Frequency-Adaptive Gaussian Grids for Efficient Image Representation

Anatolii Evdokimov; Stavros Diolatzis; Bingxuan Li; Akshay Jindal; Patsorn Sangkloy; Qi Sun

cs.CVOct 27, 2026

Frequency-Adaptive Gaussian Grids for Efficient Image Representation

Authors: Anatolii Evdokimov, Stavros Diolatzis, Bingxuan Li, Akshay Jindal, Patsorn Sangkloy, Qi Sun

Organizations: New York University · Intel Corporation

Abstract

Neural representation has emerged as a powerful tool to mitigate the demand for increasingly higher-resolution images. However, a fundamental trade-off remains between structured (e.g., feature grids, which offer spatial consistency) and unstructured (e.g., 2D Gaussians, which adaptively capture local variances) representations. This choice significantly affects how well an algorithm represents an image. Here, we present Gaussian-Grid, a hybrid neural representation that adaptively integrates structured and unstructured primitives. By dynamically partitioning computation between 2D Gaussian distributions and their attached multi-resolution feature grids based on the local frequency content, Gaussian-Grid automatically optimizes its architecture to match the underlying complexity of the target signal. We evaluate our approach using high-resolution natural images and demonstrate its applications as GPU-native dataset storage, showing $1.24$ - $2.14\times$ decoding speed improvement and PCIe data traffic reduction of $23\times$ - $176\times$ over JPEG + ZIP data archiving. The results show that Gaussian-Grid maintains small memory footprints while achieving high reconstruction fidelity across a variety of datasets.

Figures & tables

Panel of Figure 1 — Figure 1: Gaussian-Grid leverages Gaussian-defined feature grid layout that is guided by image frequency map to generate high quality reconstruction.

Figure 2: Overview of our method. Gaussian-Grid optimizes 2D Gaussian primitives to a spatial frequency map derived from the target image. Gaussians are then used to place feature grids of variable resolutions, aligning with the underlying frequency of the content. In the final stage, the feature grids are optimized to reconstruct the final image.

Figure 3: Pixel Rendering. For a given pixel location, Gaussian-Grid bilinearly samples top-K overlapping Gaussian grids as well as the canvas grid. The interpolated features are decoded by a small MLP dec to render the final pixel value.

Panel of Figure 4 — Figure 4: Qualitative comparison with other recent image representation baselines. Examples are from the DIV2K dataset. Columns (a)-(e) show model reconstructions with error map insets, (f) and (g) show reference crops and the full reference images respectively. The budgets of I-NGP, GI, SAD, IMG-GS, and Gaussian-Grid are 0.1 bits per pixel respectively.

		DIV2K				CLIC2020
Method	Metric	0.1 bpp	0.4 bpp	0.7 bpp	1.0 bpp	0.1 bpp	0.4 bpp	0.7 bpp	1.0 bpp
SAD [ 13 ]	PSNR $\uparrow$	24.72	29.36	31.24	32.43	26.95	31.29	33.04	34.10
	SSIM $\uparrow$	0.6573	0.7816	0.8309	0.8597	0.7210	0.8096	0.8465	0.8698
	MS-SSIM $\uparrow$	0.7954	0.9196	0.9495	0.9632	0.8353	0.9271	0.9518	0.9643
	LPIPS $\downarrow$	0.4512	0.2980	0.2399	0.2040	0.4238	0.2996	0.2524	0.2197
GI [ 39 ]	PSNR $\uparrow$	24.21	27.27	28.71	29.69	26.36	29.15	30.40	31.33

Table 1: Comparison of Gaussian-Grid with recent primitive-based and neural baselines on DIV2K (100 images) and CLIC2020 validation (41 images) natural image datasets across four budgets (0.1 bits per pixel to 1.0 bits per pixel). The best result is in dark blue bold and the second best is in light blue.

bpp	JPEG	Ours	JPEG	Ours	$\Delta$	Unzip	Dec	PCIe	Total	Load	PCIe	Dq	Krnl	Total	Speedup	$\text{PCIe}\times$
	Dataset size (MB)		PSNR (dB)			JPEG decode (ms / image)				Our decode (ms / image)
0.1	4.85	4.62	$23.25$	$\mathbf{25.58}$	$\mathbf{+2.33}$	0.87	10.05	0.73	11.65	0.17	0.14	0.64	4.48	$\mathbf{5.44}$	$\mathbf{2.14\times}$	$176\times$
0.4	14.71	14.85	$29.05$	$\mathbf{29.36}$	$\mathbf{+0.31}$	1.20	9.46	0.67	11.33	0.16	0.15	0.63	6.78	$\mathbf{7.73}$	$\mathbf{1.47\times}$	$55\times$
0.7	24.97	25.28	$\mathbf{31.75}$	$31.22$	$-0.53$	1.26	9.88	0.64	11.78	0.16	0.15	0.57	8.25	$\mathbf{9.13}$	$\mathbf{1.29\times}$	$32\times$
1.0	35.48	35.71	$\mathbf{33.55}$	$32.55$	$-1.00$	1.30	10.61	0.64	12.55	0.16	0.16	0.59	9.19	$\mathbf{10.11}$	$\mathbf{1.24\times}$	$23\times$
All	80.0	80.5	—		$\mathbf{+0.27}$	1.16	10.00	0.67	11.83	0.16	0.15	0.61	7.18	$\mathbf{8.10}$	$\mathbf{1.46\times}$	$40\times$

