What is image-3d?

image-3d is a local web application that turns a photo or drawing into a 3D printable file. You upload an image, the app generates a 3D mesh from it, and you can preview the result in a browser based 3D viewer before downloading it as STL, 3MF, GLB, or OBJ, the common formats a 3D printer or slicer software expects. Right now the project ships with a mock generator by default, meaning it produces a fixed, predictable test shape rather than one that actually reflects your uploaded image. This lets the full pipeline, from upload through mesh cleanup to download, run and be tested even without a GPU. When you have an NVIDIA GPU with recent CUDA drivers, you can switch to a real image to 3D model called Hunyuan3D-2, which does generate a shape that matches the picture you provide. A third option, Pixal3D, is also supported and can add color and surface detail to the output. The project also supports feeding in multiple views of the same subject, such as front and back photos, automatically splitting a single character reference sheet into separate views, applying a four color mode for printers that support multi color output, and generating texture paint on the model surface, though that last feature needs an extra CUDA build step. Setup involves creating a Python virtual environment and installing dependencies, then starting the app with a provided script, after which it runs a small local web server you open in your browser. Configuration, such as which generator to use, the server address, and upload size limits, is controlled through environment variables. The project includes a REST API for creating generation jobs, checking their status, and fetching the resulting files, along with a test suite covering the mesh processing, the API, and the color and texture handling. Minimum GPU memory for the real generator is documented, ranging from about 16GB for basic shape generation to 32GB when texture painting is enabled. The full README is longer than what was shown.