MrDeepFakes Forums
  • New and improved dark forum theme!
  • Guests can now comment on videos on the tube.
   
tutsmybarreh[GUIDE] - DeepFaceLab 2.0 EXPLAINED AND TUTORIALS (recommended)
#21
(02-12-2020, 02:32 AM)Allesdrop Wrote: You are not allowed to view links. Register or Login to view.DFL 2.0 is 03.02.2020 version?

2.0 is from 22.01.2020. 03.02.2020 is the newest stable release.
At the end of the guide there is a changelog and it contains all the update notes.
Raising money for a new GPU, if you enjoy my fakes or my work on forums consider donating via bitcoin, tokens or paypal/patreon, any amount helps!
Paypal/Patreon: You are not allowed to view links. Register or Login to view.
Bitcoin: 1C3dq9zF2DhXKeu969EYmP9UTvHobKKNKF
Want to request a paid deepfake or have any questions reagarding the forums or deepfake creation using DeepFaceLab? Write me a message.
TMB-DF on the main website - You are not allowed to view links. Register or Login to view.
#22
I got some weird issues with a couple of tries. The first try I got the orange spot, the second try I got this: You are not allowed to view links. Register or Login to view.

I used the same settings from the 1.0 guide on the first try and the second try I just used default settings, but here's the settings:



Code:
================== Model Summary ==================
==                                               ==
==            Model name: Tint01_SAEHD           ==
==                                               ==
==     Current iteration: 92598                  ==
==                                               ==
==---------------- Model Options ----------------==
==                                               ==
==            resolution: 128                    ==
==             face_type: f                      ==
==     models_opt_on_gpu: False                  ==
==                 archi: df                     ==
==               ae_dims: 256                    ==
==                e_dims: 64                     ==
==                d_dims: 64                     ==
==           d_mask_dims: 22                     ==
==            learn_mask: True                   ==
==            lr_dropout: False                  ==
==           random_warp: True                   ==
==             gan_power: 0.0                    ==
==       true_face_power: 0.0                    ==
==      face_style_power: 10.0                   ==
==        bg_style_power: 10.0                   ==
==               ct_mode: none                   ==
==              clipgrad: False                  ==
==              pretrain: False                  ==
==       autobackup_hour: 0                      ==
== write_preview_history: False                  ==
==           target_iter: 0                      ==
==           random_flip: True                   ==
==            batch_size: 8                      ==
==                                               ==
==----------------- Running On ------------------==
==                                               ==
==          Device index: 0                      ==
==                  Name: GeForce RTX 2070 SUPER ==
==                  VRAM: 8.00GB                 ==
==                                               ==
===================================================


I have the latest video card drivers, Here are my system specs:

GPU: Nvidia RTX 2070 Super
CPU: AMD 3900x
RAM: 32GB @3600 MHz
#23
(02-12-2020, 11:23 AM)whatsis Wrote: You are not allowed to view links. Register or Login to view.I got some weird issues with a couple of tries. The first try I got the orange spot, the second try I got this: You are not allowed to view links. Register or Login to view.

I used the same settings from the 1.0 guide on the first try and the second try I just used default settings, but here's the settings:



Code:
================== Model Summary ==================
==                                               ==
==            Model name: Tint01_SAEHD           ==
==                                               ==
==     Current iteration: 92598                  ==
==                                               ==
==---------------- Model Options ----------------==
==                                               ==
==            resolution: 128                    ==
==             face_type: f                      ==
==     models_opt_on_gpu: False                  ==
==                 archi: df                     ==
==               ae_dims: 256                    ==
==                e_dims: 64                     ==
==                d_dims: 64                     ==
==           d_mask_dims: 22                     ==
==            learn_mask: True                   ==
==            lr_dropout: False                  ==
==           random_warp: True                   ==
==             gan_power: 0.0                    ==
==       true_face_power: 0.0                    ==
==      face_style_power: 10.0                   ==
==        bg_style_power: 10.0                   ==
==               ct_mode: none                   ==
==              clipgrad: False                  ==
==              pretrain: False                  ==
==       autobackup_hour: 0                      ==
== write_preview_history: False                  ==
==           target_iter: 0                      ==
==           random_flip: True                   ==
==            batch_size: 8                      ==
==                                               ==
==----------------- Running On ------------------==
==                                               ==
==          Device index: 0                      ==
==                  Name: GeForce RTX 2070 SUPER ==
==                  VRAM: 8.00GB                 ==
==                                               ==
===================================================


