A synthetic data generator for video caption pairs.
0.mp4
Medium shot, orbiting The Moon to the right. It's mid-day and the sky is clear. The background is an air museum with a few planes parked.
Simian creates synthetic data that is usable for generative video and video captioning tasks. The data consists of videos and captions. The videos are generated using Blender, a 3D modeling software.
This project is in open development phase and is not ready for public use. Many things are still being actively developed for an initial release. Use at your own risk.
-
Install Python 3.10. If you're on Linux, set it up with this (gist)[https://gist.github.com/lalalune/986704a935d202ab2350ca90b2fc9755]
-
Download the datasets:
./scripts/get_data.sh
- [OPTIONAL] If you're on a headless Linux server, install Xorg and start it:
sudo apt-get install xserver-xorg -y && \
sudo python3 scripts/start_x_server.py start
python3 simian/combiner.py --count 1000 --seed 42
You can generate individually:
# MacOS
/Applications/Blender.app/Contents/MacOS/Blender --background --python simian/render.py
# Linux
blender --background --python simian/render.py
## Kitchen sink
blender --background --python simian/render.py -- --width 1920 --height 1080 --combination_index 0 --output_dir ./renders --background_path ./backgrounds --start_frame 1 --end_frame 25
Configure the flags as needed:
--width
and--height
are the resolution of the video.--combination_index
is the index of the combination to render.--output_dir
is the directory to save the rendered video.--background_path
is the directory containing the background images.--start_frame
and--end_frame
are the start and end frames of the video.
Or generate all or part of the combination set using the batch.py
script:
python3 simian/batch.py --start_index 0 --end_index 1000 --width 1024 --height 576 --start_frame 1 --end_frame 25
Rendering can be distributed across multiple machines using the "distributed.py" and "worker.py" scripts.
First, make sure you have Redis set up
scripts/setup_redis.sh
Now, start your workers
celery -A simian.worker worker --loglevel=info
Now issue work to your task queue
python3 simian/distributed.py --start_index 0 --end_index 10 --width 1024 --height 576
If you want to use a custom or hosted Redis instance (recommended), you can add th redis details like this:
EXPORT REDIS_URL=<my_redis_url>
We are currently using the following datasets: Objaverse
Backgrounds are loaded from: Poly Haven
We welcome contributions! We're especially interested in help adding and refining datasets, improving generation quality, adding new features and dynamics and allowing the project to meet more use cases.
- Check out the issues here: https://github.com/RaccoonResearch/simian/issues
- Join our Discord here: https://discord.gg/JMfbmHdPNB
- Get in touch with us so we can coordinate on development.
- Or, you know, just YOLO a pull request. We're pretty chill.
This project is licensed under the MIT License - see the LICENSE file for details.
If you use it, please cite us:
@misc{Simian,
author = {Raccoon Research},
title = {Simian: A Synthetic Data Generator for Video Caption Pairs},
year = {2024},
publisher = {GitHub},
howpublished = {\url{https://github.com/RaccoonResearch/simian}}
}