Pith. sign in

REVIEW 4 major objections 5 minor 141 references

ATSplat: Compact Feed-forward 3D Gaussian Splatting with Adaptive Token Expansion

T0 review · 4 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

Pith's one-line read ATSplat restores scene-adaptive Gaussian allocation in feed-forward 3DGS by starting from sparse 3D anchor tokens and expanding only the tokens predicted to be under-reconstructed, matching or beating dense pixel-aligned baselines with more

desk verdict Core idea is solid and DL3DV results are strong, but the abstract overclaims SOTA and the high-res baseline is run off its training resolution. read the letter →

arxiv 2607.20417 v2 pith:JMSDEE2B submitted 2026-07-22 cs.CV

classification cs.CV
keywords 3DGaussianSplattingfeed-forwardreconstructionnovel-viewsynthesisadaptivecapacityallocationuncertainty-guidedexpansionsparsetokensmulti-viewreal-timerendering
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

ATSplat argues that the dense pixel-aligned design shared by most feed-forward 3D Gaussian Splatting methods is wasteful: when one Gaussian is regressed per input pixel, the Gaussian budget and its spatial distribution are set by image resolution and camera sampling rather than by scene complexity. The paper proposes instead a sparse-to-adaptive pipeline: coarse patch-level depths are unprojected into sparse 3D anchor tokens, each token decodes into a cluster of Gaussians whose centers are learned offsets from the anchor, and an Adaptive Token Expansion (ATE) module predicts which tokens sit in poorly reconstructed regions and expands those tokens. On RealEstate10K and DL3DV, the authors report that this formulation matches or exceeds the rendering quality of dense feed-forward baselines while using more than 5.7x fewer Gaussians, reconstructing a high-resolution scene in under a second on one GPU and rendering at 1136 FPS. A sympathetic reader would care because it suggests feed-forward 3DGS quality is governed by where capacity is allocated, not by how densely primitives are sampled.

What carries the argument

The two central components are the adaptive 3D anchor tokens and the Adaptive Token Expansion (ATE) module. Anchor tokens are obtained by unprojecting coarse patch features along predicted ray-depth and aggregating local 3D context; each token subsequently predicts K Gaussians whose centers are learned offsets from the anchor, so primitive placement is no longer constrained to input pixel grids. The ATE module is a lightweight MLP that outputs a per-token uncertainty score and is supervised by comparing a rasterized uncertainty map to the D-SSIM error map of intermediate renderings. At decoder blocks, the top fraction of tokens by predicted uncertainty are each replaced with M child tokens t

What would settle it

Measure the rank correlation between the predicted uncertainty scores and the actual D-SSIM error maps on a held-out dataset with different scene statistics than training (e.g., cross-dataset evaluation). If the correlation is near zero or negative, the expansion targets are effectively arbitrary and the claimed compression gains would not survive when the selection ratios are replaced by oracle or random selection; alternatively, compare against an oracle that expands the true highest-error regions and check whether the quality gap closes.

Watch

Extended reading notes

Core claim

The core discovery is that a feed-forward 3D Gaussian Splatting system can recover the scene-adaptive capacity allocation of optimization-based 3DGS. ATSplat does this by building a sparse 3D scaffold from coarse depth patches, decoding each anchor into K local Gaussians placed at learnable 3D offsets, and progressively expanding tokens whose predicted uncertainty exceeds a threshold, where the uncertainty head is trained to reproduce the model's own intermediate D-SSIM error maps. The reported result is that from 12 input images at 512x960, the model reconstructs a scene in less than a second into 311K Gaussians and renders novel views at 1136 FPS, matching or beating dense pixel-aligned fe

Load-bearing premise

The learned per-token uncertainty score, supervised only by the model's own intermediate rendering errors during training, continues to rank the right tokens as needing expansion on unseen scenes at inference.

Editorial extensions

If this is right

  • Feed-forward 3DGS can be compact without sacrificing rendering quality, since primitive count and placement become functions of scene complexity rather than input resolution or number of views.
  • A single forward pass on a commercial GPU can produce a high-resolution 3D representation in under a second and render it at interactive rates (1136 FPS), making on-the-fly reconstruction practical for real-time applications.
  • The uncertainty scores are a usable confidence map: regions that get expanded are exactly regions the model expects to render poorly, so the same signal can drive view selection, active capture, or downstream geometry tasks.
  • Because expansion happens inside the decoder, the additional tokens continue to cross-attend to input images, so allocated capacity comes with extra computation where it is needed, not just extra memory.
  • The authors' own dynamic-budget experiments show the fixed selection ratios can be replaced by uncertainty thresholds at inference, yielding scene-dependent Gaussian counts and pointing toward a fully adaptive compute allocator.

