Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Instant GaussianImage: A Generalizable and Self-Adaptive Image Representation via 2D Gaussian Splatting

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A generalizable network initializes 2D Gaussian splats in one forward pass, then minimal fine-tuning matches or beats GaussianImage at a fraction of the training time.

desk verdict Network-initialized 2D Gaussians with dithering-based adaptive counts give a real speedup; the color-sampling ambiguity in Eq. 14 needs clarification before the headline claim is fully trusted. read the letter →

arxiv 2506.23479 v1 pith:G4OPAKJ4 submitted 2025-06-30 cs.CV

Zhaojie Zeng , Yuesong Wang , Chao Yang , Tao Guan , Lili Ju This is my paper
classification cs.CV
keywords 2DGaussianSplattingImageRepresentationImplicitNeuralGeneralizableInitializationFloyd–SteinbergDitheringPositionProbabilityMapSelf-AdaptiveCountFastRendering
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Instant-GI proposes a generalizable image representation based on 2D Gaussian Splatting. A ConvNeXt-based UNet extracts features and predicts a Position Probability Map (PPM), which is discretized by Floyd–Steinberg dithering to place Gaussians adaptively. A set of MLP fields then predicts each Gaussian's position, scale, rotation, and opacity, with colors sampled directly from the input image. After a single feed-forward pass and minimal fine-tuning, the rendered image matches or exceeds the quality of GaussianImage, which trains from random initialization for much longer. The same network also controls the number of Gaussians via a kernel-size parameter, allowing a compression-quality trade-off without retraining.

What carries the argument

The central object is the Position Probability Map (PPM), a per-pixel probability of placing a Gaussian, predicted by a small MLP from a ConvNeXt UNet feature map. Floyd–Steinberg dithering discretizes the PPM, distributing a self-adaptive number of Gaussians that matches local image entropy. Delaunay triangulation on the sampled points yields triangles whose fitted ellipses provide a geometric reference scale and orientation for the Gaussian fields; MLP fields predict barycentric positions, scale offsets, rotation bias, and opacity, while color is obtained by sampling the input image at the Gaussian center and multiplying by a sigmoid-activated opacity. This design turns Gaussian decomposition into a feed-forward operation followed by a short fine-tune.

What would settle it

Inspect the inference code path after fine-tuning: if rendering a Gaussian representation still requires the original image to supply the color term I(pi) of Eq. (14), then the representation is not self-contained and the training-time comparison is unfair. Alternatively, retrain the fine-tuning stage with color as a free parameter while hiding the source image, and check whether the reported ~10x speed-up and PSNR gains persist.

Watch

Extended reading notes

Core claim

The paper's central claim is that a network-initialized Gaussian decomposition can replace random initialization in GaussianImage, achieving comparable or better rendering fidelity in about one tenth of the training time. The key finding is that with the same number of Gaussians, starting from a network's structured prediction—rather than uniform random positions—lets optimization converge to a higher PSNR in far fewer iterations. On the DIV2K ×2 test set the method reports 42.80 dB versus 40.26 dB for GaussianImage with equal Gaussian count; on Kodak it reports 42.92 dB, slightly below GaussianImage's 44.08 dB but with fewer points and more consistent performance across datasets. The paper also demonstrates that rendering quality after 2 seconds of network initialization and fine-tuning can exceed the 20-second result of random initialization.

Load-bearing premise

The method assumes that sampling a Gaussian's color directly from the input image at its center is a legitimate part of the representation; if the original image is still required to render after fine-tuning, the representation is not standalone and the claimed speed advantage over GaussianImage largely reflects querying ground truth rather than true learning.