I have the latest video card drivers, Here are my system specs:

GPU: Nvidia RTX 2070 Super
CPU: AMD 3900x
RAM: 32GB @3600 MHz

You have gradient clipping disabled so your model might have collapsed. Make sure to read this guide, I mention what each feature does.
If you don't have any backup I'm afraid your model is now useless. Or maybe it's some issue with 2.0 - try disabling style powers and learn mask, enable random warp, drop batch to 4, set models_opt_on_gpu: to True so it's trains super fast and enable gradient clipping, then let it train for a while and see if faces start to appear, if they do then there is some issue with one of the features, if they don't - model is collapsed (preview might even turn completely white/black) and in that case you have to start all over again.
Raising money for a new GPU, if you enjoy my fakes or my work on forums consider donating via bitcoin, tokens or paypal/patreon, any amount helps!
Paypal/Patreon: You are not allowed to view links. Register or Login to view.
Bitcoin: 1C3dq9zF2DhXKeu969EYmP9UTvHobKKNKF
Want to request a paid deepfake or have any questions reagarding the forums or deepfake creation using DeepFaceLab? Write me a message.
TMB-DF on the main website - You are not allowed to view links. Register or Login to view.
#24
(02-12-2020, 12:55 PM)tutsmybarreh Wrote: You are not allowed to view links. Register or Login to view.
(02-12-2020, 11:23 AM)whatsis Wrote: You are not allowed to view links. Register or Login to view.I got some weird issues with a couple of tries. The first try I got the orange spot, the second try I got this: You are not allowed to view links. Register or Login to view.

I used the same settings from the 1.0 guide on the first try and the second try I just used default settings, but here's the settings:



Code:
================== Model Summary ==================
==                                               ==
==            Model name: Tint01_SAEHD           ==
==                                               ==
==     Current iteration: 92598                  ==
==                                               ==
==---------------- Model Options ----------------==
==                                               ==
==            resolution: 128                    ==
==             face_type: f                      ==
==     models_opt_on_gpu: False                  ==
==                 archi: df                     ==
==               ae_dims: 256                    ==
==                e_dims: 64                     ==
==                d_dims: 64                     ==
==           d_mask_dims: 22                     ==
==            learn_mask: True                   ==
==            lr_dropout: False                  ==
==           random_warp: True                   ==
==             gan_power: 0.0                    ==
==       true_face_power: 0.0                    ==
==      face_style_power: 10.0                   ==
==        bg_style_power: 10.0                   ==
==               ct_mode: none                   ==
==              clipgrad: False                  ==
==              pretrain: False                  ==
==       autobackup_hour: 0                      ==
== write_preview_history: False                  ==
==           target_iter: 0                      ==
==           random_flip: True                   ==
==            batch_size: 8                      ==
==                                               ==
==----------------- Running On ------------------==
==                                               ==
==          Device index: 0                      ==
==                  Name: GeForce RTX 2070 SUPER ==
==                  VRAM: 8.00GB                 ==
==                                               ==
===================================================


I have the latest video card drivers, Here are my system specs:

GPU: Nvidia RTX 2070 Super
CPU: AMD 3900x
RAM: 32GB @3600 MHz

You have gradient clipping disabled so your model might have collapsed. Make sure to read this guide, I mention what each feature does.
If you don't have any backup I'm afraid your model is now useless. Or maybe it's some issue with 2.0 - try disabling style powers and learn mask, enable random warp, drop batch to 4, set models_opt_on_gpu: to True so it's trains super fast and enable gradient clipping, then let it train for a while and see if faces start to appear, if they do then there is some issue with one of the features, if they don't - model is collapsed (preview might even turn completely white/black) and in that case you have to start all over again.

