IM-3D: Iterative Multiview Diffusion and Reconstruction for High-Quality 3D Generation
This addresses the need for faster and higher-quality 3D asset generation from text, with incremental improvements over existing methods.
The paper tackles the problem of slow and artifact-prone text-to-3D generation by introducing IM-3D, which uses video generators for multi-view synthesis and Gaussian splatting for reconstruction, reducing 2D generator evaluations by 10-100x and improving quality and efficiency.
Most text-to-3D generators build upon off-the-shelf text-to-image models trained on billions of images. They use variants of Score Distillation Sampling (SDS), which is slow, somewhat unstable, and prone to artifacts. A mitigation is to fine-tune the 2D generator to be multi-view aware, which can help distillation or can be combined with reconstruction networks to output 3D objects directly. In this paper, we further explore the design space of text-to-3D models. We significantly improve multi-view generation by considering video instead of image generators. Combined with a 3D reconstruction algorithm which, by using Gaussian splatting, can optimize a robust image-based loss, we directly produce high-quality 3D outputs from the generated views. Our new method, IM-3D, reduces the number of evaluations of the 2D generator network 10-100x, resulting in a much more efficient pipeline, better quality, fewer geometric inconsistencies, and higher yield of usable 3D assets.