About
TRELLIS is a cutting-edge 3D generative model that leverages a unified Structured Latent representation and Rectified Flow Transformers to create versatile and high-quality 3D assets. This innovative approach allows for decoding into various output formats, including Radiance Fields, 3D Gaussians, and meshes. It is particularly useful for researchers and developers in the fields of computer graphics, game design, and virtual reality, enabling them to generate detailed 3D models from text or image prompts. With the ability to manipulate local regions of 3D assets and generate variants, TRELLIS offers unprecedented flexibility in 3D asset creation. The model has been trained on a large dataset of diverse objects, ensuring high-quality results that surpass existing methods. Code and data will be made available for further exploration and development.