Search found 1847 matches

by torzdf
Thu Aug 11, 2022 2:46 pm
Forum: Training Discussion
Topic: [Guide] Introducing - Phaze-A
Replies: 32
Views: 30482

Re: [Guide] Introducing - Phaze-A

Thanks for this post! This is exactly the kind of information and research that I would like to see from users.

I look forward to seeing how your model gets on :)

by torzdf
Thu Aug 11, 2022 8:25 am
Forum: Training Discussion
Topic: [Guide] Introducing - Phaze-A
Replies: 32
Views: 30482

Re: [Guide] Introducing - Phaze-A

MaxHunter wrote: Thu Aug 11, 2022 1:40 am

Wow!! I went this entire time without knowing there were presets?!!! Is this in the tutorial?

Yes. Linked from the very first post ;)
viewtopic.php?p=5367#p5367

by torzdf
Wed Aug 10, 2022 10:17 am
Forum: Installation Support
Topic: help me!“Some applications failed to install.Process Aborted.Check Details.”
Replies: 1
Views: 53

Re: help me!“Some applications failed to install.Process Aborted.Check Details.”

Ok, this is a system issue, which I cannot directly solve. However, there appears to be an SSL issue on your machine. A bit of googling comes up with some potential solutions: https://www.google.com/search?q=git+error+openssl+ssl_read+connection+was+reset+errno+10054&client=ubuntu&hs=NQ7&...
by torzdf
Wed Aug 10, 2022 10:13 am
Forum: Training Discussion
Topic: [Guide] Introducing - Phaze-A
Replies: 32
Views: 30482

Re: [Guide] Introducing - Phaze-A

I couldn't tell you batch size, just based on settings. See here for finding the best batch size yourself:

viewtopic.php?p=388

by torzdf
Tue Aug 09, 2022 2:20 pm
Forum: Extract Discussion
Topic: Possible to Delete Source Pictures / Clean Alignments File?
Replies: 2
Views: 57

Re: Possible to Delete Source Pictures / Clean Alignments File?

Your use is very edge-case, as I have not seen it raised before.

Short answer: no, and not something I'm likely to add, I'm afraid.

by torzdf
Mon Aug 08, 2022 5:48 pm
Forum: Training Support
Topic: empty face
Replies: 2
Views: 40

Re: empty face

Corrupted model.

Lower learning rate + rollback 50k

by torzdf
Mon Aug 08, 2022 5:47 pm
Forum: Training Discussion
Topic: Dlight worked to about 6000 its, then flatlined
Replies: 3
Views: 50

Re: Dlight worked to about 6000 its, then flatlined - addt'l questions

That leaves me with a jumbled folder of source faces from which I remove unwanted faces for training and alignments files that exist but which are mostly from the original extract (extract n= usually 15-40 depending on fps, length and quality of video) and which would be difficult to tie back to sp...
by torzdf
Mon Aug 08, 2022 2:07 pm
Forum: Training Discussion
Topic: Dlight worked to about 6000 its, then flatlined
Replies: 3
Views: 50

Re: Dlight worked to about 6000 its, then flatlined

It's been a long time since I used DLight, but this is not an issue I have seen before.

Generally that model needs a lower LR (3.5e-5) as it tends to collapse otherwise.

That may solve your issue, it may not.

by torzdf
Mon Aug 08, 2022 1:53 pm
Forum: Installation Support
Topic: Windows Installer error: “Some applications failed to install.Process Aborted.Check Details.”
Replies: 3
Views: 214

Re: Windows Installer error: “Some applications failed to install.Process Aborted.Check Details.”

This is a connection issue with your setup/conda.

