Search found 27 matches

by ugramund
Sat Apr 03, 2021 8:18 am
Forum: Training Discussion
Topic: RealFace or Dlight when training for 128px output?
Replies: 7
Views: 7181

Re: RealFace or Dlight when training for 128px output & RealFace settings

My Final RealFace model settings. Hope it would help someone. ------------------------------------------------------------------------------ My main use of FaceSwap was always for doing PHOTO swaps & not VIDEO swaps since :- 1)Easy to do & gets completed quickly 2)Limitation of my current ha...
by ugramund
Fri Apr 02, 2021 3:06 am
Forum: General Discussion
Topic: April Fools -- Big news, FaceKoin announcement!
Replies: 16
Views: 22889

Re: Big news for April 1st, FaceKoin announcement!

bryanlyon wrote: Fri Apr 02, 2021 1:34 am

Thank you all for participating in our April Fools day announcement. We hope you all had fun.

:) , I hope & pray that this awesome software always remain FREE, FULLY FUNCTIONAL & OPEN SOURCE forever.

by ugramund
Thu Apr 01, 2021 9:37 am
Forum: General Discussion
Topic: April Fools -- Big news, FaceKoin announcement!
Replies: 16
Views: 22889

Re: Big news, FaceKoin announcement!

money ruins everything Then I will have to just disconnect my Internet during training to make sure it can't mine or may be I will never update my FaceSwap version now. The good thing is, I have already made an offline installer of my current FaceSwap version so that I can install it on any PC. I d...
by ugramund
Thu Apr 01, 2021 7:36 am
Forum: General Discussion
Topic: April Fools -- Big news, FaceKoin announcement!
Replies: 16
Views: 22889

Re: Big news, FaceKoin announcement!

When turned off, it'll still mine 15% of the time. This is done to cover dev costs. But you can make that back if you allow FaceSwap devs to have 50% commission on the NFT instead. You mean to say that if someone like me who does not want do mining, turns OFF the mining option, it will still mine i...
by ugramund
Thu Apr 01, 2021 5:07 am
Forum: General Discussion
Topic: April Fools -- Big news, FaceKoin announcement!
Replies: 16
Views: 22889

Re: Big news, FaceKoin announcement!

Just asking, can we turn off the mining, if someone doesn't want participate in mining
or will it be always ON in the background while training ?

by ugramund
Thu Apr 01, 2021 4:45 am
Forum: Training Discussion
Topic: Making UNBALANCED model into BALANCED model
Replies: 1
Views: 2942

Making UNBALANCED model into BALANCED model

If we set the Complexity Decoder A & B to EQUAL values of 512 in UNBALANCED model,
will it make it a BALANCED model resulting in equal swapping of A->B and B->A?

by ugramund
Wed Mar 31, 2021 3:25 am
Forum: Training Discussion
Topic: RealFace or Dlight when training for 128px output?
Replies: 7
Views: 7181

Re: RealFace or Dlight when training for 128px output?

One last question [mention]bryanlyon[/mention], since I feel like I have found my "DREAM MODEL" in the RealFace model. Supposing that in the near future, I can rent a GPU farm or become an owner of a very powerful machine, at what values should I keep the :- Dense Nodes Complexity Encoder ...
by ugramund
Sun Mar 28, 2021 12:44 pm
Forum: Training Discussion
Topic: Changing output size mid-train
Replies: 2
Views: 2580

Re: Changing output size mid-train

Hi, is it possible to change the output size for one trainer in mid-train process or will it mess up the model? For example change dfaker output from 128 to 256? Can i expect better results in the endphase of training or is it not really helpful? Im aware of that a bigger output size needs more VRa...
by ugramund
Thu Mar 25, 2021 5:35 pm
Forum: Training Discussion
Topic: Is Learning Rate inversely proportional to Batch size ?
Replies: 1
Views: 3305

Is Learning Rate inversely proportional to Batch size ?

Training on my CPU for more than a month now with different image sets on 3 different models :- Original - Fastest EGs/sec batch size = 12 Dlight (128in/out) - Fastest EGs/sec batch size = 9 RealFace (64in/128out)- Fastest EGs/sec batch size =5 What I have seen (not one time but many times) that if ...
by ugramund
Sat Mar 20, 2021 9:13 am
Forum: General Discussion
Topic: Option of adding a switch between CPU/GPU training under training tab
Replies: 1
Views: 10189

Option of adding a switch between CPU/GPU training under training tab

Mr[mention]torzdf[/mention] , is it possible for you to add a CPU /GPU training switch under training tab so that we can switch training between CPU & GPU like we can switch between different models under training tab? I am asking this question because right now I am training on a RealFace model...
by ugramund
Sat Mar 20, 2021 6:42 am
Forum: Training Discussion
Topic: RealFace or Dlight when training for 128px output?
Replies: 7
Views: 7181

Re: RealFace or Dlight when training for 128px output?