Reading between the lines

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

  • Editorial: If the uncertainty head stays calibrated across distribution shift, the same sparse-to-adaptive recipe could be applied to other primitive-based reconstruction models (e.g., large reconstruction models or voxel-based splatting) to reduce memory and bandwidth in on-device or streaming settings with minimal changes.
  • Editorial: A natural testbed is to make the selection policy explicitly budget-aware — e.g., train a controller to choose expansion ratios or thresholds per scene to hit a target Gaussian count — which would turn the current fixed-ratio mechanism into a true scene-dependent resource allocator.
  • Editorial: The paper's uncertainty signal could be reused as a heuristic for subsequent densification or pruning stages, effectively giving feed-forward methods the adaptive density-control loop that per-scene 3DGS enjoys, but in a single pass.
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

4 major / 5 minor

Summary. ATSplat is a feed-forward 3D Gaussian Splatting method that replaces pixel-aligned, per-pixel Gaussian prediction with a sparse set of 3D anchor tokens derived from coarse patch features and predicted depths. An Adaptive Token Expansion (ATE) module predicts per-token uncertainty scores, supervised by D-SSIM error maps of the model's own intermediate renderings, and selectively expands high-uncertainty tokens into additional Gaussians. The method is evaluated on RealEstate10K and DL3DV under standard protocols, including a high-resolution setting, and the paper claims state-of-the-art rendering quality with more than 5.7× fewer Gaussians than dense feed-forward baselines. The central technical contribution is the sparse-to-adaptive allocation that decouples primitive placement from the input image grid.

Significance. If the claims are substantiated, the work makes a useful contribution to feed-forward 3DGS by showing that adaptive capacity allocation — a key strength of per-scene optimized 3DGS — can be restored in a single forward pass. The proposed anchor-plus-offset design and the ATE module are clearly motivated, and the ablations (Tables 4, 5 and supplementary C) demonstrate that both components contribute to the reported gains. The paper also provides useful additional experiments: dynamic Gaussian budgets via uncertainty thresholds (Table 12), zero-shot transfer (Table 9), extrapolation (Table 10), and a unified model across datasets/view counts (Table 11). These strengthen the empirical case that the representation budget can be made scene-dependent rather than resolution-dependent. However, the headline claim that ATSplat achieves 'state-of-the-art rendering quality' is not consistently supported by the paper's own tables, and the high-resolution baseline comparison includes a protocol mismatch. The core efficiency idea is valuable, but the claims need to be narrowed and qualified.