Editorial extensions

If this is right

  • Training time for a usable Gaussian image representation drops from tens of seconds to a few seconds: on image 0844, the method reaches 46.68 dB PSNR after 2 s including initialization, versus 45.90 dB after 20 s with random initialization.
  • The number of Gaussians can be adjusted at test time by changing the dithering kernel size k (e.g., from 3 to 4), trading quality for compression without retraining the network.
  • Because the Gaussian model matches GaussianImage, existing compression techniques for GaussianImage apply directly to the method's output.
  • On the DIV2K ×2 dataset the network-initialized representation reaches 42.80 dB PSNR, outperforming GaussianImage's 40.26 dB under the same Gaussian count.
  • Network-based initialization also yields smoother spatial distributions than gradient-based point placement, avoiding over- and under-sampling in high- and low-entropy regions.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A testable implication is that the same network could amortize initialization across many images in a video sequence, potentially turning per-frame Gaussian fitting into a short residual update.
  • Because colors are sampled from the source image during initialization, a fair standalone representation requires that fine-tuning converts those colors into independent parameters; if the image remains needed at inference, the reported speed gains over GaussianImage are partly an artifact of information leakage.
  • The kernel-size control could serve as an explicit rate-distortion knob for progressive compression, but the available experiments only demonstrate it on kernel sizes 3 and 4.
  • The PPM supervision depends on a pre-trained GaussianImage decomposition; a different teacher model would likely shift both the point distribution and the final rendering quality.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes Instant-GI, a generalizable initialization method for 2D Gaussian splatting image representation. A ConvNeXt-based UNet extracts features, a Position Field predicts a Gaussian Position Probability Map supervised by a pseudo-PPM computed from a converged GaussianImage model, and Floyd-Steinberg dithering discretizes this map to adaptively select Gaussian locations. A second set of MLP fields predicts position, scaling, rotation, and opacity, and the final Gaussian color is defined in Eq. (14) as the image color sampled at the Gaussian center multiplied by the predicted opacity. After a feed-forward initialization, the representation is fine-tuned with a small number of iterations. Experiments on Kodak and DIV2K report PSNR comparable to or better than GaussianImage with fewer iterations and, for three selected images, show large PSNR gains at 2s compared with a randomly initialized baseline at 20s.

Significance. If the correctness issues are resolved, the paper addresses a real limitation of GaussianImage: slow training from random initialization and a fixed number of Gaussians. The idea of a learned prior for Gaussian placement and a dithering-based adaptive count is practical and the code is promised publicly. The best-case result, matching or exceeding GaussianImage in a fraction of the training time, would be useful for image representation and compression. However, the current manuscript leaves open a load-bearing question: whether the final representation is self-contained or still requires the original image to render. Because Eq. (14) defines colors as direct samples from the target image, the headline speed advantage may be partly an artifact of ground-truth access. The pseudo-PPM supervision from a converged GaussianImage model also means the 'self-adaptive' behavior is a distillation of the baseline rather than an independent entropy-based criterion. These issues need to be clarified and re-tested before the central claims can be accepted.

major comments (3)
  1. [Sec. 3.4, Eq. (14)] The definition ci = I(pi)·σ(o) means that, as written, the Gaussian color is read from the target image at the Gaussian center. The paper never states whether this expression is evaluated only during the network's feed-forward pass and then detached, or whether the renderer continues to access I during fine-tuning and at inference. The headline result in Sec. 4.3 (2s vs 20s) is only a fair test of an image representation if the final stored Gaussians do not require I for rendering; otherwise the speed advantage is in part an artifact of querying ground-truth colors. Please clarify this explicitly, and if colors are detached, state so and include an ablation that replaces I(pi) with a learned color to quantify the contribution of this initialization.
  2. [Sec. 3.3 and Supplementary Sec. 2.3] The pseudo PPM used to supervise the Position Field is computed from a converged GaussianImage model (Supplementary Sec. 2.3). The claim that the method 'adaptively determines the number of Gaussians based on information entropy' is therefore a claim about matching the baseline's own placement, not an independently derived entropy-based criterion. This is a form of self-distillation; it should be acknowledged and tested against the alternative of sampling directly from the pseudo PPM at test time to show that the learned PPM adds robustness or quality beyond the teacher.
  3. [Sec. 4.3, Table 2] The order-of-magnitude speedup is supported by PSNR measurements on only three images, with no standard deviations or repeated-run statistics. In addition, the '2s' interval must be specified precisely (network forward pass, dithering, Delaunay triangulation, number of fine-tuning iterations) and, given Eq. (14), the reported 2s PSNRs may include the effect of GT color lookup. Please report metrics over a larger set with error bars and an experiment that isolates the contribution of GT color sampling from the contribution of position, scaling, and opacity initialization.
minor comments (4)
  1. [Table 1] The second GI row appears as 'GI ' rather than 'GI†' in the printed table, and the dagger is not visibly attached to the row. Please ensure the marker matches the caption and report the number of Gaussians for each row so the 'same number of Gaussians' comparison is easy to verify.
  2. [Sec. 4.3, Table 2] The GPU memory column for Net. Init. lists '3038 + 408' but the text says 'approximately 3GB'; clarify what the two numbers represent.
  3. [Throughout] There are several typographical errors: 'hass' in Sec. 2.1, 'availiable' in the abstract, 'upper bounder' in Sec. 4.2, and 'x g 0' in Eq. (11) should be 'x ≥ 0'.
  4. [Sec. 4.2] The sentence 'Kodak's lower resolution, leading to a 61% reduction in Gaussian points' should be rephrased: it was not Kodak's resolution per se but the adaptive algorithm that selected fewer Gaussians; please state the comparison baseline for the 61% reduction.

