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× speedup over the fine-tuned autoregressive baseline without sacrificing geometric fidelity.
Figures & tables
Figure 2: Comparison of 3D token generation schedules. (a) Token-wise generation predicts one shape token at a time and cannot revise emitted tokens. (b) Full-sequence denoising updates all positions in parallel but repeatedly processes the complete representation. (c) Block3D generates blocks from left to right and denoises all positions in the current block in parallel; a filled token can still be corrected before its block is finalized.
Figure 3: Block3D overview. Text and optional bounding-box conditions guide left-to-right block generation. Completed blocks form a frozen causal prefix, while M2T and T2T [ 15 ] edit only the active block before the frozen decoder maps the completed codes to a mesh.
Figure 4: Qualitative Comparison of Text-to-3D Generation. Compared to existing paradigms, Block3D produces coherent front and back views in these examples, while several baselines exhibit missing or distorted geometry. Complete prompts and the selection protocol are documented in Appendix Section D.2.
Method
CD-L1 ↓
NC ↑
F@1% ↑
CLIP ↑
ShapeLLM-Omni
0.229
0.490
0.089
19.08
TRELLIS-text
0.222
0.496
0.090
20.41
AR3D-R1
0.145
0.583
0.162
21.92
Cube
0.094
0.632
0.219
23.87
Block3D
0.078
0.668
0.309
23.24
Table 1: Geometry and text-shape alignment on the 100-object TRELLIS-500K [ 2 ] evaluation set. F@1 denotes an F-score threshold of 1% [ 53 , 54 ] . Geometry metrics are rounded to three decimals and CLIPScore [ 55 ] to two decimals.
Method
Mean ↓
Median ↓
P90 ↓
Std. ↓
TRELLIS-text
11.65
9.75
20.06
5.59
ShapeLLM-Omni
36.89
33.69
40.64
8.48
AR3D-R1
79.80
82.48
94.19
17.34
Cube
25.71
25.43
26.68
0.80
Block3D
4.99
4.96
5.60
0.43
Table 2: End-to-end generation time in seconds over the 100 TRELLIS-500K [ 2 ] prompts, measured on one NVIDIA A100 80GB GPU.
Model
Time ↓
CD-L1 ↓
NC ↑
F@1% ↑
Block3D ( B=32 )
12.98
0.076
0.667
0.296
Block3D ( B=64 )
4.99
0.078
0.668
0.309
Block3D ( B=96 )
3.62
0.085
0.660
0.279
Block3D ( B=128 )
2.90
0.090
0.647
0.258
Block3D ( B=256 )
2.15
0.186
0.578
0.103
Table 3: Effect of block size on Block3D with four denoising steps per block. Time is reported in seconds. Every configuration is trained independently.
Model
Time ↓
CD-L1 ↓
NC ↑
F@1% ↑
Block3D ( T=4 )
4.99
0.078
0.668
0.309
Block3D ( T=8 )
7.33
0.079
0.665
0.300
Block3D ( T=12 )
9.37
0.084
0.668
0.284
Block3D ( T=20 )
13.66
0.078
0.676
0.303
Table 4: Effect of denoising steps on Block3D with a block size of 64. Time is reported in seconds. Every configuration is trained independently.
Model
CD-L1 ↓
NC ↑
F@1% ↑
CLIP ↑
Block3D (M2T)
0.081
0.666
0.287
22.74
Block3D (M2T+T2T)
0.078
0.668
0.309
23.24
Table 5: Effect of token-to-token editing [ 15 ] . Geometry metrics are rounded to three decimals and CLIPScore [ 55 ] to two decimals.
Appendix figures & tables6 assets
Supplementary material from the paper’s appendix.
Appendix
Figure A: Block-level visualization of the training attention mask. Green cells denote visible query–key group pairs. A corrupted query reads the condition, its clean teacher-forced prefix, and the corrupted copy of its own block; it cannot read the clean target of that block, a future block, or another corrupted block. The diagram expands each block to a single group for readability; the corresponding token-level rules are given in Equations A.3 – A.5 .
Figure B: Analytical diagnostics for the main configuration. Left: the deterministic quota reveals at least 16 positions per step, so the number of masks is bounded by 64→48→32→16→0 . Middle: Equation A.10 yields guidance coefficients (3.00,2.25,1.50,0.75) . Right: excluding the K cache-building calls, Block3D performs at most KT=64 conditional active-block evaluations, or 2KT=128 logical branches with CFG, whereas token-wise Cube performs 1,024 sequential token evaluations. These counts describe the generator schedule and are not substitutes for measured end-to-end latency.
