(unofficial) DFL 2.0 GUI (user interface)

[Roy_B]

One issue I have tho, is that I can't change anything under the SAEHD button, everything is greyed. I would especially like to change the "number of workers" parameters as it's 1 by default on your GUI, but I usually use 8 with standard DFL.

To enable parameters you have to select your model first, then click "command line mode" button to enable parameters change, then click on "saehd" to start training in "old" mode (without preview).

Thanks for the thread link, i'll take a look and see if can contribute or have new idea to share

 

[Nickrus]

There's no file for language at all, nor the ability to create one (to my knowledge): everything is embedded within the Visual Studio code itself

You need to export your stings to an XML resource file and let the app load it on startup.
Then you can let others translate this XML into other languages and add them to our setup.
It takes effort but it is definitely possible to implement. You're not in the first 1000 of the developers who had this task.

 

[Roy_B]

You need to export your stings to an XML resource file and let the app load it on startup.
Then you can let others translate this XML into other languages and add them to our setup.
It takes effort but it is definitely possible to implement. You're not in the first 1000 of the developers who had this task.

Easy to say but complicated to do: atm i can't find the solution to export dedicated files for text parts (xml etc.): both for "button text" part and especially the "tooltips" part that use different coding sources.

Keep in mind that this GUI use different libraries instead of the "official" microsoft ones: krypton libraries are more complete but at the same time are more complicated to manage (at least for my very basic basic coding knowledge), so it gets more tricky than it seems on paper...

Regarding the "1000 developers" thing: as i said in a previous post, i'm not a developer at all (this is an "hobby") and i don't have much time to spend for it, especially in this period. So i'll try at my best if i have spare time.

 

[Roy_B]

Updated version with some minor fixes and adjustments

 

[kaemen]

Does this GUI work with deepface lab Opencl 1.3?

 

[Roy_B]

Does this GUI work with deepface lab Opencl 1.3?

It works only on DFL 2.0

 

[kaemen]

Does this GUI work with deepface lab Opencl 1.3?

It works only on DFL 2.0

Is there an GUI for the Opencl version?

 

[Roy_B]

There are other GUI in this forum section for older DFL versions. I don't know if they works with opencl but you can try for yourself

 

[lennygordon]

This is looking AWESOME!
Do you happen to have a video tutorial on how to do DST face alignment fixing for existing projects?

 

[Roy_B]

This is looking AWESOME!
Do you happen to have a video tutorial on how to do DST face alignment fixing for existing projects?

There's a video tutorial in DFL 2.0 official guide here. In alternative take a look at this topic for alignment fixing with Machine Video Editor

 

[Roy_B]

Hi,

there is a new DFL-GUI update. There are several changes, including: 

- compatibility with DeepFaceLab update 03/23/2021
- customization for datasets, models, videos and backups
- modified interface for better compatibility with high resolutions
- various optimizations/fixes to console, functions and utilities

For any bug, ideas etc. please send me a feedback

 

[Roy_B]

New 06/19/2021update:

-Compatibility with DeepFaceLab June 6th 2021 update: added support for AMP model
-Modified and fixed XSeg interface: added function for custom SRC/DST training


Cheers!


EDIT: If you are used to have multiple config files, at every DFL update also GUI configurations are modified. So your old DFL-GUI config files doesn't work in the new version.
Backup them or extract/run the new gui from a different folder (without "DFL-GUI.ini"): you can use the config editor (in "utilities" section) to apply your custom parameters into the gui, then save them to a new .ini file.

 

[Shkeke]

I am running as administrator but I can't get this thing to work, its command line interface is full of errors, it freezes a lot and makes my other programs freeze and act up as well.

---

I am running as administrator but I can't get this thing to work, its command line interface is full of errors, it freezes a lot and makes my other programs freeze and act up as well.

Oh ignore that it turns out I have to restart it for it to work

 

[Roy_B]

New update:

- Compatibility with DeepFaceLab update October 20, 2021
- Changed timings during parameters entry to avoid errors (especially on slow PCs)
- Added some extra features like 1-click workspace change, siltent start and more

 

[Roy_B]

Update 20.11.2021: fixed (again) parameter sequence entries for training

 

[CptCrunch83]

This is absolutely awesome, mate. Thanks a lot.

 

[NoWayMan]

Another graphical user interface for Deep Face Lab 2.0, made with Visual Studio. All DFL features are present, including additional tooltips with useful info for every function and parameter settings.

 
 

UPDATE: 21.11.2021, compatibility with Deep Face Lab build 20.11.2021  ►  REQUIREMENTS:   |  DOWNLOAD:   |  AV CHECK:   |  SEE README FOR DETAILS

What a great job, impressive.

Imagine being able to automate the various stages of training, starting it only once.

For example it would be useful for DFL to start for 10k iterations RW = on, then between 10k and 100k toggle LRD and disable RW. Finally between 100k and 150k activates gan 0.10.

In this way it would no longer be necessary to pay attention to the training process: one start, one project. Maybe one solution can be to create option1.txt, 2.txt, etc... each one with different option and a launcher that tell to run XK iteration for 1.txt, than run Yk for 2.txt etc...

Any ideas for doing this?

 

[Roy_B]

It is something i wanted to do, but i think you need to intervene directly on the DFL code, which i can't because of my poor coding knowledge...
the only thing i could do would be some crappy "time-based" commands like:

- start training with default settings for hh:mm:ss, stop
- apply EMP for a certain time, stop
- disable EMP, enable UY...stop
- etc.


The ideal thing would be create functions based on iteration numbers or even better, based on loss values  :-/

 

[Roy_B]

New update

 

[Roy_B]

Fixed: Extraction (Cleanup Faceset, DST), Training (Silent Start)