AI for Text-to-3D Image Synthesis
This AI transforms text into 3D models using a pre-trained 2D text-to-image diffusion model, enabling visualization from any angle and lighting condition. It doesn’t need 3D training data or modifications to the image diffusion model.
Pricing
Conversion
For area
Platform
Category
This AI leverages a pre-trained 2D text-to-image diffusion model to execute text-to-3D synthesis. It introduces a loss based on probability density distillation that allows the use of a 2D diffusion model as a prior for optimizing a parametric image generator. This AI optimizes a randomly-initialized 3D model using DeepDream-like procedure. The final 3D model can be viewed from any angle, illuminated in any way or composited into any 3D environment. This AI requires no 3D training data and no modifications to the image diffusion model.
Reviews
There are no reviews yet.