Search found 50 matches

by Replicon
Wed Jul 28, 2021 1:21 am
Forum: Extract Discussion
Topic: How important is it for the jawline and eyebrows to be perfect, assuming other landmarks are good?
Replies: 3
Views: 9477

Re: How important is it for the jawline and eyebrows to be perfect, assuming other landmarks are good?

Aah ok. Then I should probably keep fixing them, since I'm basically defaulting to extended.

It's funny though, even with a well-defined chin, videos converted with extended tend to cut off the bottom just below the mouth, rather than including the full contour of the chin.

by Replicon
Wed Jul 21, 2021 9:35 pm
Forum: Extract Discussion
Topic: Manual Tool - Mask not updating with changes to bounding box
Replies: 5
Views: 14219

Re: Manual Tool - Mask not updating with changes to bounding box

OH! Looks like you can update through the UI. I just did, and it looks good now, mask is propagating with box change. Thanks for the fix!

by Replicon
Wed Jul 21, 2021 8:44 pm
Forum: Extract Discussion
Topic: How important is it for the jawline and eyebrows to be perfect, assuming other landmarks are good?
Replies: 3
Views: 9477

How important is it for the jawline and eyebrows to be perfect, assuming other landmarks are good?

I know the answer is "as perfect as possible", but I'm hoping for some more qualitative answers, to help decide if something will be good enough, or will be a complete mess if I don't spend the hours adjusting it. :) To give a specific example that I constructed from stock photos for illus...
by Replicon
Wed Jul 21, 2021 8:35 pm
Forum: Extract Discussion
Topic: Manual Tool - Mask not updating with changes to bounding box
Replies: 5
Views: 14219

Re: Manual Tool - Mask not updating with changes to bounding box

Yes, but I have not yet done a faceswap update on my box since the first time installing. I'll re-install to get the latest and get back to you... hopefully I can just run the install script and it'll do a graceful update.

by Replicon
Mon Jul 05, 2021 2:43 pm
Forum: Cloud Discussion
Topic: No GPU detected (on subsequent commands - first command is fine)
Replies: 1
Views: 11971

Re: No GPU detected (on subsequent commands - first command is fine)

Aha, running 'nvidia-smi' also results in: 'Failed to initialize NVML: Driver/library version mismatch' And I can see nvidia updates in /var/log/apt/history.log Sure enough, bringing up a fresh instance and running 'sudo apt-get dist-upgrade' puts it into the bad state. Unfortunately, even rebooting...
by Replicon
Mon Jul 05, 2021 2:13 pm
Forum: Cloud Discussion
Topic: No GPU detected (on subsequent commands - first command is fine)
Replies: 1
Views: 11971

No GPU detected (on subsequent commands - first command is fine)

I was running a slightly elaborate thing on a cloud instance, and it looked like the GPU disappeared between commands. While training a model for a few hours, everything ran smoothly, then when it saved it and ran convert, I got the "no gpu detected" error: An unhandled exception occured l...
by Replicon
Fri Jun 25, 2021 9:52 pm
Forum: Training Discussion
Topic: Training from multiple sources - how to combine alignments?
Replies: 4
Views: 10995

Re: Training from multiple sources - how to combine alignments?

Oh BTW, I just updated my script to support extracting from multiple sources to the same output directory, but it fails on the second iteration (second source), with: ERROR: Output faces folder should be empty Is this by design? Seems we should be able to use the tool to generate a faces folder for ...
by Replicon
Thu Jun 24, 2021 10:38 pm
Forum: Training Discussion
Topic: Training from multiple sources - how to combine alignments?
Replies: 4
Views: 10995

Re: Training from multiple sources - how to combine alignments?

Ahaaaaaaa, that was the missing piece! I didn't realize the alignments were actually stored in the png headers. Ok, the world makes sense again, thank you!

by Replicon
Mon Jun 21, 2021 6:45 pm
Forum: Training Discussion
Topic: Training from multiple sources - how to combine alignments?
Replies: 4
Views: 10995

Training from multiple sources - how to combine alignments?

So far, I've been combining source videos using Kdenlive, and then the "multiple alignment files" problem just goes away. But now, I'm faced with a situation where I want to incorporate a new source into the pool of sources for an identity. I'm not about to recombine the videos and redo th...
by Replicon
Mon Jun 21, 2021 6:16 pm
Forum: Training Discussion
Topic: Putting single photo B face on model?
Replies: 2
Views: 8472

Re: Putting single photo B face on model?

