OpenAI has developed a new model called Shap-E to generate 3D objects from text, similar to how Dall-E can create 2D images. This generative model was designed to enable users to generate 3D assets and neural radiance fields. Shap-E is trained on a combination of mapping 3D assets and a diffusion model.
OpenAI offers Shap-E as a free-to-run program on Microsoft Paint 3D as well as convert it into an STL file which allows creating 3D prints. However, installing and setting up Shap-E can be challenging for some. Avram Piltch, the editor-in-chief of a publication, had to invest 8 hours of his time in order to understand how to use the Shap-E model.
Shap-E is compatible only with Nvidia GPUs and requires high-performance CPUs to render assets quickly – in a matter of minutes, not hours. Users can upload 2D images and give prompts such as shark, a Minecraft creeper or an airplane that looks like a banana, to render objects in different formats and with varying degrees of resolution.
OpenAI is a San Francisco-based artificial intelligence company which is wide-known for its groundbreaking research. It was co-founded in 2015 by tech-entrepreneurs including Elon Musk, Reid Hoffman, and Greg Brockman. The OpenAI team works towards the development of “safe general artificial intelligence” for benefiting humanity in the fields of medicine, economics, etc. OpenAI’s research is based on unsupervised training of deep neural networks in order to improve our understanding of general AI.
