Faceswap Forum

I have a video where I want to replace the face. Naturally I want the extracted face to have the best alignment possible. So I would like to use "Normalization", "Re Feed" and "Re Align". If I understand it correctly. Each of them reduces the speed. And if the video I h...

It's pretty hot where I am. I'm already sitting in my underwear and I don't like how the coolers on my video card sound. ) And they say it'll get even hotter. So I began thinking that maybe I need to take a break for a couple of months with training the model. Which I don't really want to do. And a ...

I made an experiment some time ago. Not exactly a pure one, but still, I think it could be interesting/useful. I intended to make a video comparison in the end, but ran out of gas. ) But I still made some images for comparison which would fit forum format, I think. The end-result of the faceswap fee...

So I just recently noticed this part of the forum exists and I wanted to leave something of mine here. ) As a foreword - I had my reasons to do this deepfake and those reasons were not malicious. :D I'm kinda fan of MxR Plays channel. But, well... He's a simp. That bothers me. Hope I don't get banne...

At extract time, the process does not know which frames have faces, and of those frames with faces, which of those faces are of the individual you are interested in. It's the same level of uncertainty as with extracting every N frame. But it's easier to calculate in your head. The way I see it, bot...

So, it's been several times I wished I had this option, so I'm going to ask if it's possible to add it in the future - if it's not too inconvenient or against some other reasoning. I would love to have "Extract M amount" option when extracting faces. Let's say I have a 30m interview of a p...

I have never seen it happen in Faceswap (and I have trained models a VERY long way). That is not to say it isn't a thing, just that I've never hit it (I train with a LOT of data). The fix for overtraining is giving it new data, but the prevention is to give it that data all along. In other words, b...

I recently found out that overtraining is a thing that can happen. (I know it's in the guide, but when you (I mean, me) read the guide for the first time, you filter out everything that doesn't answer the question "how do you make this thing work at all?") I searched the forum and if I'm n...

The scenario is somewhat like this: I start training the model. Some time after on the preview I see an error in the mask I want to fix. While fixing this error, I see some more errors in the mask and landmarks. Nothing too damaging, I presume (the resulting model looks ok, think), but still. The qu...

Cool.

And I totally agree - it would be an overkill to fit it into the preview window.

So I did try to mess up the mask (one eye and a half of the mouth). Looks like the process itself is fine - I see some of the results of messing up the masks. On the 03 image of preview during training you can definitely see one eye masked out. If you know what to expect - you can see a part of the ...

Did you re-extract your face images after updating the masks and saving them to the alignments file? Yeah, I did that. My workflow was a bit different (I extract with the masker checked for example). But when I encountered the problem, after checking the masks with the manual tool I did extract the...

Sure, I'll roll the previews to get one that would illustrate. Before I go do that I have to explain - it's not my proudest deepfake, but it's not as bad as it probably looks like. :D (I can explain though the explanation isn't that interesting I think) I attached two images. First one (01.png) is t...

The preview I see during training makes me think some parts of the mask were not exported properly. Or maybe my eyes deceive me due to bright vibrant colors of the parts where I expect to see the mask. Maybe the mask is just too transparent and neutral. I checked the mask with Tools -> Manual and ex...

I would say yes. With Phaze-A you could freeze the entire B side and just get decoder C caught up. Ah... Interesting. I've been using Villain so far since it looked like an easy option. Phaze-A could be interesting to look at in the future. Thanks for the answer. Good to know that previously traine...

I've red some explanations, but I'm not sure if I understand correctly, so I'd want to get some clarification. I started with the desire to exchange face A with face B. Trained the correlating model. All good. Now I want to exchange face C with face B. Seems like a waste to start training both faces...