Circularity Check

1 steps flagged · score 6.0 of 10

Eq. 14 reads Gaussian colors directly from the target image, so the 2s-vs-20s speedup is largely an artifact of querying ground truth rather than predicting the representation.

  1. fitted input called prediction [Sec. 3.4 (Eq. 14); Sec. 4.3 (Tab. 2)]
    "The final color attribute ci is then obtained by multiplying the color sampled from the image I at the Gaussian position pi with the opacity: ci = I(pi) · σ(o). ... Even for 0858×2 ... our method still achieves a 5× speed-up by predicting more accurate scaling, rotation, and color parameters."

    Eq. 14 defines the Gaussian color as a direct lookup from the target image I at the Gaussian center, scaled by opacity. The renderer (Eq. 3) then splats these exact target colors at every point, so the initial rendering already contains ground-truth color values at all Gaussian locations. Table 2's headline comparison, Net. Init. reaching PSNR 46.68 at 2s versus 45.90 for random init at 20s, therefore pits a method that is handed the correct color for the color channel against a baseline that must regress RGB from scratch.

full rationale

The paper's central efficiency claim is substantially undermined by one circular shortcut. In Sec. 3.4, the Gaussian color attribute is defined as the target-image color at the Gaussian center (Eq. 14), meaning the network does not predict color at all; the renderer composes ground-truth colors from the very first iteration. The 2s-vs-20s PSNR comparison in Sec. 4.3 is therefore not a clean test of learned initialization: the color channel, which dominates a splatting render's signal, is initialized by direct target lookup. The paper also never states whether the sampled colors become independent stored parameters after the brief fine-tune; its claim that 'all compression techniques for GI can be applied directly' assumes this, while Eq. 14 suggests rendering still requires I. This makes the order-of-magnitude speed advantage partially circular by construction. The position/scale/rotation prediction, the dithering-based adaptive point placement, and the final converged 20s results retain independent content. The pseudo-PPM supervision derived from a converged GaussianImage model is a legitimate teacher-student training signal rather than a logical circularity, though it qualifies the originality of the 'self-adaptive' claim. Overall, one central 'prediction' reduces to reading the input, so a score of 6 is appropriate.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim rests on several hand-chosen hyperparameters (K, k, loss weights) and on the ad hoc assumption that a converged GaussianImage model provides the 'correct' Gaussian distribution to imitate. The color-from-image mechanism (Eq. 14) is the most load-bearing assumption, since it may allow the renderer to query ground truth directly.

