Turn a reference photo of an object into a 3D model for game development.
Refine a rough 3D model with realistic textures that respond to lighting.
Add a skeleton to a 3D model so it can be animated in a browser.
Paint custom textures directly onto a 3D model using Blender.
| intisargit/aismith-3d | bytedtsinghua-sia/direct-opd | cheshiremew/price-action-analysis | |
|---|---|---|---|
| Stars | 54 | 54 | 54 |
| Language | Python | Python | Python |
| Setup difficulty | hard | hard | easy |
| Complexity | 4/5 | 5/5 | 2/5 |
| Audience | vibe coder | researcher | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires an NVIDIA CUDA GPU, ~18 GB of model downloads, and Blender 4.2 for texture painting.
AISmith 3D is a Windows application for creating 3D models from reference images. It runs entirely on your local machine, meaning your images and generated files never leave your computer. The app covers four main tasks: generating 3D geometry from an image, refining that geometry with better detail and textures, rigging a skeleton for animation, and painting textures directly on a model. The generation and refinement steps require an NVIDIA graphics card with CUDA support. These steps use AI models to convert a flat image into a 3D object. Refinement adds surface detail and PBR textures, which are textures that respond realistically to lighting. The rigging and animation step works in a web browser and does not need a graphics card or any 3D software. Texture painting requires Blender 4.2 or later. If your machine lacks the right GPU, the app clearly tells you when a feature will not work rather than failing silently. Installation is handled through a single PowerShell script. You clone the repository, run the start script with an install flag, and it sets up everything: a Python environment, frontend dependencies, and hardware detection. The app then opens in your browser at a local address. The project does not bundle the large AI model files with the code. Instead, you download them from within the app interface when you choose to use a specific workflow. The generation models take about 9.6 GB of disk space, and refinement models take about 8.8 GB. The app keeps heavy processing isolated in short-lived background processes to manage memory. It passes artifact identifiers between its browser interface and backend rather than moving large 3D files through browser memory. The original application code is released under the MIT license, though it integrates several third-party projects under their own terms. Model weights are downloaded separately and may carry their own usage restrictions.
A Windows app that turns 2D images into 3D models you can refine, rig, and paint, all locally on your machine. Generation requires an NVIDIA GPU, rigging works in any browser.
Mainly Python. The stack also includes Python, PowerShell, CUDA.
The app code is free to use for any purpose under MIT, but the AI model weights you download separately may have their own restrictions.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
Verify against the repo before relying on details.