Search found 95 matches

by cosmico
Sun Jun 06, 2021 8:32 pm
Forum: General Discussion
Topic: Challenges faceswapping VR videos
Replies: 12
Views: 33259

Re: Challenges faceswapping VR videos

@cosmico - how did you get on with this project? .... Any additional information or pointers from your end would be great - there really isn't much online about this. Hey man, sorry for the long delay. I've been working a lot and doing a lot of crypto mining recently. How did it get on? Pretty dece...
by cosmico
Sat Apr 03, 2021 11:44 pm
Forum: Extract Support
Topic: Extractions and training error using old model and data since recent update
Replies: 1
Views: 9544

Extractions and training error using old model and data since recent update

I’m currently away from the computer at the moment, but when I was on it a couple days ago I was having an error with both extractions and training when using older models. It started off first with me wanting to try and train some more on an older project I’m doing on my spare time between paid req...
by cosmico
Thu Mar 25, 2021 5:34 pm
Forum: Training Support
Topic: hardware requirements of different models?
Replies: 2
Views: 4925

Re: hardware requirements of different models?

There's are definitely some hard requirements, Some trainer will not work on some cards, but there also some soft requirements. When a model is training at a speed as low of 1 eg/s, it is technically working, but is practically unviable do to how long that model would take to train. Nobody wants tha...
by cosmico
Thu Mar 11, 2021 9:17 pm
Forum: Swap Showcase
Topic: Original Model can still produce great results!
Replies: 5
Views: 21921

Re: Original Model can still produce great results!

Here's the link to the model files. It's about 626mb. For anyone wanting to download, just don't use them for anything bad or try to take credit for them :D The Original Model Also I uploaded a slightly better version of the swap in the original post up top, and I made them both public on youtube. I...
by cosmico
Thu Mar 11, 2021 1:58 am
Forum: Swap Showcase
Topic: Creating a decent faceswap only using 4 images of B data
Replies: 6
Views: 27912

Re: Creating a decent faceswap only using 4 images of B data

Honestly now that I think about it, the app I used is extremely similar to first order AI, only I could make on the fly adjustments to 3d skeleton of the face, say if I ever felt the nose was too big or small or the eyebrows werent moving right. Check your DM's

by cosmico
Wed Mar 10, 2021 8:02 pm
Forum: Swap Showcase
Topic: Creating a decent faceswap only using 4 images of B data
Replies: 6
Views: 27912

Re: Creating a decent faceswap only using 4 images of B data

No way!? This wacky idea was the perfect info for you? lol If you don't mind me asking, which part?

by cosmico
Wed Mar 10, 2021 4:20 pm
Forum: Hardware
Topic: Can a 2060s or 2060 run dfl-sae at 256 resolution?
Replies: 5
Views: 13493

Re: Can a 2060s or 2060 run dfl-sae at 256 resolution?

To help you out, I just tried to create a DFLSAE @ 256 model on my project that I'm currently using with original. Ans I was unable to get a single 256 model starting with 6 different settings. The 6 settings I tried were: . DFLSAE 256 Allow Growth, Mixed precision, Architecture DF, Encoders: Defaul...
by cosmico
Tue Mar 09, 2021 5:55 pm
Forum: Content Creator Requests
Topic: Need DeepFake Creator
Replies: 1
Views: 14824

Re: Need DeepFake Creator

I know there's a template, but a lot of the information on this request seems similar to the trof3039 request. Maybe spam? Or maybe they just copied trod3039's request because they had similar request parameters? Well VisionWanda, If you are a separate person, trof3039 has not responded to my DM ye...
by cosmico
Mon Mar 08, 2021 10:39 pm
Forum: Content Creator Requests
Topic: Creatotor needed.
Replies: 4
Views: 16119

Re: Creatotor needed.

I might be interested in it. Check your DM's

by cosmico
Mon Mar 08, 2021 10:34 pm
Forum: Training Discussion
Topic: Would a B&W data set (on both sides) train faster than color?
Replies: 6
Views: 4423

Re: Would a B&W data set (on both sides) train faster than color?