free parameters (6)
  • K in pseudo density = 10
    K-nearest neighbors in Eq. 6 for local Gaussian density; chosen by hand.
  • dithering kernel size k = 3 (train), 3/4/5 (test)
    Controls the number of Gaussians generated from the PPM; user-adjustable for compression ratio.
  • loss weights lambda, alpha, gamma = lambda=2, alpha=1.0, gamma=2.0
    Set empirically in Sec. 4.1.
  • MSE threshold and min block size for pseudo PPM quadtree = 0.02, 4x4
    Parameters in supplementary 2.3 for generating pseudo PPM; chosen by hand.
  • learning rate range and epochs = 1e-3 to 1e-5, 100 epochs
    Adam with cosine annealing; standard but chosen.
  • activation function constants in Eq. 11 = 0.5, 5, 3
    Constants in the scaling activation function fs; chosen by hand.
assumptions (5)
  • domain assumption GI accumulated blending rendering without opacity accumulation is a valid image representation model.
    The paper builds directly on GaussianImage's rasterization model (Eq. 3-5) as a given.
  • ad hoc to paper A converged GaussianImage model's Gaussian density is a good supervision signal for the Position Probability Map.
    The pseudo PPM is generated from a trained GI model (Sec. 3.3, Supplementary 2.3); this choice is not independently motivated.
  • ad hoc to paper Colors can be sampled from the target image at Gaussian positions during rendering.
    Eq. 14 defines the final color as I(p_i) * sigma(o); the paper does not clarify whether the image remains necessary at inference.
  • standard math Delaunay triangulation of sampled points ensures full coverage with no overlap, making fitted ellipses a good reference scale.
    Standard geometric property used in Sec. 3.4 to justify the ellipse fitting reference.
  • domain assumption The network trained on DIV2K generalizes to unseen images (Kodak, DIV2K test).
    Standard generalization assumption for learned initialization.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Instant GaussianImage: A Generalizable and Self-Adaptive Image Representation via 2D Gaussian Splatting." pith.science (2026). https://pith.science/paper/G4OPAKJ4

@misc{pith2026250623479,
  author       = {Pith},
  title        = {Pith review of: Instant GaussianImage: A Generalizable and Self-Adaptive Image Representation via 2D Gaussian Splatting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G4OPAKJ4}},
  note         = {Machine review of arXiv:2506.23479}
}
read the original abstract

Implicit Neural Representation (INR) has demonstrated remarkable advances in the field of image representation but demands substantial GPU resources. GaussianImage recently pioneered the use of Gaussian Splatting to mitigate this cost, however, the slow training process limits its practicality, and the fixed number of Gaussians per image limits its adaptability to varying information entropy. To address these issues, we propose in this paper a generalizable and self-adaptive image representation framework based on 2D Gaussian Splatting. Our method employs a network to quickly generate a coarse Gaussian representation, followed by minimal fine-tuning steps, achieving comparable rendering quality of GaussianImage while significantly reducing training time. Moreover, our approach dynamically adjusts the number of Gaussian points based on image complexity to further enhance flexibility and efficiency in practice. Experiments on DIV2K and Kodak datasets show that our method matches or exceeds GaussianImage's rendering performance with far fewer iterations and shorter training times. Specifically, our method reduces the training time by up to one order of magnitude while achieving superior rendering performance with the same number of Gaussians.

Figures

Figures reproduced from arXiv: 2506.23479 by the authors.

Figure 1
Figure 1. Teaser. Comparison of the network-based initialization (Net. Init.) and the random initialization (Rand. Init.) regarding training efficiency and rendering quality using the same number of Gaussians. The top graph shows that Net. Init. converges faster and achieves higher quality. The bottom row visualizes the ini￾tialization state (left), results after 20,000 iterations (middle), and zoomed-in details (right), high… view at source ↗
Figure 2
Figure 2. Overview of our Instant-GI pipeline. Given an input image, we first extract a feature map using a ConvNeXt-based UNet. The Position Probability Map (PPM) is then predicted by an MLP-based Position Field and discretized via Floyd–Steinberg Dithering. The resulting points are structured using Delaunay Triangulation, from which geometric features are extracted. Gaussian attributes, including position, scaling, rotation… view at source ↗
Figure 3
Figure 3. Feature organization for MLP input. We construct the MLP input by combining multiple feature components. Geometric features are extracted from the triangle ti and its fitted ellipse ei, while grid sampling provides local image features and sampled colors from the input image. The image features undergo feature reduction and layer normalization before concatenation. points are allocated to smooth regions, leading to … view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Reconstruction Progress Visualization. Each row sequentially shows the GT image, frequency spectrum, initialization results, and results at 2s, 10s, and 20s. The results compare Rand. Init (left) and Net. Init. (right). with random initialization and our network-based …
Figure 5
Figure 5. Figure 5: PPM Comparison. Visualization of different Position Probability Maps (PPMs). The top row shows the pseudo PPM, predicted PPM, and gradient-based PPM. The bottom row presents their corresponding probability distributions along the indicated horizontal line. Our predicte…
Figure 1
Figure 1. Figure 1: CDF Sampling Analysis. A simulated experiment analyzing the effects of CDF sampling under different entropy distributions. Top left: Pixel distribution for balanced, low-entropy dominant, and high-entropy dominant cases. Top right: Sampled point distribution after appl…
Figure 2
Figure 2. Figure 2: Feature organization for MLP input. We construct the MLP input by combining multiple feature components. Geometric features are extracted from the triangle ti and its fitted ellipse ei, while grid sampling provides local deep features and sampled col￾ors from the input…
Figure 3
Figure 3. Figure 3: PSNR curves during training process. We use image features as the default part and analyze the effects of including or excluding color features, ellipse fea￾tures, and triangle features. The results show that ellipse features are crucial for the final performance, whil…
Figure 4
Figure 4. Figure 4: More visualization Results. Please zoom in for more details. [PITH_FULL_IMAGE:figures/full_fig_p016_4.png]
Figure 5
Figure 5. Figure 5: More visualization Results. Please zoom in for more details. [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