major comments (4)
  1. [Abstract and Sec. 7.2, Table 1] The abstract and conclusion claim 'state-of-the-art rendering quality' with 5.7× fewer Gaussians, but Table 1 on RealEstate10K shows iLRM achieves higher PSNR (28.65 vs 28.46), higher SSIM (0.900 vs 0.894), and lower LPIPS (0.110 vs 0.118). ATSplat is not state-of-the-art on this dataset. Section 7.2 carefully says 'comparable to state-of-the-art feed-forward methods,' which is accurate, but the abstract and conclusion do not match. The claim should be narrowed to DL3DV or to 'state-of-the-art among compact/sub-40K-Gaussian methods.' This is a load-bearing overstatement of the central result.
  2. [Sec. 7.3, Table 3] The high-resolution DL3DV comparison includes DepthSplat evaluated at 512×960 using a checkpoint trained at 448×768. This is an off-protocol evaluation and makes DepthSplat's 21.33 PSNR not a fair feed-forward baseline; the gap to ATSplat (24.85) is thereby inflated. The paper acknowledges this in the caption, but the abstract and conclusion still rely on the resulting 'state-of-the-art' claim. I recommend either evaluating at DepthSplat's native resolution (with appropriate scaling) or clearly labeling the number as an out-of-distribution point and excluding it from SOTA claims. Also, the 5.7× reduction figure is computed against the densest pixel-aligned baselines (131K/688K); against iLRM the reduction is roughly 4.7× at high resolution (1474K/311K) and only 1.4× in the DL3DV 6-view setting (172K/120K). The quantitative claim should be stated relative to a specific baseline class.
  3. [Sec. 5, Eq. (8); Sec. 7.4, Fig. 4] The ATE module and the central efficiency claim depend on the learned uncertainty score u_i being a reliable proxy for actual rendering error on unseen scenes. The paper provides only qualitative visualizations (Fig. 4) and an ablation (Table 5) showing uncertainty-guided selection beats random selection. No quantitative calibration is reported, nor any analysis of failure cases or sensitivity to the fixed selection ratios ρ_l. Given the method's entire premise is that capacity is allocated to high-error regions, I ask for a quantitative validation, e.g., rank correlation between predicted uncertainty and D-SSIM, or an AUROC-style measure, and an ablation over selection ratios (the supplementary's threshold-based dynamic budgets partially address the latter, but not the calibration). This would substantiate the robustness of the adaptive mechanism.
  4. [Supplementary Sec. B.1, Table 7] The paper's own supplementary shows ATSplat falls behind pixel-aligned baselines in the large-overlap regime (e.g., PSNR 29.13 vs iLRM's 29.43, DepthSplat's 28.32). This is a legitimate limitation, but it is not mentioned in the main paper's discussion or conclusion. Since the paper claims general superiority of the anchor-offset design, this domain-dependent behavior should be disclosed and discussed in the main text, especially because it affects the interpretation of where the method is applicable.
minor comments (5)
  1. [Sec. 7.2, RealEstate10K paragraph] The phrase 'comparable to state-of-the-art feed-forward methods' is accurate, but the following sentence says '5.7× reduction compared to dense pixel-aligned formulations' — the comparison class is not stated. Please specify that the reduction is against pixelSplat/MVSplat/DepthSplat, not against iLRM.
  2. [Abstract and Fig. 1] The abstract claims real-time rendering at 1136 FPS (512×960), but no table or experiment in the main text measures rendering FPS. Please provide the measurement protocol (GPU, rasterizer, batch size) or remove the number from the abstract if it is not reproducible from the reported experiments.
  3. [Eq. (8)] The uncertainty loss uses L1 between predicted and stop-gradiated D-SSIM error maps. The choice of D-SSIM vs MSE is not discussed; a one-sentence justification would help, since the two losses emphasize different types of artifacts.
  4. [Table 2, footnote] The footnote says iLRM predicts Gaussians at 2× downsampled resolution, resulting in 4× fewer Gaussians. This is correct for a uniform downsampling, but iLRM's 172K at 6 views is only modestly fewer than the dense 688K of DepthSplat. A clear explanation of why iLRM is not considered a 'dense' baseline would avoid confusion about the 5.7× claim.
  5. [Sec. A.3, Eq. (11)] The notation N_init and N_final is clear, but the variable L_ATE is only defined in the text. Consider defining it in the equation or just below it to improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the uncertainty head is a self-supervised training signal, not a fitted prediction, and the central efficiency/quality claims are externally benchmarked on held-out splits.

full rationale

ATSplat's derivation chain is self-contained against external benchmarks. The sparse anchor scaffold is constructed from predicted patch depths (Eq. 3) and refined by cross-attention; Gaussian centers are anchor-plus-offset (Eqs. 4-5), so primitive placement is genuinely decoupled from pixel grids. The ATE uncertainty head (Eq. 6) is trained with L_unc (Eq. 8) to match the model's own intermediate D-SSIM error maps; this is an auxiliary self-supervised training signal, not a parameter fitted to the final evaluation metric and then renamed as a prediction. At inference the uncertainty head must generalize to new scenes, and the paper supports its usefulness with ablations (Tables 5 and 15) showing uncertainty-guided selection outperforms random selection, FPS, and no expansion. The central efficiency and rendering-quality claims are measured on held-out RealEstate10K and DL3DV splits (Tables 1-3) and are therefore externally falsifiable rather than forced by construction. The only author self-citation (FreeSplat/FreeSplat++, Sec. 2.3, Refs. [Wang et al. 2024, 2025b]) is a related-work description and is not load-bearing; no uniqueness theorem or ansatz is imported from prior work by the same authors. The paper does omit a quantitative calibration of predicted uncertainty versus actual error (Sec. 7.4 provides only qualitative visualizations), and the abstract's 'state-of-the-art' wording is not fully supported by Table 1, but these are correctness/support concerns, not circularity. No equation or claim in the paper reduces to its own inputs by definition, so the circularity score is 0.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