Table 3: End-to-end decode comparison against JPEG + ZIP on 100 DIV2K images at four budgets. At 0.1 bpp our representation yields

+2.33

dB over a size-matched JPEG while decoding

2.14\times

faster; The decode speedup (

1.24

–

2.14\times

, Speedup column) persists across all budgets. The rightmost column (

\text{PCIe}\times

) shows the most distinctive property: transferring quantized model weights (47–366 KB) instead of decoded pixel maps (

8{,}306

KB per image) reduces CPU-to-GPU traffic by

23

–

176\times

. Pipeline columns. Ours: weights load ( Load ), GPU transfer ( PCIe ), GPU dequantize ( Dq ), GPU kernel forward pass ( Krnl ). JPEG: one-time ZIP extraction on CPU amortized over 100 images ( Unzip ), JPEG CPU decode ( Dec ), and pixel transfer to GPU ( PCIe ).

Implicit Neural Representations

Implicit neural representations encode signals and fields as continuous functions that can be queried at arbitrary coordinates and resolutions.

See all 188 papers

SGI: Structured 2D Gaussians for Efficient and Compact Large Image Representation
Mar 8, 2026Zixuan Pan, Kaiyuan Tang, Jun Xia +5Gaussian Splatting Compression Gaussian Splatting
University of Notre Dame · Tohoku University · Leibniz-Institut für Analytische Wissenschaften – ISAS – e.V.
Locality-Aware Density Control for Efficient Gaussian-based Image Representation
Jul 20, 2026Jiacong Chen, Qingyu Mao, Xiandong Meng +53D Gaussian Splatting Adaptive Gaussian Densification
Shenzhen University Shenzhen, China · Pengcheng Laboratory Shenzhen, China · Harbin Institute of Technology (Shenzhen) Shenzhen, China
Efficient Compression of Structured and Unstructured Volumes via Learned 3D Gaussian Representation
Jul 1, 2026Landon Dyken, Sharmistha Chakrabarti, Nathan Debardeleben +43DGS Compression 3D Gaussian Splatting

Mar 8, 2026J/K move · Enter open · S save

SGI: Structured 2D Gaussians for Efficient and Compact Large Image Representation

Zixuan Pan, Kaiyuan Tang, Jun Xia, Yifan Qin+4

University of Notre Dame · Tohoku University · Leibniz-Institut für Analytische Wissenschaften – ISAS – e.V.

Gaussian Splatting Compression Gaussian Splatting Implicit Neural Representations

Open paper AskPDF arXiv

2D Gaussian Splatting has emerged as a novel image representation technique that can support efficient rendering on low-end devices. However, scaling to high-resolution images requires optimizing and storing millions of unstructured Gaussian primitives independently, leading to slow convergence and redundant parameters. To address this, we propose Structured Gaussian Image (SGI), a compact and efficient framework for representing high-resolution images. SGI decomposes a complex image into multi-scale local spaces defined by a set of seeds. Each seed corresponds to a spatially coherent region and, together with lightweight multi-layer perceptrons (MLPs), generates structured implicit 2D neural Gaussians. This seed-based formulation imposes structural regularity on otherwise unstructured Gaussian primitives, which facilitates entropy-based compression at the seed level to reduce the total storage. However, optimizing seed parameters directly on high-resolution images is a challenging and non-trivial task. Therefore, we designed a multi-scale fitting strategy that refines the seed representation in a coarse-to-fine manner, substantially accelerating convergence. Quantitative and qualitative evaluations demonstrate that SGI achieves up to 7.5x compression over prior non-quantized 2D Gaussian methods and 1.6x over quantized ones, while also delivering 1.6x and 6.5x faster optimization, respectively, without degrading, and often improving, image fidelity. Code is available at https://github.com/zx-pan/SGI.

Abstract

Implicit Neural Representations

Explore similar work

SGI: Structured 2D Gaussians for Efficient and Compact Large Image Representation

Locality-Aware Density Control for Efficient Gaussian-based Image Representation

Efficient Compression of Structured and Unstructured Volumes via Learned 3D Gaussian Representation