It's more complicated than that. The encoder is trained to take a face and encode it so that the decoder can re-create the face. Giving it a 64px or 128px it should be able to get all the face data it needs to be able to feed the decoder. The decoder is responsible for all the detail above that. As...
by ugramund
Tue Mar 16, 2021 3:05 pm
Forum: Training Discussion
Topic: Can setting the Learning Rate to lower values help in getting the details better?
Replies: 1
Views: 2474

Can setting the Learning Rate to lower values help in getting the details better?

After training for a good amount of time (with warp on), if we reduce the learning rate to very low values like 1e or 2e on a small batch of 4 or 8 (with warp off) & train for some time & again set the learning rate to normal, will it help to get the details of the facial feature better?

by ugramund
Mon Mar 15, 2021 1:42 pm
Forum: General Discussion
Topic: Future support of current Models?
Replies: 1
Views: 10401

Future support of current Models?

Mr[mention]torzdf[/mention], as your the main developer of the Faceswap project, I would like to ask you that in the future, say after 2-3 years when many more development would have taken place in the GPU hardware, AI programming & perhaps with newer Models for face swapping, will the current m...
by ugramund
Sun Mar 14, 2021 8:11 am
Forum: Training Discussion
Topic: Is a Loss value of perfect 0.0 possible? Has anybody achieved it?
Replies: 2
Views: 3106

Is a Loss value of perfect 0.0 possible? Has anybody achieved it?

Supposing that somebody is an owner of a GPU farm & also a deepfake specialist, having a high quality data sets of 2 persons with similar faces & he is only training to convert for a video of only 10 seconds length in which the face is facing the center all the time, is it possible that he c...
by ugramund
Fri Mar 12, 2021 3:51 pm
Forum: Training Discussion
Topic: RealFace or Dlight when training for 128px output?
Replies: 7
Views: 7181

RealFace or Dlight when training for 128px output?

Mr [mention]torzdf[/mention] has already said in https://forum.faceswap.dev/viewtopic.php?t=418 that dlight > dfaker > h128 & so I am not including Dfaker & DflH128 here. For a given high quality data, what should I use :- RealFace - 64px input & 128px output Dlite - 128px input/output I...
by ugramund
Fri Mar 12, 2021 2:39 pm
Forum: Training Discussion
Topic: Does Dfaker supports 128px input?
Replies: 1
Views: 2345

Does Dfaker supports 128px input?

[mention]torzdf[/mention] you have written in the training manual that :- Dfaker (64/128px input, 128/256px output). So when does it take 128px as input. Does it happen when we select 256px as output? Similarly in DFL-SAE (64-256px input, 64-256px output) & Unbalanced (64-512px input, 64-512px o...
by ugramund
Fri Mar 12, 2021 2:27 pm
Forum: Swap Showcase
Topic: Linus sebastian - Obama Dlight Swap
Replies: 1
Views: 15158

Linus sebastian - Obama Dlight Swap

This video is not mine. I was just browsing the YouTube, going through some tutorials & found this. Thought it may be of no harm if I share it here. Video is very small of 9 seconds only. I think he will release the full version later. https://youtu.be/EEHKnOJ2JR0 This YouTuber WowEpic released ...
by ugramund
Fri Mar 12, 2021 4:32 am
Forum: Swap Showcase
Topic: Original Model can still produce great results!
Replies: 5
Views: 21872

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. ...
by ugramund
Tue Mar 09, 2021 3:54 am
Forum: Training Discussion
Topic: Can interchanging of image sets A & B between training help in better prediction?
Replies: 2
Views: 2397

Can interchanging of image sets A & B between training help in better prediction?

I was training with around 300+ images for 2 different persons on sides A & B for some time but noticed that some faces of side B were not at all predicted well on side A in the preview window because those facial angles & expressions were missing in my B image set. Then I interchanged my se...
by ugramund
Mon Mar 08, 2021 7:28 am
Forum: Swap Showcase
Topic: Original Model can still produce great results!
Replies: 5
Views: 21872

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 ...