Search papers, labs, and topics across Lattice.
This paper introduces Block3D, a novel block-wise diffusion framework for efficient text-to-3D generation that partitions shape-token sequences into contiguous blocks, allowing for autoregressive generation while jointly denoising tokens within each block. By implementing confidence-guided intra-block correction, the method effectively mitigates error accumulation, leading to significant improvements in generation speed. On the TRELLIS-500K dataset, Block3D achieves a remarkable 5.15x speedup in end-to-end generation time compared to traditional autoregressive methods, all while maintaining high geometric fidelity.
Block3D slashes text-to-3D generation time from over 25 seconds to under 5, revolutionizing efficiency without compromising quality.
While text-to-3D generation has advanced rapidly, achieving high geometric fidelity at low inference cost remains challenging. Existing text-to-3D methods either decode discrete shape tokens autoregressively or iteratively refine global 3D representations with diffusion or flow models. However, autoregressive decoding is sequential and cannot revise errors, whereas diffusion and flow-matching models repeatedly process the full representation, making high-quality generation increasingly expensive. In this paper, we propose Block3D, a block-wise diffusion framework that partitions the discrete shape-token sequence into contiguous blocks, generates the blocks autoregressively, and jointly denoises all tokens within the current block. To alleviate error accumulation, we introduce confidence-guided intra-block correction, which revises low-confidence tokens before each block is finalized. On a held-out set from TRELLIS-500K, Block3D reduces mean end-to-end generation time from 25.71 seconds to 4.99 seconds, achieving a $5.15\times$ speedup over the fine-tuned autoregressive baseline without sacrificing geometric fidelity.
