Skip to content
Preprint

Block3D: Efficient Text-to-3D Generation via Block-Wise Diffusion

Aug 2026 · 1 citation · 61 references
Computer Science

TL;DR

Block3D is proposed, 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 and introduces confidence-guided intra-block correction, which revises low-confidence tokens before each block is finalized.

Abstract

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.

View source

Similar papers

Preprint Sep 2026

ZipTok3D: High-Fidelity 3D Tokenization with Compact Token Prefixes

Compact token sequences are essential for efficient 3D generation. However, existing 3D tokenizers typically organize latent representations either over spatial regions or as fixed-size sets of global tokens, both suffering sharp reconstruction degradation when compressed to extremely low token budgets. In this paper,...

Mingda Lin, Weijie Wang, Zeyu Zhang et al. · 0 citations
#artificial intelligence Preprint Sep 2026

DraftAttention2: Fast Video Diffusion with Low-Resolution-Guided Mixed-Precision Attention

DraftAttention2 is presented, a training-free framework that uses the low-resolution draft attention map to jointly select attention blocks and assign their numerical precision, and achieves a superior quality-efficiency trade-off over existing efficient video generation methods.

Rui Ding, Haopeng Li, Wei-Ze Ma et al. · 0 citations
Preprint Aug 2026

NoisEasier: Test-Time Noise Optimization for Text-to-Video Generation

NoisEasier serves as both a flexible alternative and a complementary enhancement to reward-based fine-tuning, establishing test-time scaling as an effective paradigm for controllable text-to-video generation.

Yu-Jiang Pu, Yu Kong · 1 citation
Review Aug 2026

MLLM-Guided Semantic Correction for Text-to-Video Generation

A training-free, interpretable mid-generation correction framework that integrates multimodal large language model (MLLM) feedback directly into the diffusion sampling loop and achieves diffusion trajectory correction by injecting semantic evaluation signals during video synthesis, enabling the model to optimize the ge...

Jun-Hao Chen, Zheqi Lv, Keting Yin et al. · 0 citations
Preprint Aug 2026

Spatially-Grounded Text-to-Video Generation via Inference-Time Gradient-Free Optimization

This work presents Gradient-free Analytical Trajectory Optimization Video Generation (GATO-Vid), a novel training-free and gradient-free approach for precise spatial guidance that significantly outperforms existing baselines in localization accuracy while introducing minimal computational overhead.

Guillaume Jeanneret, Mathis Koroglu, H. Caselles-Dupré et al. · 0 citations
Aug 2026

Text-to-Motion Generation With Discrete Representations and Large Language Models.

This work investigates a simple yet effective conditional generative framework for text-to-motion generation and proposes T2M-GIT+, which employs a non-autoregressive method to generate discrete motion representations in parallel, and is therefore more efficient than T2M-GPT+ while achieving comparable results.

Jian-Rong Zhang, Yang-Song Zhang, Xiaodong Cun et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.