The company responsible for ChatGPT, OpenAI, has made strides toward converting text into 3D objects. The next step was the artificial intelligence model shape with open source.
Image source: github.com/openai
The preliminary version was released in December – at that time it was called Point-E and allowed to create simple 3D models in the form of clouds of points. Shap-E is significantly faster and generates models as “implicit functions” – machine learning helped the system create 3D models from 2D images. These objects can later work with other applications.
Essentially, Shap-E is an AI model that responds to natural language, much like ChatGPT, only acting as a 3D artist – it generates buildings, sculptures, machine parts and much more; In addition, the user does not even have to touch the mouse and other manipulators used in the creation of graphics. The results of his work will eventually appear both in virtual worlds and in 3D printing. The latter will be useful in the near future, when the quality of the models will improve, but at the moment they look quite modest, and some even primitive.
Add Comment