Search found 1838 matches

by torzdf
Sun Jul 10, 2022 8:51 am
Forum: Training Discussion
Topic: Is my CPU a bottleneck?
Replies: 6
Views: 130

Re: GPU Not being Used

That CPU should be ok. Stats seem reasonable. I would expect GPU utilization to be higher, but every system is different. The system CPU/RAM is utilized for data augmentation/feeding the neural network. Glad you got up and running. ________ Edit: Just checked your CPU and it is dual core. You may we...
by torzdf
Sat Jul 09, 2022 1:52 am
Forum: Installation Support
Topic: GPU Not being Used
Replies: 6
Views: 219

Re: GPU Not being Used

It is possible to train lightweight on 2GB Vram, but I don' t like to guarantee it (it was built and tested on 1.6GB a long time ago). However, every system is different, and Tensorflow/Cuda will change requirements up and down on different iterations for the same task. System RAM is never shared. Y...
by torzdf
Sat Jul 09, 2022 12:23 am
Forum: Installation Support
Topic: GPU Not being Used
Replies: 6
Views: 219

Re: GPU Not being Used

It's not speed that will be the issue. It's unlikely to be enough VRAM (under Windows) to load any models. You might be able to run the lightweight model, but I would rank your chances at 'slim'.

Also, this:
https://www.quora.com/Does-Nvidia%E2%80 ... rogramming

by torzdf
Fri Jul 08, 2022 11:29 pm
Forum: Convert Support
Topic: output video saved but terminal frozen
Replies: 5
Views: 118

Re: output video saved but terminal frozen

That error message is just saying that the process was forced to shut down (which is correct, because you hit the Stop button)

by torzdf
Fri Jul 08, 2022 11:28 pm
Forum: Convert Discussion
Topic: How to determine if Convert is still processing or complete?
Replies: 8
Views: 208

Re: How to determine if Convert is still processing or complete?

Most likely an issue with the video file.

The process just goes through the video file from first frame to last frame. Again, without the video or alignments file I cannot give any more feedback as this is not an issue I have ever had.

by torzdf
Fri Jul 08, 2022 11:27 pm
Forum: Installation Support
Topic: GPU Not being Used
Replies: 6
Views: 219

Re: GPU Not being Used

Cuda and cudnn are installed fine in your environment: cudatoolkit 11.2.2 h933977f_10 conda-forge cudnn 8.1.0.77 h3e0f4f4_0 conda-forge The first step is to always to do this, it solves 99% of issues, but only if you don't skip any steps: https://forum.faceswap.dev/app.php/faqpage#f1r1 However, with...
by torzdf
Thu Jul 07, 2022 11:58 pm
Forum: Cloud Discussion
Topic: [Resource] Google Colab Notebook
Replies: 27
Views: 1877

Re: [Resource] Google Colab Notebook

That was code I added in a recent commit to stop the preview window from stealing focus (on normal systems).

figure.raise_window definitely is a valid rcParam, so either you have an outdated version of matplotlib or Colab just doesn't like it for whatever reason.

by torzdf
Thu Jul 07, 2022 11:54 pm
Forum: Convert Discussion
Topic: How to determine if Convert is still processing or complete?
Replies: 8
Views: 208

Re: How to determine if Convert is still processing or complete?

If it's hit 100% and all the frames are in the output folder, it's finished. I have never had a problem with convert getting stuck though. If it's possible to provide the video + alignments file then I can look to see if I can recreate and identify the issue with it not completing. If not, no bother...
by torzdf
Wed Jul 06, 2022 9:32 pm
Forum: Convert Support
Topic: output video saved but terminal frozen
Replies: 5
Views: 118

Re: output video saved but terminal frozen

This may be an M1 specific issue. Unfortunately I don't own an M1, so am unable to investigate.

by torzdf
Wed Jul 06, 2022 9:27 pm
Forum: Training Support
Topic: How to Restore from backup
Replies: 1
Views: 70

Re: How to Restore from backup

It's been moved (will update the docs). If you go tools > model you'll find the restore job there.

by torzdf
Tue Jul 05, 2022 6:34 pm
Forum: Training Discussion
Topic: AutoClip: Any Feedback From Users?
Replies: 7
Views: 227

Re: AutoClip: Any Feedback From Users?