Do this and try again. Hopefully it will resolve any Conda related issues:
app.php/faqpage#f1r1

by torzdf
Sun Aug 07, 2022 6:46 pm
Forum: Training Discussion
Topic: Clarificaton on Training options: Mask selection
Replies: 4
Views: 111

Re: Clarificaton on Training options: Mask selection

The easiest way to see it is visually. Look at the "red" area in the training previews. These are masked out. Features are badly learned in the masked out area (this is the point of masking without 'learn mask enabled', it tells the NN which parts of the image it is interested in, and whic...
by torzdf
Sat Aug 06, 2022 4:17 pm
Forum: Training Discussion
Topic: [Guide] Introducing - Phaze-A
Replies: 32
Views: 30482

Re: [Guide] Introducing - Phaze-A

Thank you for your advise , Would you please suggest some setting about 1080p video for using V2 with rtx 3080 n 3090 ? because for now , v2-s with stojo present , 1080p video , some close-up shot the face go blur , but v2-l with stojo present much better . suggestion like : extraction px phaze-A o...
by torzdf
Sat Aug 06, 2022 4:13 pm
Forum: General Discussion
Topic: Masking with CPU
Replies: 4
Views: 118

Re: Masking with CPU

Ok, good to know. Thanks. I will look at the potential for choosing to put that on cpu.

by torzdf
Sat Aug 06, 2022 4:12 pm
Forum: Convert Discussion
Topic: wired highlight patches
Replies: 1
Views: 41

Re: wired highlight patches

Looks like model corruption to me.

You'd need to try rolling back 50k or so, lowering learning rate a bit and continuing.

by torzdf
Sat Aug 06, 2022 4:11 pm
Forum: Training Discussion
Topic: Potential VRAM Saving techniques
Replies: 11
Views: 274

Re: Potential VRAM Saving techniques

Just an update. I'm noticing that there is a much longer delay between iterations compared with, for example, the 256 StoJo model. The GPU is not used to its full potential. For example, this is a graph for I believe the slowdown is caused by the larger amount of data (larger image sizes - which ar...
by torzdf
Mon Aug 01, 2022 1:22 am
Forum: Training Support
Topic: Bug: ValueError: Please use `tf.keras.losses.Reduction.SUM` or `tf.keras.losses.Reduction.NONE`
Replies: 19
Views: 720

Re: Bug: ValueError: Please use `tf.keras.losses.Reduction.SUM` or `tf.keras.losses.Reduction.NONE`

Removing the bug tag from this as -d, --distributed is deprecated in favour of -D, --distribution-strategy

by torzdf
Mon Aug 01, 2022 12:52 am
Forum: General Discussion
Topic: saving individual frames from manual
Replies: 1
Views: 40

Re: saving individual frames from manual

No. It is purely a numbers thing. I doubt I would implement anything like this, because if you're at the point of reviewing every single frame, then you might as well review every single face. At which point, you're manually going through everything anyway, so could build your training set that way.

by torzdf
Mon Aug 01, 2022 12:49 am
Forum: General Discussion
Topic: Masking with CPU
Replies: 4
Views: 118

Re: Masking with CPU

Which mask were you using? If landmarks based mask, then the GPU isn't used at all. The mask is generated from landmarks.

If it was an NN based mask, it would seem unlikely. However, let me know which one, because if it is as fast on CPU, then that is something I would want to look at.

by torzdf
Mon Aug 01, 2022 12:48 am
Forum: Training Discussion
Topic: Clarificaton on Training options: Mask selection
Replies: 4
Views: 111

Re: Clarificaton on Training options: Mask selection

Yes, there are huge benefits for multiple reasons. Off the top of my head: 1) Landmarks based masks are only as good as the landmarks. It really requires very precise landmark alignments for the mask to be effective. For face alignment, only the core landmarks are used (all landmarks, excluding the ...
by torzdf
Mon Aug 01, 2022 12:44 am
Forum: Training Discussion
Topic: AutoClip: Any Feedback From Users?
Replies: 7
Views: 247

Re: AutoClip: Any Feedback From Users?

Gradient clipping is a mechanism to help prevent exploding/vanishing gradients (that is numbers that go to +/- infinity or to 0). Both of these will cause a model to NaN (Mixed Precision is more prone to this, as infinity in limited precision space is a smaller number than infinity in full precision...
by torzdf
Sat Jul 30, 2022 10:53 pm
Forum: Convert Support
Topic: How do I change the face of only one of two people in one picture?
Replies: 2
Views: 3587

Re: How do I change the face of only one of two people in one picture?

Protip: Don't report the forum owner for spam.

It is in the convert guide. Literally covered in the first section. Also this:
viewtopic.php?t=2090