Figure C: Recomputable profile of the fixed random100 evaluation manifest. The source counts sum to 100; prompt-length bins use the stored text verbatim; and anisotropy uses Equation B.1 . The distribution reflects the seeded TRELLIS-500K sample and is not manually balanced by source, prompt length, or shape extent.
Figure D: Four views of the manually assembled teaser arena. The layout, repeated placement of objects, display colors, lighting, and background are Blender presentation choices; Block3D generates the component meshes rather than the complete arena composition.
Figure F: Full-prompt version of the qualitative comparison in the main paper. Columns show Block3D, Cube, TRELLIS-text, AR3D-R1, and ShapeLLM-Omni; each result is rendered from front and back views.
Figure G: Additional Block3D text-to-shape results. Each group contains multiple viewpoints of a single generated mesh, with its complete text prompt shown directly below.
Continual learning enables models to acquire new knowledge over time while retaining previously learned capabilities. However, its application to text-to-3D generation remains unexplored. We present ReConText3D, the first framework for continual text-to-3D generation. We first demonstrate that existing text-to-3D models suffer from catastrophic forgetting under incremental training. ReConText3D enables generative models to incrementally learn new 3D categories from textual descriptions while preserving the ability to synthesize previously seen assets. Our method constructs a compact and diverse replay memory through text-embedding k-Center selection, allowing representative rehearsal of prior knowledge without modifying the underlying architecture. To systematically evaluate continual text-to-3D learning, we introduce Toys4K-CL, a benchmark derived from the Toys4K dataset that provides balanced and semantically diverse class-incremental splits. Extensive experiments on the Toys4K-CL benchmark show that ReConText3D consistently outperforms all baselines across different generative backbones, maintaining high-quality generation for both old and new classes. To the best of our knowledge, this work establishes the first continual learning framework and benchmark for text-to-3D generation, opening a new direction for incremental 3D generative modeling. Project page is available at: https://mauk95.github.io/ReConText3D/.
Muhammad Ahmed Ullah Khan, Muhammad Haris Bin Amir, Didier Stricker +1
Recent progress in text-to-3D object generation enables the synthesis of detailed geometry from text input by leveraging 2D diffusion models and differentiable 3D representations. However, the approaches often suffer from limited controllability and texture ambiguity due to the limitation of the text modality. To address this, we present SIC3D, a controllable image-conditioned text-to-3D generation pipeline with 3D Gaussian Splatting (3DGS). There are two stages in SIC3D. The first stage generates the 3D object content from text with a text-to-3DGS generation model. The second stage transfers style from a reference image to the 3DGS. Within this stylization stage, we introduce a novel Variational Stylized Score Distillation (VSSD) loss to effectively capture both global and local texture patterns while mitigating conflicts between geometry and appearance. A scaling regularization is further applied to prevent the emergence of artifacts and preserve the pattern from the style image. Extensive experiments demonstrate that SIC3D enhances geometric fidelity and style adherence, outperforming prior approaches in both qualitative and quantitative evaluations.
Ming He, Zhixiang Chen, Steve Maddock
School of Computer Science, University of Sheffield
Current 3D generation methods lack explicit local control: geometric adherence is often defined by a global control strength, and appearance cannot be specified locally. We present SpaceFlow, a training-free pipeline for locally controllable 3D generation from text descriptions and a collection of geometric primitives. Each primitive serves as a proxy for an object part and is assigned a local control level, enabling users to specify whether regions should strictly follow the input shape or allow generative completion. During structure generation, we enforce these spatial constraints within the generative flow process. For appearance synthesis, the generated structure is segmented and matched to the primitives. Each generated part is conditioned only on its assigned text or image cue, thereby limiting cross-part leakage. Regional geometry metrics demonstrate that SpaceFlow preserves the specified geometry in high-control regions and enables plausible shape variation in low-control areas. A user study further indicates that the resulting balance between geometric fidelity and generative freedom remains competitive in overall quality. When evaluating appearance on fixed geometry, text-conditioned routing achieves state-of-the-art prompt faithfulness and color/material accuracy. Qualitative results additionally show localized routing of image cues. The project page is available at SpaceFlow3D.github.io.
Neil De La Fuente, Joan Lafuente, Mukhammadali Sayfiddinov +5