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

    cs.CV 2026-07 conditional novelty 6.0 of 10

    A locality-aware density-control framework for 2D Gaussian image representation that densifies coherent high-error regions and merges redundant similar Gaussians, improving PSNR at fixed budgets.

Reference graph

Works this paper leans on

51 extracted references · 33 canonical work pages · cited by 1 Pith paper

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    Ntire 2017 challenge on single image super-resolution: Dataset and study

    Eirikur Agustsson and Radu Timofte. Ntire 2017 challenge on single image super-resolution: Dataset and study. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops, pages 126--135, 2017

    work page 2017

  3. [3]

    Mip-nerf: A multiscale representation for anti-aliasing neural radiance fields

    Jonathan T Barron, Ben Mildenhall, Matthew Tancik, Peter Hedman, Ricardo Martin-Brualla, and Pratul P Srinivasan. Mip-nerf: A multiscale representation for anti-aliasing neural radiance fields. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 5855--5864, 2021

    2021

  4. [4]

    Barron, Ben Mildenhall, Dor Verbin, Pratul P

    Jonathan T. Barron, Ben Mildenhall, Dor Verbin, Pratul P. Srinivasan, and Peter Hedman. Zip-nerf: Anti-aliased grid-based neural radiance fields. ICCV, 2023

    work page 2023

  5. [5]

    Discontinuity-aware 2d neural fields

    Yash Belhe, Micha \"e l Gharbi, Matthew Fisher, Iliyan Georgiev, Ravi Ramamoorthi, and Tzu-Mao Li. Discontinuity-aware 2d neural fields. ACM Transactions on Graphics (TOG), 42 0 (6): 0 1--11, 2023

    work page 2023

  6. [6]

    Delaunay

    N Boris. Delaunay. sur la sphere vide. Izvestia Akademia Nauk SSSR, VII Seria, Otdelenie Matematicheskii i Estestvennyka Nauk, 7: 0 793--800, 1934

    work page 1934

  7. [7]

    Pgsr: Planar-based gaussian splatting for efficient and high-fidelity surface reconstruction

    Danpeng Chen, Hai Li, Weicai Ye, Yifan Wang, Weijian Xie, Shangjin Zhai, Nan Wang, Haomin Liu, Hujun Bao, and Guofeng Zhang. Pgsr: Planar-based gaussian splatting for efficient and high-fidelity surface reconstruction. IEEE Transactions on Visualization and Computer Graphics, 2024 a

    work page 2024

  8. [8]

    Nerv: Neural representations for videos

    Hao Chen, Bo He, Hanyu Wang, Yixuan Ren, Ser Nam Lim, and Abhinav Shrivastava. Nerv: Neural representations for videos. Advances in Neural Information Processing Systems, 34: 0 21557--21568, 2021 a

    work page 2021

Show all 51 references
  1. [9]

    Hnerv: A hybrid neural representation for videos

    Hao Chen, Matthew Gwilliam, Ser-Nam Lim, and Abhinav Shrivastava. Hnerv: A hybrid neural representation for videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10270--10279, 2023 a

    2023

  2. [10]

    Learning continuous image representation with local implicit image function

    Yinbo Chen, Sifei Liu, and Xiaolong Wang. Learning continuous image representation with local implicit image function. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8628--8638, 2021 b

    2021

  3. [11]

    Gaussianeditor: Swift and controllable 3d editing with gaussian splatting

    Yiwen Chen, Zilong Chen, Chi Zhang, Feng Wang, Xiaofeng Yang, Yikai Wang, Zhongang Cai, Lei Yang, Huaping Liu, and Guosheng Lin. Gaussianeditor: Swift and controllable 3d editing with gaussian splatting. In Proceedings of the IEEE/CVF conference on computer vision and pattern ...

