SILSA: Sliding-Window Slice Latents for Topology-Preserving High-Resolution 3D Generation

Tianjiao Yu; Xinzhuo Li; Yifan Shen; Ying Shen; Kiet A. Nguyen; Adheesh Sunil Juvekar; Ismini Lourentzou

cs.CVOct 1, 2026

SILSA: Sliding-Window Slice Latents for Topology-Preserving High-Resolution 3D Generation

Authors: Tianjiao Yu, Xinzhuo Li, Yifan Shen, Ying Shen, Kiet A. Nguyen, Adheesh Sunil Juvekar, Ismini Lourentzou

Organizations: University of Illinois Urbana-Champaign

Abstract

High-resolution 3D generation increasingly relies on voxel latents and multi-stage pipelines that first predict active structure and then synthesize local geometry. While effective, this design fragments continuous surfaces into many local tokens, inflates generation cost, and often weakens topological consistency for thin or highly connected shapes. We introduce SILSA, a topology-aware 3D generation framework that represents shapes with compact sliding-window slice latents. Instead of generating expensive voxel tokens, SILSA uses a fixed set of overlapping slices along the three canonical axes, where each token summarizes a local depth window to preserve cross-sectional continuity and support single-stage rectified-flow generation. A Slice VAE encodes oriented surface samples into multi-axis slice latents and reconstructs them with a sparse volumetric decoder, while a Volumetric Anchor Lattice coordinates directional slice streams through a shared 3D workspace. To preserve structural correctness, we introduce slice-level topology supervision that matches persistence diagrams and aligns Betti transitions across neighboring slices. Experiments show that SILSA improves structural fidelity while substantially reducing generation cost. SILSA improves PSNR by $8.7\%$ , coverage by $5.96$ absolute points, and Betti error by $9.2\%$ over the strongest baseline, while using $70.0\%$ fewer tokens than the next-most compact baseline and over $98\%$ fewer tokens than sparse or hierarchical tokenizers, effectively reducing training memory by $40.4\%$ and inference time by $58.5\%$ . Qualitative results further show improved preservation of thin structures, repeated components, and long-range connectivity.

Figures & tables

Figure 1: High-resolution 3D generation through topology-preserving slice latents. Given a single input image, 0.64314 0.68627 0.84706S0.34902 0.7451 0.88235I \__color_backend_reset: \__color_backend_reset: 0.41569 0.76863 0.6902L0.47843 0.78824 0.50196S0.5451 0.81176 0.3098A \__color_backend_reset: \__color_backend_reset: \__color_backend_reset: generates high-resolution 3D shapes with coherent structure and detailed local geometry, including thin structures, holes, repeated components, and long-range connectivity. By modeling shapes through compact overlapping slice latents and cross-axis volumetric coordination, SILSA maintains cross-sectional continuity and produces coherent global structure. — Figure 1: High-resolution 3D generation through topology-preserving slice latents. Given a single input image, 0.64314 0.68627 0.84706S0.34902 0.7451 0.88235I __color_backend_reset: __color_backend_reset: 0.41569 0.76863 0.6902L0.47843 0.78824 0.50196S0.5451 0.81176 0.3098A __color_backend_reset: __color_backend_reset: __color_backend_reset: generates high-resolution 3D shapes with coherent structure and detailed local geometry, including thin structures, holes, repeated components, and long-range connectivity. By modeling shapes through compact overlapping slice latents and cross-axis volumetric coordination, SILSA maintains cross-sectional continuity and produces coherent global structure.

