For all those who want to create animated deepfaked gifs/videos
Follow this Github tutorial
Stable-Diffusion/Tutorials/1-Click-DeepFake-Tutorial.md at main · FurkanGozukara/Stable-Diffusion
Check out my thread "Alia Bhatt Animated Gifs (Hardcore) - Bollywood Actress -" to see what level of deepswaps can you do with this tool.
The github tutorial has everything that you need to know to successfully install it.
I recommend firstly watch this youtube tutorial link to understand and get an overview of the whole tool and how to use it.
Note: Only watch the video to know about the working of the tool, don't follow the installing instructions there, the video is a bit outdated, but is still the best explanation of the tool.
Follow the github tutorial for installation. (it is updated and covers everything)
(You can skip watching the video if you want to, it's not necessary to watch, but recommended to understand the tool)
After you've watched the video(or skipped it) follow the Github Tutorial properly.
Make sure you do every single installation as it is mentioned in there, I cannot stress this enough.
Make sure you do every single installation as it is mentioned in there, I cannot stress this enough.
VERY IMPORTANT STEP
Now, the main thing, the original tool blocks NSFW(nude) content by default.
So, in Step 6 of installation(as mentioned in github readme), instead of the original one,
either use this repository
"GitHub - hassan-sd/roop-unlocked: one-click deepfake (face swap)"
or use this repository
"GitHub - C0untFloyd/roop-unleashed: Uncensored Fork of roop with many enhancements"
(roop unleashed has features to select the faces out of multiple ones, so I'd recommend this one)
The code becomes:
git clone <link for any of the above two github repositories> . . . other commands |
How to run the tool efficiently so that your PC doesn't lag so much?
In Step 8,
when you are creating a "start.bat" file to quick start the tool.
edit the final line of the code from this
python run.py --keep-frames --keep-fps --execution-provider cuda |
|
These values are recommended for a PC where,
- CPU has 8 threads (google your processor to know how many threads yours has)
- RAM is 16 GB
You don't need a really high end pc for this. A low level-mid level GPU will work perfectly fine.
It consumes a good amount of RAM, so make sure you have atleast 8 gb ram.
What to do if you get any errors?
The best place to find the solution is to look for that error in the Issues section of the github repository.
Chances are very high that someone else has already got this error and there is solution for that.
Issues · FurkanGozukara/Stable-Diffusion
Last edited: