You can now save your favourite tools. Register/Login to try.

A Generative Model of High Quality 3D Textured Shapes Learned from Images.

A Generative Model of High Quality 3D Textured Shapes Learned from Images.

0.0
Rated 0.0 out of 5
0.0 out of 5 stars (based on 0 reviews)
Excellent0%
Very good0%
Average0%
Poor0%
Terrible0%

About

As several industries are moving towards modeling massive 3D virtual worlds, the need for content creation tools that can scale in terms of the quantity, quality, and diversity of 3D content is becoming evident. In their work, they aim to train performant 3D generative models that synthesize textured meshes which can be directly consumed by 3D rendering engines, thus immediately usable in downstream applications. Prior works on 3D generative modeling either lack geometric details, are limited in the mesh topology they can produce, typically do not support textures, or utilize neural renderers in the synthesis process, which makes their use in common 3D software non-trivial. In this work, they introduce GET3D, a Generative model that directly generates Explicit Textured 3D meshes with complex topology, rich geometric details, and high fidelity textures. They bridge recent success in the differentiable surface modeling, differentiable rendering as well as 2D Generative Adversarial Networks to train their model from 2D image collections. GET3D is able to generate high-quality 3D textured meshes, ranging from cars, chairs, animals, motorbikes and human characters to buildings, achieving significant improvements over previous methods.

They generate a 3D SDF and a texture field via two latent codes. They utilize DMTet to extract a 3D surface mesh from the SDF, and query the texture field at surface points to get colors.

They train with adversarial losses defined on 2D images. In particular, they use a rasterization-based differentiable renderer to obtain RGB images and silhouettes. They utilize two 2D discriminators, each on RGB image, and silhouette, respectively, to classify whether the inputs are real or fake. The whole model is end-to-end trainable.

Leave your review

Recent Reviews

There are no reviews yet. Be the first one to write one.

Suggest an edit

Reccomended tools

Screenshot 2022-12-21 alle 16.49
Fliki
Turn text into videos with AI voices
Screenshot-2022-12-23-alle-00.11
Writesonic
Best AI Writer for Creating
Facebook
Twitter
LinkedIn
WhatsApp
Reddit
Pinterest