Local implementation of text-to-3D using Shap-E

The Shap-E-local project provides a way to run text-to-3D sample code locally, enabling the generation of 3D models offline and boasting ample GPU power. The project includes prompts for generating 3D models of an apple, shark, and table which can be imported into Blender for rendering. The process involves downloading the zip folder, creating a virtual environment, installing required libraries, testing Torch and Pytorch3D versions, and finally generating a 3D model. The project also provides commands for previewing the generated 3D models using required libraries. The demo was run on Ubuntu 20.04.5 LTS with an Intel Core i7-6800K CPU, NVIDIA GeForce RTX 3090 GPU, Cuda 12.1, pip 23.1.2, and conda 23.5.0.


To top