Figure 2: 0.64314 0.68627 0.84706S0.34902 0.7451 0.88235I \__color_backend_reset: \__color_backend_reset: 0.41569 0.76863 0.6902L0.47843 0.78824 0.50196S0.5451 0.81176 0.3098A \__color_backend_reset: \__color_backend_reset: \__color_backend_reset: Overview. SILSA represents each shape with overlapping slice latents along the $x$ , $y$ , and $z$ axes. A topology-aware SliceVAE encodes surfaces into fixed multi-axis slice latents and decodes them through sparse volumetric upsampling into a high-resolution mesh. An image-conditioned rectified-flow transformer generates these latents, using a Volumetric Anchor Lattice as shared 3D memory for cross-axis coordination. Slice-wise persistent-homology and Betti-transition losses supervise connected components, holes, and topological consistency during VAE training. — Figure 2: 0.64314 0.68627 0.84706S0.34902 0.7451 0.88235I __color_backend_reset: __color_backend_reset: 0.41569 0.76863 0.6902L0.47843 0.78824 0.50196S0.5451 0.81176 0.3098A __color_backend_reset: __color_backend_reset: __color_backend_reset: Overview. SILSA represents each shape with overlapping slice latents along the $x$ , $y$ , and $z$ axes. A topology-aware SliceVAE encodes surfaces into fixed multi-axis slice latents and decodes them through sparse volumetric upsampling into a high-resolution mesh. An image-conditioned rectified-flow transformer generates these latents, using a Volumetric Anchor Lattice as shared 3D memory for cross-axis coordination. Slice-wise persistent-homology and Betti-transition losses supervise connected components, holes, and topological consistency during VAE training.

Model	CLIP $\uparrow$	FD $\downarrow$	KD $\downarrow$	PSNR $\uparrow$	LPIPS $\downarrow$	COV(%) $\uparrow$	MMD(‰) $\downarrow$
Shap-E	80.16	34.64	0.87	16.84	0.21	61.41	19.19
LN3Diff	82.79	26.98	0.76	18.73	0.19	55.21	19.84
Direct3D	74.12	24.97	0.33	22.36	0.17	58.72	18.46
3DTopia-XL	76.46	24.21	0.29	22.06	0.18	58.93	17.62
InstantMesh	84.41	20.13	0.29	25.72	0.11	66.84	16.72
Gau.Any.	80.91	22.46	0.44	23.84	0.15	60.01	15.47

Table 1: Image-to-3D generation. We compare SILSA with representative image-conditioned and native 3D generation methods. Best and second best highlighted.

Model	CD $\downarrow$	F-Score@0.01 $\uparrow$	F-Score@0.005 $\uparrow$	IoU $\uparrow$	Betti-Err $\downarrow$
Dora	9.76	67.92	38.71	64.85	4.916
XCube	3.21	81.74	52.06	78.13	4.382
Trellis (SLAT)	1.12	92.83	71.45	86.29	2.871
SparseFlex	0.61	96.18	83.62	92.54	1.743
0.64314 0.68627 0.84706S0.34902 0.7451 0.88235I __color_backend_reset: __color_backend_reset: 0.41569 0.76863 0.6902L0.47843 0.78824 0.50196S0.5451 0.81176 0.3098A __color_backend_reset: __color_backend_reset: __color_backend_reset:	0.59	96.79	84.03	93.01	1.582

Table 2: VAE reconstruction quality. We compare SILSA against representative 3D reconstruction tokenizers using geometric, volumetric, and topology-aware metrics. Betti-Err measures the average mismatch in connected components and holes. Best and second best results highlighted.

Model	#Tokens $\downarrow$	Stages $\downarrow$	Params (M) $\downarrow$	Train Mem. $\downarrow$	Train Time $\downarrow$	Inference $\downarrow$	CD $\downarrow$
				(GB)	(s/iter)	(s/shape)
Dora	1,280	1	124	14.6	0.38	0.82	9.76
XCube	64,821	2	87	68.4	1.18	2.74	3.21
Trellis	19,847 $\pm$ 4,312	2	347	42.7	0.71	1.93	1.12
SparseFlex	87,453 $\pm$ 18,264	2	213	55.4	1.47	1.15	0.61
0.64314 0.68627 0.84706S0.34902 0.7451 0.88235I __color_backend_reset: __color_backend_reset: 0.41569 0.76863 0.6902L0.47843 0.78824 0.50196S0.5451 0.81176 0.3098A __color_backend_reset: __color_backend_reset: __color_backend_reset:	384 $(N{=}128)$	1	96	8.7	0.21	0.34	0.59

Table 3: Efficiency comparison. Token counts for variable-length methods are reported as mean

\pm

std over the test set. Memory and training time are measured with batch size 4 on a single A100. Inference time is reported end-to-end per shape. Best and second best highlighted.