    2024

  4. [12]

    Neurbf: A neural fields representation with adaptive radial basis functions

    Zhang Chen, Zhong Li, Liangchen Song, Lele Chen, Jingyi Yu, Junsong Yuan, and Yi Xu. Neurbf: A neural fields representation with adaptive radial basis functions. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 4182--4194, 2023 b

    2023

  5. [13]

    Image super-resolution using deep convolutional networks

    Chao Dong, Chen Change Loy, Kaiming He, and Xiaoou Tang. Image super-resolution using deep convolutional networks. IEEE transactions on pattern analysis and machine intelligence, 38 0 (2): 0 295--307, 2015

    2015

  6. [14]

    A buyer's guide to conic fitting

    Andrew W Fitzgibbon, Robert B Fisher, et al. A buyer's guide to conic fitting. Citeseer, 1996

    1996

  7. [15]

    Stochastic floyd-steinberg dithering on gpu: image quality and processing time improved

    Giorgia Franchini, Roberto Cavicchioli, and Jia Cheng Hu. Stochastic floyd-steinberg dithering on gpu: image quality and processing time improved. In 2019 Fifth International Conference on Image Information Processing (ICIIP), pages 1--6, 2019

    2019

  8. [16]

    Gaussiansr: High fidelity 2d gaussian splatting for arbitrary-scale image super-resolution, 2024

    Jintong Hu, Bin Xia, Bin Chen, Wenming Yang, and Lei Zhang. Gaussiansr: High fidelity 2d gaussian splatting for arbitrary-scale image super-resolution, 2024

    2024

  9. [17]

    Meta-sr: A magnification-arbitrary network for super-resolution

    Xuecai Hu, Haoyuan Mu, Xiangyu Zhang, Zilei Wang, Tieniu Tan, and Jian Sun. Meta-sr: A magnification-arbitrary network for super-resolution. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1575--1584, 2019

    2019

  10. [18]

    2d gaussian splatting for geometrically accurate radiance fields

    Binbin Huang, Zehao Yu, Anpei Chen, Andreas Geiger, and Shenghua Gao. 2d gaussian splatting for geometrically accurate radiance fields. In ACM SIGGRAPH 2024 conference papers, pages 1--11, 2024

    2024

  11. [19]

    Ice-g: Image conditional editing of 3d gaussian splats

    Vishnu Jaganathan, Hannah Hanyun Huang, Muhammad Zubair Irshad, Varun Jampani, Amit Raj, and Zsolt Kira. Ice-g: Image conditional editing of 3d gaussian splats. arXiv preprint arXiv:2406.08488, 2024

    2024 arXiv

  12. [20]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk \"u hler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph., 42 0 (4): 0 139--1, 2023

    2023

  13. [21]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  14. [22]

    Hypernetwork functional image representation

    Sylwester Klocek, ukasz Maziarka, Maciej Wo czyk, Jacek Tabor, Jakub Nowak, and Marek \'S mieja. Hypernetwork functional image representation. In International Conference on Artificial Neural Networks, pages 496--510. Springer, 2019

  15. [23]

    Kodak lossless true color image suite, 1999

    Kodak. Kodak lossless true color image suite, 1999

    1999

  16. [24]

    Efficient frequency domain-based transformers for high-quality image deblurring

    Lingshun Kong, Jiangxin Dong, Jianjun Ge, Mingqiang Li, and Jinshan Pan. Efficient frequency domain-based transformers for high-quality image deblurring. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5886--5895, 2023

    2023

  17. [25]

    Neumip: Multi-resolution neural materials

    Alexandr Kuznetsov. Neumip: Multi-resolution neural materials. ACM Transactions on Graphics (ToG), 40 0 (4), 2021

    2021

  18. [26]

    Adaptive local implicit image function for arbitrary-scale super-resolution

    Hongwei Li, Tao Dai, Yiming Li, Xueyi Zou, and Shu-Tao Xia. Adaptive local implicit image function for arbitrary-scale super-resolution. In 2022 IEEE International Conference on Image Processing (ICIP), pages 4033--4037. IEEE, 2022

