Lunar is a neural network aimbot that uses real-time object detection accelerated with CUDA on Nvidia GPUs.
Lunar can be modified to work with a variety of FPS games; however, it is currently configured for Fortnite. Besides being general purpose, the main advantage of using Lunar is that it is virtually undetectable by anti-cheat software (no memory is meddled with).
The basis of Lunar's player detection is the YOLOv5 architecture written in the PyTorch framework.
A demo video (outdated) can be found here.
I have sold a closed-source version of this project and will no longer be updating this repo. It is now being run by Lunar Enhancements.
-
Install a version of Python 3.8 or later.
-
Navigate to the root directory. Use the package manager pip to install the necessary dependencies.
pip install -r requirements.txt
python lunar.py
To update sensitivity settings:
python lunar.py setup
To collect image data for annotating and training:
python lunar.py collect_data
- The method of mouse movement (SendInput) is slow. For this reason, the crosshair often lags behind a moving detection. This problem can be lessened by increasing the pixel_increment (e.g. to 4) so fewer calls to that function are made.
- The model is trained on a dataset of Fortnite players, and it will not work well for other games. False positives can also happen under certain lighting conditions.
- There is a known issue that occurs with PyTorch and the GeForce 16 series GPUs on Windows. Unfortunately, if you are using one of these GPUs, the aimbot will not work for you.
- Train a model on a custom dataset
- Implement better player tracking
- Add support for other FPS games (e.g. Valorant, Warzone)
- Use TensorRT for faster inference
- Make an easy-to-use GUI
Pull requests are welcome. If you have any suggestions, questions, or find any issues, please open an issue and provide some detail. If you find this project interesting or helpful, please star the repository.