Figure 3: Image-to-3D generation in the wild.

Figure 4: VAE reconstruction quality. We compare reconstructed meshes from SILSA and representative 3D models. Normal maps are shown in the top-right inset, and surface-error maps are shown in the bottom-right inset. Surface error is visualized as from low (blue) to high (red).

Variant	CD $\downarrow$	IoU $\uparrow$	Betti $\downarrow$
Slice-wise topology loss
No $\mathcal{L}_{\mathrm{topo}}$	0.78	81.76	4.43
Only $\mathcal{L}_{\mathrm{PH}}$	0.72	84.41	2.91
Only $\mathcal{L}_{\mathrm{trans}}$	0.69	87.16	2.87
Full $\mathcal{L}_{\mathrm{topo}}$	0.59	93.01	1.58
Cross-axis communication

Table 4: Key ablations. We ablate topology supervision and VAL. Best and second best highlighted.

Appendix figures & tables7 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 5: Topology signals for slice-wise supervision. For each canonical slicing direction, cross-sections form a sequence over depth. Blue intervals denote connected components $(\beta_{0})$ and red intervals denote holes $(\beta_{1})$ that persist across ranges of slices. Our loss uses these signals in two ways: per-slice persistence matching supervises the topology within each cross-section, while Betti-transition matching supervises where components and holes appear, disappear, merge, or split across neighboring slices.

Model	CD $\downarrow$	F-Score@0.01 $\uparrow$	F-Score@0.005 $\uparrow$	IoU $\uparrow$	Betti-Err $\downarrow$
Trellis (SLAT)	0.07	99.71	91.18	96.84	0.01
SparseFlex	0.05	100.00	93.07	98.42	0.01
SILSA	0.04	100.00	93.21	98.71	0.00

Appendix

Table 5: VAE reconstruction on open-surface shapes. Best and second best highlighted.

Figure 6: Additional image-to-3D results.

Variant	#Tokens	CD $\downarrow$	F@0.01 $\uparrow$	IoU $\uparrow$	Betti-Err $\downarrow$
Number of slices $N$
$N{=}64$	192	0.82	94.36	88.24	2.86
$N{=}96$	288	0.67	95.81	91.27	2.04
$N{=}128$ (default)	384	0.59	96.79	93.01	1.58
$N{=}256$	768	0.57	97.02	93.28	1.53
$N{=}512$	1536	0.55	97.24	93.51	1.63

Appendix

Table 6: Ablation on slice representation. We ablate the number of slices per axis

N

and the sliding-window width

w

. Default settings are

N{=}128

w{=}8

, and three canonical axes, yielding

3N{=}384

slice tokens. Reported on the VAE reconstruction task.

Variant	CD $\downarrow$	F@0.01 $\uparrow$	F@0.005 $\uparrow$	IoU $\uparrow$	Betti-Err $\downarrow$
Loss components
No $\mathcal{L}_{\mathrm{topo}}$	0.78	93.84	78.16	81.76	4.43
Only $\mathcal{L}_{\mathrm{PH}}$	0.72	94.51	79.72	84.41	2.91
Only $\mathcal{L}_{\mathrm{trans}}$	0.69	95.03	80.46	87.16	2.87
Full $\mathcal{L}_{\mathrm{topo}}$ (default)	0.59	96.79	84.03	93.01	1.58
Topology loss weight $\lambda_{\mathrm{topo}}$

Appendix

Table 7: Ablation of the slice-wise topology-preserving loss. All variants are evaluated on VAE reconstruction. Default settings are shaded.

