Multi-scale Latent Point Consistency Models for 3D Shape Generation

TL;DR

This paper introduces a multi-scale latent point consistency model for 3D shape generation that significantly speeds up sampling and outperforms existing diffusion models in quality and diversity.

Contribution

It proposes a hierarchical latent diffusion framework with multi-scale integration and a latent consistency model for efficient 3D shape synthesis.

Findings

Achieves 100x faster sampling speed.

Outperforms state-of-the-art diffusion models in shape quality.

Demonstrates superior diversity in generated shapes.

Abstract

Consistency Models (CMs) have significantly accelerated the sampling process in diffusion models, yielding impressive results in synthesizing high-resolution images. To explore and extend these advancements to point-cloud-based 3D shape generation, we propose a novel Multi-scale Latent Point Consistency Model (MLPCM). Our MLPCM follows a latent diffusion framework and introduces hierarchical levels of latent representations, ranging from point-level to super-point levels, each corresponding to a different spatial resolution. We design a multi-scale latent integration module along with 3D spatial attention to effectively denoise the point-level latent representations conditioned on those from multiple super-point levels. Additionally, we propose a latent consistency model, learned through consistency distillation, that compresses the prior into a one-step generator. This significantly improves sampling efficiency while preserving the performance of the original teacher model. Extensive experiments on standard benchmarks ShapeNet and ShapeNet-Vol demonstrate that MLPCM achieves a 100x speedup in the generation process, while surpassing state-of-the-art diffusion models in terms of both shape quality and diversity.