No new physical entities are postulated. 'Adaptive 3D anchor tokens' and 'child tokens' are internal learned representations without independent falsifiable handles, so they are not listed as invented physical entities. The central claim rests mainly on learned components (encoder, depth head, uncertainty head) and hand-set expansion hyperparameters.

free parameters (4)
  • Token selection ratios ρ_l = 0.5, 0.5, 0.25; 0.8 for first block at 512×960
    Hand-set hyperparameters that determine how many anchors are expanded at each decoder block; the high-resolution setting changes ρ_1, so the reported budget depends on this choice.
  • Expansion ratio M = 2
    Number of child tokens per expanded anchor; fixed across experiments and not ablated.
  • Gaussians per anchor K = 16
    Controls total Gaussian count; Table 13 shows quality still rises at K=32/64, so 16 is a budget tradeoff rather than a converged optimum.
  • Loss weights λ_p, λ_int, λ_unc = 0.5, 0.5, 0.1
    Training hyperparameters chosen in all experiments.
assumptions (4)
  • domain assumption Frozen DINOv2-B patch features provide a strong enough multi-view representation for coarse depth and anchor tokens.
    Multi-view encoder relies on DINOv2-B (§7.1, §A.2); no analysis of feature-sufficiency is given.
  • domain assumption Coarse patch-level depths from a lightweight MLP are accurate enough to form a useful 3D scaffold.
    Eq. 3 unprojects all anchors using these depths; ablations show anchors matter, but no depth accuracy evaluation or failure cases are provided.
  • domain assumption D-SSIM error maps at intermediate decoder blocks are a valid supervision target for uncertainty.
    Eq. 8 uses D-SSIM on the model's own intermediate renderings; no comparison to other error metrics or calibration analysis.
  • domain assumption Reported baseline numbers/checkpoints are directly comparable to ATSplat.
    Tables mix reported numbers and public checkpoints; DepthSplat is evaluated at 512×960 with a 448×768 checkpoint, so comparability is imperfect.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ATSplat: Compact Feed-forward 3D Gaussian Splatting with Adaptive Token Expansion." pith.science (2026). https://pith.science/paper/JMSDEE2B

@misc{pith2026260720417,
  author       = {Pith},
  title        = {Pith review of: ATSplat: Compact Feed-forward 3D Gaussian Splatting with Adaptive Token Expansion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JMSDEE2B}},
  note         = {Machine review of arXiv:2607.20417}
}
abstract

3D Gaussian Splatting (3DGS) achieves high-quality novel-view synthesis by optimizing freely placed primitives in 3D and adaptively densifying them in under-reconstructed regions. However, this scene-adaptive capacity allocation is largely lost in existing feed-forward 3DGS methods, which commonly regress Gaussians at input pixels and lift them along camera rays. Such pixel-aligned formulations make the number and placement of primitives depend on image resolution and input viewpoints rather than scene complexity, resulting in dense and often redundant Gaussian sets. We present ATSplat, a feed-forward 3DGS framework that restores the adaptive allocation capability of 3DGS optimization through Adaptive 3D Tokens. ATSplat first lifts coarse patch-level depth and camera cues into sparse 3D anchor tokens, forming a compact scaffold of the scene. Each token is then regressed into local Gaussians with learnable 3D offsets, decoupling primitive placement from input image grids. An Adaptive Token Expansion module predicts a token-level uncertainty score, supervised by rendering error maps, and selectively expands high-uncertainty tokens through learnable expansion layers. This sparse-to-adaptive formulation enables ATSplat to concentrate primitives in challenging regions while maintaining a compact representation. Experiments on two representative datasets, RealEstate10K and DL3DV, show that ATSplat achieves state-of-the-art rendering quality while reducing the number of Gaussians by more than $5.7\times$ compared with dense feed-forward 3DGS methods. From 12 input images at $512 \times 960$ resolution, ATSplat completes reconstruction in less than a second using a single commercial GPU, and renders high-quality novel views at 1136 FPS ($512 \times 960$) with only 311K Gaussians.

Figures

Figures reproduced from arXiv: 2607.20417 by the authors.

