An Instagram Bot serving as an account, people can use to create DeepFakes on Instagram. The DeepFakes are created using First Order Model for Image Animation and the Faces are cropped using Ultra-Light-Fast-Generic-Face-Detector
Explain Video
Disclaimer:
The code needed a lot of configuring in order to work in my specific workplaces. I tried making it as easy as possible to run it yourself. Errors may occur! (Also this code only serves as a “show – off” and is not intended to be run since Instagram’s terms of services strictly prohibits this.) Run it at your own risk.
Prerequisites & Installation
- Code was written and tested in Python 3.6.5 and 3.6.10
- Install the requirements.txt pip install -r requirements.txt
- Install autocrop pip install autocrop
- locate autocrop (/usr/local/lib/python3.6/dist-packages/autocrop)
- replace entire folder with my updated autocrop
- deepfake.py requires numpy=1.15 -> we need a feature from numpy=1.16
- locate numpy-core (/usr/local/lib/python3.6/dist-packages/numpy/core)
- replace function_base.py with my updated numpy function_base.py
- Additional things:
- ffmpeg
- cuda
- opencv (I had to recompile it on Ubuntu 18.04.4) -> Install Opencv and follow this guide
- GPU (I used a GTX 1080 renting on vast.ai)
Running it
- Edit main.py on line 596 with your account details.
- python3.6 main.py
- Wait for the Instagram-Login to happen.
- Wait for Checker-,Editor-,and Sender-Class to be initialised.
Adding ranks to users
- Go to other_tools
- Run python3.6 add_donator username rank
- Rank 1: 30 seconds videos + priority queue
- Rank 2: 45 seconds videos + priority queue + watermark removal
- Rank 3: 60 seconds videos + priority queue + watermark removal
- (Every other user is initially Rank 0)
Checking stats
all stats can be found in log.json