Generative AI keeps growing exponentially, both in terms of fundamental breakthroughs as well as in regard to new interesting products. However, most of the generative AI art seems to be plagued with weird and undesirable artifacts.
1/5
1/5
One of the potential reasons could be due to the fact that most of the generative AI image models have been trained on 2D images - these models don't have any intrinsic understanding of spatial relationships between various image components.
2/5
2/5
In recently published work a group of @NVIDIA researchers has built Magic3D, a generative AI model that can create high-quality 3D textured mesh models from input text prompts.
3/5
3/5
It utilizes a coarse-to-fine strategy that leverages both low- and highresolution diffusion priors for learning the 3D representation of the target content. Magic3D synthesizes 3D content with 8× higher-resolution supervision than DreamFusion while also being 2× faster.
4/5
4/5
An Ars Technica article: arstechnica.com
Research paper: arxiv.org
Project homepage: deepimagination.cc
#MachineLearning #DeepLearning #ArtificialIntelligence #ML #DL #AI #GenerativeAI
5/5
Research paper: arxiv.org
Project homepage: deepimagination.cc
#MachineLearning #DeepLearning #ArtificialIntelligence #ML #DL #AI #GenerativeAI
5/5
Loading suggestions...