Variant	Axes	$N$	#Tokens	CD $\downarrow$	IoU $\uparrow$	Betti-Err $\downarrow$
Single axis
1 axis ( $z$ only)	$z$	384	384	1.34	76.43	5.92
1 axis ( $z$ only)	$z$	128	128	1.87	71.29	7.83
1 axis ( $x$ only)	$x$	384	384	1.41	75.18	6.18
1 axis ( $x$ only)	$x$	128	128	1.94	70.42	8.07
Two axes

Appendix

Table 8: Ablation on the number of canonical axes. Default setting is shaded.

Variant	CD $\downarrow$	F@0.01 $\uparrow$	IoU $\uparrow$	Betti-Err $\downarrow$
Overwrite (no gating)	0.66	95.61	90.86	1.91
Additive write (no gating)	0.63	96.12	92.04	1.74
Gated write (default)	0.59	96.79	93.01	1.58

Appendix

Table 9: Ablation on the VAL update mechanism. Default setting is shaded.

Explore similar work

FILIGREE3D: Scaling Sparse Latent Flow Matching for Ultra-High-Resolution Image-to-3D Generation
Sep 28, 2026Hongjie Li, Xinran Yang, Xiuchao Wu +23D Generative Models 3D Generation
AlibabaGroup
SuperVoxelGPT: Adaptive and Ordered 3D Tokenization for Autoregressive Shape Generation
May 28, 2026Yuan Li, Congyi Zhang, Xifeng Gao +1Voxel Autoregressive Generation
University of Texas at Dallas, USA · LightSpeed, USA
ZipTok3D: High-Fidelity 3D Tokenization with Compact Token Prefixes
Sep 1, 2026Mingda Lin, Weijie Wang, Zeyu Zhang +73D Generative Models 3D Generation
1Zhejiang University · 2Monash University · University of Adelaide

Sep 28, 2026, cs.CVJ/K move · Enter open · S save

FILIGREE3D: Scaling Sparse Latent Flow Matching for Ultra-High-Resolution Image-to-3D Generation

Hongjie Li, Xinran Yang, Xiuchao Wu, Jiangjing Lyu+1

AlibabaGroup

3D Generative Models 3D Generation High-Resolution Latent Flow Diffusion Transformers

Open paper AskPDF arXiv

Scaling image-to-3D generation to ultra-high resolutions requires controlling rapidly growing computational costs without sacrificing fine geometric detail. We present \textbf{Filigree3D}, a sparse latent flow-matching framework that generates 3D geometry from a single image at voxel resolutions up to

2048^3

, with straightforward extensibility to

4096^3

. To make training tractable, we introduce Structure-Aware Sparse Scaling, which combines spatial bounding with alternating local-global attention to constrain token growth while preserving both fine-scale details and long-range structural context. To enhance detail reconstruction, we curate training samples based on their high-resolution geometric gains and inject multi-scale image features into a sparse 3D DiT, effectively coupling structural semantics with fine-grained visual cues. Furthermore, a visibility-aware voxel regularization strategy improves robustness against sparse perturbations and facilitates the completion of unobserved geometry. Under our default configuration, Filigree3D maintains peak GPU memory consumption within practical limits for contemporary hardware, enabling the generation of highly intricate 3D geometry in approximately one minute. Extensive experiments demonstrate that our method yields substantial improvements in overall geometric fidelity and fine-detail preservation compared to existing baselines, validating practical, detail-preserving 3D generation at unprecedented resolutions.

Abstract

Explore similar work

FILIGREE3D: Scaling Sparse Latent Flow Matching for Ultra-High-Resolution Image-to-3D Generation

SuperVoxelGPT: Adaptive and Ordered 3D Tokenization for Autoregressive Shape Generation

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