Figure 1
Figure 1. ATSplat reconstructs compact 3D Gaussians from multi-view captures in a single forward pass. Unlike dense pixel-aligned feed-forward methods that [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. (a) Framework overview. Given posed multi-view images, a multi-view image encoder extracts coarse patch features and estimates patch-level depths. The predicted depths and patch features are used to initialize a sparse set of 3D anchor tokens, serving as a sparse scene scaffold. An image-to-3D decoder refines these anchors through 𝐿𝑑 decoder blocks, with an Adaptive Token Expansion (ATE) module that allocates additi… view at source ↗
Figure 3
Figure 3. Visualization of expanded tokens. The coordinate of each ex￾panded token is computed as the mean center of its decoded Gaussians. geometry. In contrast, ATSplat begins with sparse anchors and se￾lectively expands them in uncertain regions, achieving high-quality results with a compact Gaussian set. 7.3 High-Resolution Novel-View Synthesis We further evaluate ATSplat on DL3DV at 512 × 960 resolution to demonstrate th… view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Visualization of uncertainty scores and selected tokens at the last block. We visualize actual error maps (D-SSIM), predicted uncertainty maps, and alpha maps of Gaussians decoded from the selected tokens. Behavior of expanded tokens. We first analyze how ATE increases…
Figure 5
Figure 5. Figure 5: Qualitative comparisons on RealEstate10K at 256 × 256 resolution with 2 input views. ATSplat reconstructs fine details in challenging regions while accurately reconstructing geometry, while using 5.7× fewer Gaussians compared to the dense pixel-aligned baselines. Input…
Figure 6
Figure 6. Figure 6: Qualitative comparisons on DL3DV at 256 × 448 resolution with 6 input views. ATSplat produces high-fidelity renderings with only 120K Gaussians, which is 5.7× fewer than DepthSplat. It also faithfully reconstructs thin structures, intricate details, and view-dependent …
Figure 7
Figure 7. Figure 7: Qualitative comparisons on DL3DV at 512 × 960 resolution with 12 input views. ATSplat recovers fine details and thin structures in less than a second, with a significantly reduced number of Gaussians, whereas 3DGS requires over 10 minutes of per-scene optimization to o…
Figure 8
Figure 8. Figure 8: Effect of progressively enabling token expansion across decoder blocks. Starting from a configuration with expansion disabled at all decoder blocks, we gradually enable expansion from the earliest block to the latest. As more blocks are activated, fine-grained texture …
Figure 9
Figure 9. Figure 9: Additional qualitative comparisons on RealEstate10K at 256 × 256 resolution with 2 input views [PITH_FULL_IMAGE:figures/full_fig_p016_9.png]
Figure 10
Figure 10. Figure 10: Additional Qualitative comparisons on DL3DV at 256 × 448 resolution with 6 input views [PITH_FULL_IMAGE:figures/full_fig_p017_10.png]
Figure 11
Figure 11. Figure 11: Additional Qualitative comparisons on DL3DV at 512 × 960 resolution with 12 input views [PITH_FULL_IMAGE:figures/full_fig_p018_11.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

141 extracted references · 7 linked inside Pith

  1. [1]

    Ye, Botao and Chen, Boqi and Xu, Haofei and Barath, Daniel and Pollefeys, Marc , journal =

  2. [2]

    , author=

    3d gaussian splatting for real-time radiance field rendering. , author=. ACM Trans. Graph. , volume=

  3. [3]

    and Tancik, Matthew and Barron, Jonathan T

    Mildenhall, Ben and Srinivasan, Pratul P. and Tancik, Matthew and Barron, Jonathan T. and Ramamoorthi, Ravi and Ng, Ren , booktitle =

  4. [4]

    and Mildenhall, Ben and Tancik, Matthew and Hedman, Peter and Martin-Brualla, Ricardo and Srinivasan, Pratul P

    Barron, Jonathan T. and Mildenhall, Ben and Tancik, Matthew and Hedman, Peter and Martin-Brualla, Ricardo and Srinivasan, Pratul P. , booktitle =

  5. [5]

    and Mildenhall, Ben and Verbin, Dor and Srinivasan, Pratul P

    Barron, Jonathan T. and Mildenhall, Ben and Verbin, Dor and Srinivasan, Pratul P. and Hedman, Peter , booktitle =

  6. [6]

    Yu, Alex and Li, Ruilong and Tancik, Matthew and Li, Hao and Ng, Ren and Kanazawa, Angjoo , booktitle =