    2022

  19. [27]

    Gaussiandiffusion: 3d gaussian splatting for denoising diffusion probabilistic models with structured noise

    Xinhai Li, Huaibin Wang, and Kuo-Kun Tseng. Gaussiandiffusion: 3d gaussian splatting for denoising diffusion probabilistic models with structured noise. arXiv preprint arXiv:2311.11221, 2023

    2023 arXiv

  20. [28]

    Enhanced deep residual networks for single image super-resolution

    Bee Lim, Sanghyun Son, Heewon Kim, Seungjun Nah, and Kyoung Mu Lee. Enhanced deep residual networks for single image super-resolution. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops, pages 136--144, 2017

    2017

  21. [29]

    Mvsgaussian: Fast generalizable gaussian splatting reconstruction from multi-view stereo

    Tianqi Liu, Guangcong Wang, Shoukang Hu, Liao Shen, Xinyi Ye, Yuhang Zang, Zhiguo Cao, Wei Li, and Ziwei Liu. Mvsgaussian: Fast generalizable gaussian splatting reconstruction from multi-view stereo. In European Conference on Computer Vision, pages 37--53. Springer, 2024

    2024

  22. [30]

    A convnet for the 2020s

    Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, and Saining Xie. A convnet for the 2020s. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 11976--11986, 2022

    2022

  23. [31]

    Sgdr: Stochastic gradient descent with warm restarts

    Ilya Loshchilov and Frank Hutter. Sgdr: Stochastic gradient descent with warm restarts. arXiv preprint arXiv:1608.03983, 2016

    2016 arXiv

  24. [32]

    Scaffold-gs: Structured 3d gaussians for view-adaptive rendering

    Tao Lu, Mulin Yu, Linning Xu, Yuanbo Xiangli, Limin Wang, Dahua Lin, and Bo Dai. Scaffold-gs: Structured 3d gaussians for view-adaptive rendering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20654--20664, 2024

    2024

  25. [33]

    Acorn: Adaptive coordinate networks for neural scene representation

    Julien NP Martel, David B Lindell, Connor Z Lin, Eric R Chan, Marco Monteiro, and Gordon Wetzstein. Acorn: Adaptive coordinate networks for neural scene representation. arXiv preprint arXiv:2105.02788, 2021

    2021 arXiv

  26. [34]

    Nerf: Representing scenes as neural radiance fields for view synthesis

    B Mildenhall, PP Srinivasan, M Tancik, JT Barron, R Ramamoorthi, and R Ng. Nerf: Representing scenes as neural radiance fields for view synthesis. In European conference on computer vision, 2020

    2020

  27. [35]

    Deep generalized unfolding networks for image restoration

    Chong Mou, Qian Wang, and Jian Zhang. Deep generalized unfolding networks for image restoration. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 17399--17410, 2022

    2022

  28. [36]

    Instant neural graphics primitives with a multiresolution hash encoding

    Thomas M \"u ller, Alex Evans, Christoph Schied, and Alexander Keller. Instant neural graphics primitives with a multiresolution hash encoding. ACM transactions on graphics (TOG), 41 0 (4): 0 1--15, 2022

    2022

  29. [37]

    Wire: Wavelet implicit neural representations

    Vishwanath Saragadam, Daniel LeJeune, Jasper Tan, Guha Balakrishnan, Ashok Veeraraghavan, and Richard G Baraniuk. Wire: Wavelet implicit neural representations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18507--18516, 2023

    2023

  30. [38]

    Implicit neural representations with periodic activation functions

    Vincent Sitzmann, Julien Martel, Alexander Bergman, David Lindell, and Gordon Wetzstein. Implicit neural representations with periodic activation functions. Advances in neural information processing systems, 33: 0 7462--7473, 2020

    2020

  31. [39]

    Compact neural graphics primitives with learned hash probing

    Towaki Takikawa, Thomas M \"u ller, Merlin Nimier-David, Alex Evans, Sanja Fidler, Alec Jacobson, and Alexander Keller. Compact neural graphics primitives with learned hash probing. In SIGGRAPH Asia 2023 Conference Papers, pages 1--10, 2023

    2023

  32. [40]

    Fourier features let networks learn high frequency functions in low dimensional domains

    Matthew Tancik, Pratul Srinivasan, Ben Mildenhall, Sara Fridovich-Keil, Nithin Raghavan, Utkarsh Singhal, Ravi Ramamoorthi, Jonathan Barron, and Ren Ng. Fourier features let networks learn high frequency functions in low dimensional domains. Advances in neural information proc...

