A research team at the University of Chicago has introduced LL3M (Large Language 3D Modeler), an advanced language model for 3D modeling. It uses text commands to automatically generate Python code, allowing for the creation and editing of complex three-dimensional objects in the Blender program.

Unlike other AIs that only create simple 3D models, LL3M operates on the principle of co-creation. The model not only generates code but is also capable of "visual self-critique": it analyzes the result and makes corrections on its own. The user can provide feedback at any time, making high-level edits.

A key feature of LL3M is its ability to create readable and well-structured code with comments, which makes it easier to edit. The model is capable of generating both individual objects (such as a piano or a rocket) and entire complex scenes, including numerous details and their hierarchy.

ORIENT