  7. [7]

    Fridovich-Keil, Sara and Yu, Alex and Tancik, Matthew and Chen, Qinhong and Recht, Benjamin and Kanazawa, Angjoo , booktitle =

  8. [8]

    Chen, Anpei and Xu, Zexiang and Geiger, Andreas and Yu, Jingyi and Su, Hao , booktitle =

Show all 141 references
  1. [9]

    ACM Transactions on Graphics (SIGGRAPH) , volume =

    Instant Neural Graphics Primitives with a Multiresolution Hash Encoding , author =. ACM Transactions on Graphics (SIGGRAPH) , volume =. 2022 , doi =

  2. [10]

    Kerbl, Bernhard and Kopanas, Georgios and Leimk\"uhler, Thomas and Drettakis, George , journal =

  3. [11]

    Yu, Zehao and Chen, Anpei and Huang, Binbin and Sattler, Torsten and Geiger, Andreas , booktitle =

  4. [12]

    Charatan, David and Li, Sizhe Lester and Tagliasacchi, Andrea and Sitzmann, Vincent , booktitle =

  5. [13]

    Chen, Yuedong and Xu, Haofei and Zheng, Chuanxia and Zhuang, Bohan and Pollefeys, Marc and Geiger, Andreas and Cham, Tat-Jen and Cai, Jianfei , booktitle =

  6. [14]

    Xu, Yinghao and Shi, Zifan and Yifan, Wang and Chen, Hansheng and Yang, Ceyuan and Peng, Sida and Shen, Yujun and Wetzstein, Gordon , booktitle =

  7. [15]

    Tang, Jiaxiang and Chen, Zhaoxi and Chen, Xiaokang and Wang, Tengfei and Zeng, Gang and Liu, Ziwei , booktitle =

  8. [16]

    Zhang, Kai and Bi, Sai and Tan, Hao and Xiangli, Yuanbo and Zhao, Nanxuan and Sunkavalli, Kalyan and Xu, Zexiang , booktitle =

  9. [17]

    and Rupprecht, Christian and Vedaldi, Andrea , booktitle =

    Szymanowicz, Stanislaw and Insafutdinov, Eldar and Zheng, Chuanxia and Campbell, Dylan and Henriques, Jo\ ao F. and Rupprecht, Christian and Vedaldi, Andrea , booktitle =

  10. [18]

    Xu, Haofei and Peng, Songyou and Wang, Fangjinhua and Blum, Hermann and Barath, Daniel and Geiger, Andreas and Pollefeys, Marc , booktitle =

  11. [19]

    Kang, Gyeongjin and Nam, Seungtae and Sun, Xiangyu and Khamis, Sameh and Mohamed, Abdelrahman and Park, Eunbyung , journal =

  12. [20]

    and Zhuang, Bohan , journal =

    Wang, Weijie and Chen, Yeqing and Zhang, Zeyu and Liu, Hengyu and Wang, Haoxiao and Feng, Zhiyuan and Qin, Wenkang and Chen, Feng and Zhu, Zheng and Chen, Donny Y. and Zhuang, Bohan , journal =

  13. [21]

    , journal =

    Kim, Injae and Kim, Chaehyeon and Bae, Minseong and Joo, Minseok and Kim, Hyunwoo J. , journal =

  14. [22]

    Itkin, Roni and Issachar, Noam and Keypur, Yehonatan and Chen, Xingyu and Chen, Anpei and Benaim, Sagie , journal =

  15. [23]

    An, Honggyu and Jung, Jaewoo and Kim, Mungyeom and Kim, Chaehyun and Jeon, Minkyeong and Han, Jisang and Fukuda, Kazumi and Narihira, Takuya and Ko, Hyunah and Kim, Junsu and Hong, Sunghwan and Mitsufuji, Yuki and Kim, Seungryong , journal =

  16. [24]

    Zhang, Zicheng and Meng, Xiangting and Wu, Ke and Ding, Wenchao , booktitle =

  17. [25]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Pointnet: Deep learning on point sets for 3d classification and segmentation , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  18. [26]

    arXiv preprint arXiv:2604.15239 , year=

    TokenGS: Decoupling 3D Gaussian Prediction from Pixels with Learnable Tokens , author=. arXiv preprint arXiv:2604.15239 , year=