Thanks for the suggestions, I'll give it another shot and report back.
#25
(02-12-2020, 01:21 AM)tutsmybarreh Wrote: You are not allowed to view links. Register or Login to view.
(02-12-2020, 12:15 AM)androsk Wrote: You are not allowed to view links. Register or Login to view.the .txt file created in model, it is the config file? can i change values there and they will be loaded when start training again?

No, it's a summary file (as the name says) and you can't use it to change any values, resolution, model architecture, dims, face type, all of those settings cannot be changed after you start training, you set them once and that's it. All you can change after starting a training is:

Autobackup every N hour
Write preview history
Target iteration
Flip faces randomly
Batch_size
Learn mask
Place models and optimizer on GPU
Use learning rate dropout
Enable random warp of samples
GAN power
True face power
Face style power
Background style power
Color transfer for src faceset
Enable gradient clipping
Enable pretraining mode

thank you.
#26
(02-12-2020, 12:55 PM)tutsmybarreh Wrote: You are not allowed to view links. Register or Login to view.
(02-12-2020, 11:23 AM)whatsis Wrote: You are not allowed to view links. Register or Login to view.I got some weird issues with a couple of tries. The first try I got the orange spot, the second try I got this: You are not allowed to view links. Register or Login to view.

I used the same settings from the 1.0 guide on the first try and the second try I just used default settings, but here's the settings:



Code:
================== Model Summary ==================
==                                               ==
==            Model name: Tint01_SAEHD           ==
==                                               ==
==     Current iteration: 92598                  ==
==                                               ==
==---------------- Model Options ----------------==
==                                               ==
==            resolution: 128                    ==
==             face_type: f                      ==
==     models_opt_on_gpu: False                  ==
==                 archi: df                     ==
==               ae_dims: 256                    ==
==                e_dims: 64                     ==
==                d_dims: 64                     ==
==           d_mask_dims: 22                     ==
==            learn_mask: True                   ==
==            lr_dropout: False                  ==
==           random_warp: True                   ==
==             gan_power: 0.0                    ==
==       true_face_power: 0.0                    ==
==      face_style_power: 10.0                   ==
==        bg_style_power: 10.0                   ==
==               ct_mode: none                   ==
==              clipgrad: False                  ==
==              pretrain: False                  ==
==       autobackup_hour: 0                      ==
== write_preview_history: False                  ==
==           target_iter: 0                      ==
==           random_flip: True                   ==
==            batch_size: 8                      ==
==                                               ==
==----------------- Running On ------------------==
==                                               ==
==          Device index: 0                      ==
==                  Name: GeForce RTX 2070 SUPER ==
==                  VRAM: 8.00GB                 ==
==                                               ==
===================================================


I have the latest video card drivers, Here are my system specs:

GPU: Nvidia RTX 2070 Super
CPU: AMD 3900x
RAM: 32GB @3600 MHz

You have gradient clipping disabled so your model might have collapsed. Make sure to read this guide, I mention what each feature does.
If you don't have any backup I'm afraid your model is now useless. Or maybe it's some issue with 2.0 - try disabling style powers and learn mask, enable random warp, drop batch to 4, set models_opt_on_gpu: to True so it's trains super fast and enable gradient clipping, then let it train for a while and see if faces start to appear, if they do then there is some issue with one of the features, if they don't - model is collapsed (preview might even turn completely white/black) and in that case you have to start all over again.

An update on my situation:

I used these settings which fixed my issue:



Code:
================== Model Summary ==================
==                                               ==
==            Model name: Tint_SAEHD             ==
==                                               ==
==     Current iteration: 332627                 ==
==                                               ==
==---------------- Model Options ----------------==
==                                               ==
==            resolution: 128                    ==
==             face_type: f                      ==
==     models_opt_on_gpu: True                   ==
==                 archi: df                     ==
==               ae_dims: 256                    ==
==                e_dims: 64                     ==
==                d_dims: 64                     ==
==           d_mask_dims: 22                     ==
==            learn_mask: False                  ==
==            lr_dropout: False                  ==
==           random_warp: True                   ==
==             gan_power: 0.0                    ==
==       true_face_power: 0.0                    ==
==      face_style_power: 0.0                    ==
==        bg_style_power: 0.0                    ==
==               ct_mode: none                   ==
==              clipgrad: False                  ==
==              pretrain: False                  ==
==       autobackup_hour: 0                      ==
== write_preview_history: False                  ==
==           target_iter: 0                      ==
==           random_flip: True                   ==
==            batch_size: 4                      ==
==                                               ==
==----------------- Running On ------------------==
==                                               ==
==          Device index: 0                      ==
==                  Name: GeForce RTX 2070 SUPER ==
==                  VRAM: 8.00GB                 ==
==                                               ==
===================================================

The quality of the final result was not as good as I was hoping, but at least the issue is gone. I'm going to mess around with my settings until I find something good.

Another issue I'm having is that I cannot enable auto backup, when choosing the options for training it just jumps straight to which GPU to use, is there something I need to do to enable auto backup?
#27
(02-13-2020, 02:48 PM)whatsis Wrote: You are not allowed to view links. Register or Login to view.The quality of the final result was not as good as I was hoping, but at least the issue is gone. I'm going to mess around with my settings until I find something good.
Another issue I'm having is that I cannot enable auto backup, when choosing the options for training it just jumps straight to which GPU to use, is there something I need to do to enable auto backup?

Because you are using old version, make sure you use 03.02 release or manually update by downloading newest version from github and replacing files inside the _internal/DeepFaceLab folder.
Raising money for a new GPU, if you enjoy my fakes or my work on forums consider donating via bitcoin, tokens or paypal/patreon, any amount helps!
Paypal/Patreon: You are not allowed to view links. Register or Login to view.
Bitcoin: 1C3dq9zF2DhXKeu969EYmP9UTvHobKKNKF
Want to request a paid deepfake or have any questions reagarding the forums or deepfake creation using DeepFaceLab? Write me a message.
TMB-DF on the main website - You are not allowed to view links. Register or Login to view.
#28
I'm trying to use the "interactive converter preview" mode which shows hotkeys for different settings and stuff but absolutely nothing happens no matter which hotkey I press. And thus merging process stays on 0%. Any ideas?

Also I keep getting this error now whenever I try to merge [Image: 6mL3xGvh.png]

GTX 1060 6GB

And now I also keep getting E tensorflow/stream_executor/cuda/cuda_dnn.cc:334] Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR when I try to continue training my model. :/ Is my model RIP?

EDIT: Keep getting the same error even if I try creating a new model from scratch. Wtf?
#29
(02-15-2020, 01:47 AM)lewutt Wrote: You are not allowed to view links. Register or Login to view.I'm trying to use the "interactive converter preview" mode which shows hotkeys for different settings and stuff but absolutely nothing happens no matter which hotkey I press. And thus merging process stays on 0%. Any ideas?

Also I keep getting this error now whenever I try to merge [Image: 6mL3xGvh.png]

GTX 1060 6GB

And now I also keep getting  E tensorflow/stream_executor/cuda/cuda_dnn.cc:334] Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR when I try to continue training my model. :/ Is my model RIP?

Try quickly starting a new SAEHD/Quick96 model and see if merger works with them. That way we can rule out if it's the issue with merger/converter or your model.
Is this the newest DFL version?
Raising money for a new GPU, if you enjoy my fakes or my work on forums consider donating via bitcoin, tokens or paypal/patreon, any amount helps!
Paypal/Patreon: You are not allowed to view links. Register or Login to view.
Bitcoin: 1C3dq9zF2DhXKeu969EYmP9UTvHobKKNKF
Want to request a paid deepfake or have any questions reagarding the forums or deepfake creation using DeepFaceLab? Write me a message.
TMB-DF on the main website - You are not allowed to view links. Register or Login to view.
#30
Hello, I am new here and have successfully swapped a face from a source to a destination video which is a solo video, thanks to this tutorial!

I am wondering how to swap one face if there are two faces in a destination video.

Forum Jump:

Users browsing this thread: jjay8787, ozingabuldak, 7 Guest(s)