This is a fork of Apple’s ml-sharp repository.
What this fork adds:
- Apple Silicon (MPS) fallback for the included demo video generation.
- A trivial web app to upload a single image and generate a left-to-right “swipe” video.
Upstream project page: https://apple.github.io/ml-sharp/
Install dependencies:
pip install -r requirements.txt -r requirements-web.txtTo test the installation, run
sharp --helpTo run prediction (upstream behavior):
sharp predict -i /path/to/input/images -o /path/to/output/gaussiansThe results will be 3D gaussian splats (3DGS) in the output folder. The 3DGS .ply files are compatible to various public 3DGS renderers. We follow the OpenCV coordinate convention (x right, y down, z forward). The 3DGS scene center is roughly at (0, 0, +z). When dealing with 3rdparty renderers, please scale and rotate to re-center the scene accordingly.
Additionally you can render videos with a camera trajectory.
- The CLI trajectory renderer (
sharp predict --render/sharp render) uses gsplat and is CUDA-only. - The web demo can still generate a swipe video on macOS/MPS (or CPU) via the fallback depth-parallax path.
sharp predict -i /path/to/input/images -o /path/to/output/gaussians --render
# Or from the intermediate gaussians:
sharp render -i /path/to/output/gaussians -o /path/to/output/renderingsThis repository also includes a tiny web app that accepts a single image and generates a left-to-right swipe trajectory video using SHARP.
- On CUDA: predicts 3D Gaussians and renders a true camera trajectory via gsplat.
- On macOS/MPS (or CPU): falls back to a depth-based parallax warp driven by SHARP's monodepth sub-network.
Run the server:
sharp webThen open http://127.0.0.1:8000 in your browser.
Our codebase is built using multiple opensource contributions, please see ACKNOWLEDGEMENTS for more details.
Please check out the repository LICENSE before using the provided code and LICENSE_MODEL for the released models.