::String) [], RemoteException + FullyQualifiedErrorId : NativeCommandError + PSComputerName : localhost File "test_sftgan.py", line 10, in <module> import util File "D:\CTT-PS Programs\SFTGAN\pytorch_test\util.py", line 10, in <module> from torchvision.utils import make_grid ModuleNotFoundError: No module named 'torchvision'
Yeah I should have specified my script lol, sorry. It looks like torchvision isn't installing for whatever reason. This should happen automatically with either the batch script included in the pack or when installing with CTT-PS. You can try to manually install it via command line.
Code: pip3.exe install https://download.pytorch.org/whl/cu100/torch-1.0.1-cp36-cp36m-win_amd64.whl I'm not sure why the installation would fail, I don't know much about python in general, but at least its a lead on the issue. Edit: Oh I think it requires CUDA installation. I know you mentioned you installed CUDA, maybe python isn't seeing it for some reason? (07-20-2019, 10:39 AM)Bighead Wrote: Yeah I should have specified my script lol, sorry. It looks like torchvision isn't installing for whatever reason. This should happen automatically with either the batch script included in the pack or when installing with CTT-PS. You can try to manually install it via command line.sorry im pretty noobish to running this is command line... do i just copy that into cmd or do i need to set directories or anything? Ill try to use the batch script again first and see if it helps UPDATE: batch says this - ERROR: Could not find a version that satisfies the requirement torch>=1.1.0 (from torchvision) (from versions: 0.1.2, 0.1.2.post1, 0.1.2.post2) ERROR: No matching distribution found for torch>=1.1.0 (from torchvision) 07-20-2019, 10:48 AM
(07-20-2019, 10:43 AM)opengrip Wrote: sorry im pretty noobish to running this is command line... do i just copy that into cmd or do i need to set directories or anything? Ill try to use the batch script again first and see if it helps https://www.dropbox.com/s/mn684bc9k07k0l...e.png?dl=0 here is an image of the batch running 07-20-2019, 10:54 AM
I did a quick search and it looks like it could be an issue with pytorch itself at the moment. I'll look more into it and see what I can find.
https://github.com/pytorch/pytorch/issues/10443 07-20-2019, 11:05 AM
(07-20-2019, 10:54 AM)Bighead Wrote: I did a quick search and it looks like it could be an issue with pytorch itself at the moment. I'll look more into it and see what I can find.Fixed it. I went to the site for pytorch and put in all the stuff to get a new command..... windows pip python 3.6 cuda 10 and re ran the code it gave me... and that seemed to fix it all. Thanks again for the great program! 07-20-2019, 11:18 AM
Awesome! I'm guessing their server is having issues resolving the latest version or something, based on the posts in that issue report. I'll wait awhile and see if the problem corrects itself in time and if more people stumble upon this issue. If that happens, I'll have to update the scripts to just download 0.3.0 directly.
|
« Next Oldest | Next Newest »
|
Users browsing this thread: 6 Guest(s)