Just to follow up, and I no longer think this is an issue we are facing. I ran some tests, and could see no noticeable difference between shared and unique layers. I have, however read this from Nvidia's guide on Mixed Precision: While many networks match FP32 training results when all tensors are s...
by torzdf
Mon Jul 04, 2022 10:29 pm
Forum: Extract Support
Topic: Unable to download vggface2_resnet50_v2.zip
Replies: 1
Views: 67

Re: Unable to download vggface2_resnet50_v2.zip

Update faceswap. I moved the hosting path for all the models recently.

by torzdf
Mon Jul 04, 2022 9:48 am
Forum: Extract Discussion
Topic: Delete/copy/paste/many - face alignments
Replies: 10
Views: 407

Re: Delete/copy/paste/many - face alignments

Unfortunately not at the moment. At some point I will revisit the manual tool to enhance it some more, but it is a particularly complex bit of code to maintain. Also, as I am the only contributor to the project at the moment, this is not currently top of my priority list (it is on my priority list, ...
by torzdf
Sun Jul 03, 2022 10:49 am
Forum: Training Discussion
Topic: AutoClip: Any Feedback From Users?
Replies: 7
Views: 227

Re: AutoClip: Any Feedback From Users?

FWIW I have been doing a lot of debugging around NaNs. They do happen in ML models, but I wanted to get a better understanding of why they occur. This has been a time-consuming and laborious process. However, my investigations have led me to conclude that the main issue is within Keras/TF implementa...
by torzdf
Sat Jul 02, 2022 10:14 am
Forum: Convert Support
Topic: output video saved but terminal frozen
Replies: 5
Views: 118

Re: output video saved but terminal frozen

I'm not sure why that would happen. I have not had a problem with frozen convert, and without being able to replicate, it will be very hard for me to fix.

by torzdf
Fri Jul 01, 2022 4:06 pm
Forum: Training Discussion
Topic: AutoClip: Any Feedback From Users?
Replies: 7
Views: 227

Re: AutoClip: Any Feedback From Users?

My experience has been that it has not been the NaN saviour that I had hoped it would be, so my battle to combat NaNs continues :( I have seen no negative impact on accuracy, though my sample set is small. This is a link to original paper if anyone wishes to investigate the claims themselves: https:...
by torzdf
Thu Jun 30, 2022 10:20 pm
Forum: Training Discussion
Topic: Blurry faces
Replies: 8
Views: 359

Re: Blurry faces

Yes, the "No Warp" option.

Don't turn it off too early as it will destroy your model. If you are seeing eye glare/defined teeth, then it is a good time to disable it.

by torzdf
Thu Jun 30, 2022 10:19 pm
Forum: Extract Support
Topic: alignment file path
Replies: 1
Views: 63

Re: alignment file path

I'm afraid not.

by torzdf
Thu Jun 30, 2022 9:33 am
Forum: Training Discussion
Topic: Blurry faces
Replies: 8
Views: 359

Re: output face is blurry

Blurry output can happen for a lot of reasons. Not enough variety in the data (this is the most important one. Data should be highly varied) Resolution of the model (IAE is 64px. It is always going to be blurry to an extent) Model needs to have warp disabled (see here: https://forum.faceswap.dev/vie...
by torzdf
Wed Jun 29, 2022 11:35 pm
Forum: Training Support
Topic: Training Continuation Error: Unknown config_item: RandomNormal
Replies: 1
Views: 106

Re: Training Continuation Error: Unknown config_item: RandomNormal

Hmmm. Not seen this before, and really don't know what the issue could be. RandomNormal is a standard initializer which definitely exists in Keras: https://github.com/keras-team/keras/blob/d8fcb9d4d4dad45080ecfdd575483653028f8eda/keras/initializers/initializers_v2.py#L302 A quick bit of googling sug...