Generator Assets
Local headless pipeline that generates game assets, AI video, and music loops with Vulkan GGUF engines.
Tested on Linux
Use it
Linux x86_64
You need first
- Runtimes:
Python 3.12+
Clone real repo laurentvv/generator-assets
rm -rf generator-assets && git clone https://github.com/laurentvv/generator-assets.git generator-assetsRun unix installer with ffmpeg only, models and check skipped
cd generator-assets && bash scripts/install_unix.sh --skip-models --skip-check --engines ffmpeg
Where the README and reality differ
- README: git clone https://github.com/votre-compte/generator-assets.git
What worked: Clone https://github.com/laurentvv/generator-assets.git (task title URL); README quickstart uses https://github.com/votre-compte/generator-assets.git which does not exist
Check that it works
cd generator-assets && uv run python main.py --helpThe output should contain “3D & 2D Workflow Examples:”.
First thing to try
cd generator-assets && uv run python main.py -w flowmap "molten lava river" --angle 45From the generator-assets folder, run uv run python main.py -w <workflow> "<your prompt>" to start a generation workflow.
What it does
These come from the README; pickfit didn’t check them one by one.
- Generate AI videos and cinematic shotsNot tested
- Generate PBR textures and 3D meshesNot tested
- Generate seamless music loopsNot tested
- Generate isolated 2D game spritesNot tested
How far pickfit got
Linux x86_64
- Installs
- Starts
- Main function works
- Works offline
Windows and macOS: the project says it runs there; pickfit hasn’t tried it.
Good for
- Generate PBR texture pack for Godot
- Convert 2D image to volumetric 3D model
- Generate seamless background music loops
- Upscale images to 4K with ESRGAN
- Generate 360-degree skybox environments
An alternative to ComfyUI and Automatic1111.
How pickfit tested it
Each run starts from a clean machine and follows the steps above. How pickfit verifies
Linux x86_64
Its main function works
- Replayed from scratch
- Works offline
- Checked on 30 Oct 2026
Differences the test machine hit
- README: bash scripts/install_unix.sh with default packs base,onnx,upscalers and 5 engines
In the test: bash scripts/install_unix.sh --skip-models --skip-check --engines ffmpeg (syncs Python deps + only 150 MB ffmpeg; skipped 10+ GB base/onnx/upscalers model packs and sd-cli/llama.cpp/audio.cpp/trellis.cpp Vulkan binaries requiring GPU, disk and time beyond the box) - README: uv run python main.py -w material3d "runic ancient dungeon stone floor" -s 1024
In the test: Procedural README workflow flowmap exactly as table writes it: uv run python main.py -w flowmap "molten lava river" --angle 45 (pure numpy/Pillow, offline, ~2 s). Headline material3d example needs Flux.1 Dev Q6_K + sd-cli Vulkan GPU and was not run here. - README: Pinned engines verified by sha256
In the test: Installer prints No known sha256 for ffmpeg BtbN latest asset and skips integrity verification (moving direct URL).