Got it, thanks! I figured that'd be the case... just trying to cut corners here and there :) BTW, if you have a small handful of photos, you can probably get some mileage out of them by using deep nostalgia's model to animate them. It's a little bit off-looking, so not sure how convincingly it'd tra...
by Replicon
Mon Jun 14, 2021 6:59 pm
Forum: Training Discussion
Topic: Putting single photo B face on model?
Replies: 2
Views: 8472

Putting single photo B face on model?

tl;dr: What would the best model/settings be to just put a single face/expression on a video? I don't care about changing expressions... just rotate and stretch the B photo enough that the alignment landmarks are "close enough", and render it. So, I was reading the "why you need more ...
by Replicon
Sun May 02, 2021 8:24 pm
Forum: Cloud Discussion
Topic: Some benchmarking numbers
Replies: 6
Views: 17111

Re: Some benchmarking numbers

Haha Money$$$$$ I should experiment with Villain, it seems like it's the state of the art right now. I wonder if you get better bang for your buck for running it overnight, vs with "original" as I've been doing. In the comparison thread, even lightweight is looking pretty good haha. Maybe ...
by Replicon
Sat May 01, 2021 6:21 pm
Forum: Training Discussion
Topic: "High quality sources" vs. "training only uses 64x64 image"
Replies: 1
Views: 5551

"High quality sources" vs. "training only uses 64x64 image"

Random thought/question that just popped into my mind: The recommendation is always that using high quality images/video sources for training is important... ... but since models seem to use small images (many only doing 64x64), does it mean the source quality isn't all that important, as long as th...
by Replicon
Sat May 01, 2021 3:31 pm
Forum: Cloud Discussion
Topic: Some benchmarking numbers
Replies: 6
Views: 17111

Re: Some benchmarking numbers

Thanks for the tip, will have to try with a n1-standard-2. I use preemptible hardware, so savings would add up to 2 cents per hour, but it's free to change my script's config. :)

by Replicon
Sat Apr 17, 2021 5:08 am
Forum: Cloud Discussion
Topic: Some benchmarking numbers
Replies: 6
Views: 17111

Some benchmarking numbers

I experimented with a variety of GPUs, and thought I'd share my numbers. I basically just ran 10K iterations of "original" with batch size 16. I figured it's representative enough for entry level stuff, but I'm sure things change once you get into bigger/fancier models. Anyway, at least fo...
by Replicon
Fri Apr 09, 2021 7:00 pm
Forum: Training Discussion
Topic: How portable are well-trained models to other videos of the same faces?
Replies: 1
Views: 2886

Re: How portable are well-trained models to other videos of the same faces?

Don't mind me, I found my answer in the best practices guide. That Harrison Ford/Nick Cage example with pulling from multiple movies is exactly my example.

by Replicon
Thu Apr 08, 2021 7:18 pm
Forum: Training Discussion
Topic: How portable are well-trained models to other videos of the same faces?
Replies: 1
Views: 2886

How portable are well-trained models to other videos of the same faces?

Suppose I have three videos: A1: Good collection of expressions/angles/lighting, for face A B: Good collection of expressions/angles/lighting, for face B A2: Some other video of face A, similar quality to A1 If I train a model to swap B's face onto A1, can I then use that same model to swap B's face...
by Replicon
Wed Apr 07, 2021 9:36 pm
Forum: Training Discussion
Topic: Do trainer options (e.g. flip chance) apply only to "Original" model or to everything?
Replies: 2
Views: 2804

Re: Do trainer options (e.g. flip chance) apply only to "Original" model or to everything?

Sweet thanks for confirming! Oh yeah I wasn't talking about manually tweaking the extracted photos, but literally appending some extra flipped clips at the end of the source video (with Kdenlive, say) and then using that as the source for the initial s3fd extract. Since the trainer already does the ...
by Replicon
Tue Apr 06, 2021 3:45 pm
Forum: Extract Support
Topic: [indexing bug?] Manual tool landmarks showing up on wrong preview thumbnail
Replies: 2
Views: 10629

[indexing bug?] Manual tool landmarks showing up on wrong preview thumbnail

I've been finding there are cases where selecting a face in the preview thumbnails will render the landmark preview in the thumbnail to the right of it. One easy way to do this (my example is with a really clean source where every frame has a face): Open up the manual tool with some video Click on s...
by Replicon
Tue Apr 06, 2021 3:31 pm
Forum: Extract Discussion
Topic: Manual Tool - Mask not updating with changes to bounding box
Replies: 5
Views: 14219

Re: Manual Tool - Mask not updating with changes to bounding box

Sweet thanks!

It's easy to work around, but the extra clicks add up if doing many frames.