    2020

  33. [41]

    Dreamgaussian: Generative gaussian splatting for efficient 3d content creation

    Jiaxiang Tang, Jiawei Ren, Hang Zhou, Ziwei Liu, and Gang Zeng. Dreamgaussian: Generative gaussian splatting for efficient 3d content creation. arXiv preprint arXiv:2309.16653, 2023

    2023 arXiv

  34. [42]

    Random-access neural compression of material textures

    Karthik Vaidyanathan, Marco Salvi, Bartlomiej Wronski, Tomas Akenine-M \"o ller, Pontus Ebelin, and Aaron Lefohn. Random-access neural compression of material textures. arXiv preprint arXiv:2305.17105, 2023

    2023 arXiv

  35. [43]

    Mirage: Editable 2d images using gaussian splatting

    Joanna Waczy \'n ska, Tomasz Szczepanik, Piotr Borycki, S awomir Tadeja, Thomas Bohn \'e , and Przemys aw Spurek. Mirage: Editable 2d images using gaussian splatting. arXiv preprint arXiv:2410.01521, 2024

    2024 arXiv

  36. [44]

    Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction

    Peng Wang, Lingjie Liu, Yuan Liu, Christian Theobalt, Taku Komura, and Wenping Wang. Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction. arXiv preprint arXiv:2106.10689, 2021

    2021 arXiv

  37. [45]

    View-consistent 3d editing with gaussian splatting

    Yuxuan Wang, Xuanyu Yi, Zike Wu, Na Zhao, Long Chen, and Hanwang Zhang. View-consistent 3d editing with gaussian splatting. In European Conference on Computer Vision, pages 404--420. Springer, 2024

    2024

  38. [46]

    Learning dynamic scale awareness and global implicit functions for continuous-scale super-resolution of remote sensing images

    Hanlin Wu, Ning Ni, and Libao Zhang. Learning dynamic scale awareness and global implicit functions for continuous-scale super-resolution of remote sensing images. IEEE Transactions on Geoscience and Remote Sensing, 61: 0 1--15, 2023

    2023

  39. [47]

    Gaussian opacity fields: Efficient adaptive surface reconstruction in unbounded scenes

    Zehao Yu, Torsten Sattler, and Andreas Geiger. Gaussian opacity fields: Efficient adaptive surface reconstruction in unbounded scenes. ACM Transactions on Graphics (TOG), 43 0 (6): 0 1--13, 2024

    2024

  40. [48]

    Rade-gs: Rasterizing depth in gaussian splatting

    Baowen Zhang, Chuan Fang, Rakesh Shrestha, Yixun Liang, Xiaoxiao Long, and Ping Tan. Rade-gs: Rasterizing depth in gaussian splatting. arXiv preprint arXiv:2406.01467, 2024 a

    2024 arXiv

  41. [49]

    Gaussianimage: 1000 fps image representation and compression by 2d gaussian splatting

    Xinjie Zhang, Xingtong Ge, Tongda Xu, Dailan He, Yan Wang, Hongwei Qin, Guo Lu, Jing Geng, and Jun Zhang. Gaussianimage: 1000 fps image representation and compression by 2d gaussian splatting. In European Conference on Computer Vision, pages 327--345. Springer, 2024 b

    2024

  42. [51]

    Diffgs: Functional gaussian splatting diffusion

    Junsheng Zhou, Weiqi Zhang, and Yu-Shen Liu. Diffgs: Functional gaussian splatting diffusion. In Advances in Neural Information Processing Systems (NeurIPS), 2024

    2024

  43. [52]

    Image-gs: Content-adaptive image representation via 2d gaussians

    Yunxiang Zhang, Alexandr Kuznetsov, Akshay Jindal, Kenneth Chen, Anton Sochenov, Anton Kaplanyan, and Qi Sun. Image-gs: Content-adaptive image representation via 2d gaussians. arXiv preprint arXiv:2407.01866, 2024

    2024 arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.