I have a follow up question along the same lines If I'm understanding this correctly you said that its hypothetically possible for it be faster if there was a model which was set up only to be B&W which there isn't, so it wont be faster. So if I went back to the videos and pictures that make up ...
by cosmico
Sun Mar 07, 2021 6:14 pm
Forum: Swap Showcase
Topic: Original Model can still produce great results!
Replies: 5
Views: 21921

Re: Original Model can still produce great results!

would like to request you that can you share your trained model for others to download because I think a trained model can help in faster training when we compare it to a model that has to be trained from scratch. I would love to share my Original model, but how am I supposed to upload a 29Gb file ...
by cosmico
Sat Mar 06, 2021 6:10 pm
Forum: Swap Showcase
Topic: Tom Cruise -- possibly best Deepfake yet
Replies: 5
Views: 18396

Re: Tom Cruise -- possibly best Deepfake yet

http://cdn.bleacherreport.net/images_root/slides/photos/000/425/798/johncena_display_image.jpg?1286172319 This man has got connections. So does that imply a yes to imposters original question? It was done at least some part on faceswap? Since he worked for you both a little? The video on cnet artic...
by cosmico
Sat Mar 06, 2021 5:59 pm
Forum: Swap Showcase
Topic: Original Model can still produce great results!
Replies: 5
Views: 21921

Original Model can still produce great results!

One of the biggest misconceptions I think the community has is wanting higher quality results, but not using higher quality data, or getting impatient and not training their models to completion. Instead they just search for a better model. Sometimes people who are unhappy with their results will ta...
by cosmico
Sat Mar 06, 2021 5:33 pm
Forum: Swap Showcase
Topic: Tom Cruise -- possibly best Deepfake yet
Replies: 5
Views: 18396

Re: Tom Cruise -- possibly best Deepfake yet

Rachel's tweet is kind of self aggrandizing I predicted this two years ago and its been two years to the dot and look what we have I mean the technology existed that this could have easily been made a year ago if not two. But I don't disagree with her message. On the surface, having a synthetic medi...
by cosmico
Thu Feb 25, 2021 5:38 pm
Forum: Training Discussion
Topic: A couple small questions
Replies: 1
Views: 1733

A couple small questions

I was training and I was rotating my data like its recommended for better training, you know... letting the model see something new after a while, and I've noticed these odd valleys coming as a result. Now I've seen this happen many times and thought it was weird but never really thought too much ab...
by cosmico
Thu Feb 18, 2021 7:57 pm
Forum: Swap Showcase
Topic: A newbie's first (actually second) test project
Replies: 2
Views: 15770

Re: A newbie's first (actually second) test project

Nicely done! This is better than my very first swap!

by cosmico
Thu Feb 18, 2021 7:52 pm
Forum: Introductions
Topic: Hello, and my recording for my face training source
Replies: 2
Views: 14374

Re: Hello, and my recording for my face training source

Hello Atlas! Nicely done on capturing plenty of angles and facial reactions in your data set!! :) I think it could be significantly improved if you 1. made it higher resolution that 480p, 2. Recorded your face closer to the camera so your face takes up more of the frame, thus more pixels can be dedi...
by cosmico
Thu Feb 18, 2021 4:07 pm
Forum: Artificial Intelligence
Topic: How to deepfake, but with input expression kept. (like snapchat)
Replies: 2
Views: 14840

Re: How to deepfake, but with input expression kept. (like snapchat)

Sounds like your asking for faceswap, but instead of Face B perfectly replicating Face A, you want to control Face B's face and voice while its on Face A's body. These tools definitely exist, like you mentioned Snapchats face swap filter, but I don't think any of these tools exists in a format as ge...
by cosmico
Sun Feb 14, 2021 6:14 pm
Forum: Convert Support
Topic: The single most common problem I encounter: Conversion previews freezing with a PermissionError: [WinError 32]
Replies: 2
Views: 10447

The single most common problem I encounter: Conversion previews freezing with a PermissionError: [WinError 32]

So I will convert a video, and the previews will show me converted thumbnails every 5 seconds, and it will get to about 33%-50% done when I'll get hit with the PermissionError: [WinError 32] The process cannot access the file because it is being used by another process: 'D:/Nueral Network programs/F...
by cosmico
Fri Feb 12, 2021 7:34 pm
Forum: Training Support
Topic: unable to train and convert or preview model
Replies: 1
Views: 2496

Re: unable to train and convert or preview model

I'm by no means an expert, but when I run into CRITICAL Error caught! Exiting... ERROR Caught exception in thread: '_training_0' I first check to make sure that all my setting are right. Sometimes I have the wrong model selected from a previous project where I used a different model or sometimes I d...