  19. [27]

    arXiv preprint arXiv:1805.09817 , year=

    Stereo magnification: Learning view synthesis using multiplane images , author=. arXiv preprint arXiv:1805.09817 , year=

  20. [28]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Dl3dv-10k: A large-scale scene dataset for deep learning-based 3d vision , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  21. [29]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    The unreasonable effectiveness of deep features as a perceptual metric , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  22. [30]

    International Conference on Learning Representations , volume=

    No pose, no problem: Surprisingly simple 3d gaussian splats from sparse unposed images , author=. International Conference on Learning Representations , volume=

  23. [31]

    arXiv preprint arXiv:2304.07193 , year=

    Dinov2: Learning robust visual features without supervision , author=. arXiv preprint arXiv:2304.07193 , year=

  24. [32]

    arXiv preprint arXiv:1711.05101 , year=

    Decoupled weight decay regularization , author=. arXiv preprint arXiv:1711.05101 , year=

  25. [33]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Long-lrm: Long-sequence large reconstruction model for wide-coverage gaussian splats , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  26. [34]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Generative densification: Learning to densify gaussians for high-fidelity generalizable 3d reconstruction , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

  27. [35]

    2019 , booktitle=

    Understanding Straight-Through Estimator in Training Activation Quantized Neural Nets , author=. 2019 , booktitle=

  28. [36]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Plenoxels: Radiance fields without neural networks , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  29. [37]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Efficient geometry-aware 3d generative adversarial networks , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  30. [38]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Ibrnet: Learning multi-view image-based rendering , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  31. [39]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Mvsnerf: Fast generalizable radiance field reconstruction from multi-view stereo , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  32. [40]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    pixelnerf: Neural radiance fields from one or few images , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  33. [41]

    ACM SIGGRAPH 2024 conference papers , pages=

    2d gaussian splatting for geometrically accurate radiance fields , author=. ACM SIGGRAPH 2024 conference papers , pages=

  34. [42]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Sugar: Surface-aligned gaussian splatting for efficient 3d mesh reconstruction and high-quality mesh rendering , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  35. [43]

    ACM Transactions on Graphics (ToG) , volume=

    Gaussian opacity fields: Efficient adaptive surface reconstruction in unbounded scenes , author=. ACM Transactions on Graphics (ToG) , volume=. 2024 , publisher=

  36. [44]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    4d gaussian splatting for real-time dynamic scene rendering , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  37. [45]

    International Conference on Learning Representations , volume=

    Real-time photorealistic dynamic scene representation and rendering with 4d gaussian splatting , author=. International Conference on Learning Representations , volume=

  38. [46]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Scaffold-gs: Structured 3d gaussians for view-adaptive rendering , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  39. [47]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Splatter image: Ultra-fast single-view 3d reconstruction , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  40. [48]

    Advances in Neural Information Processing Systems , volume=

    Mvsplat360: Feed-forward 360 scene synthesis from sparse views , author=. Advances in Neural Information Processing Systems , volume=

  41. [49]

    arXiv preprint arXiv:2410.22128 , year=

    Pf3plat: Pose-free feed-forward 3d gaussian splatting , author=. arXiv preprint arXiv:2410.22128 , year=

  42. [50]

    ACM Transactions on Graphics (TOG) , volume=

    Anysplat: Feed-forward 3d gaussian splatting from unconstrained views , author=. ACM Transactions on Graphics (TOG) , volume=. 2025 , publisher=

  43. [51]

    Advances in Neural Information Processing Systems , volume=

    Freesplat: Generalizable 3d gaussian splatting towards free view synthesis of indoor scenes , author=. Advances in Neural Information Processing Systems , volume=

  44. [52]

    arXiv preprint arXiv:2503.22986 , year=

    Freesplat++: Generalizable 3d gaussian splatting for efficient indoor scene reconstruction , author=. arXiv preprint arXiv:2503.22986 , year=

  45. [53]

    Advances in Neural Information Processing Systems , volume=

    Gaussian graph network: Learning efficient and generalizable gaussian representations from multi-view images , author=. Advances in Neural Information Processing Systems , volume=

  46. [54]

    Advances in Neural Information Processing Systems , volume=

    Learning Efficient Fuse-and-Refine for Feed-Forward 3D Gaussian Splatting , author=. Advances in Neural Information Processing Systems , volume=

  47. [55]

    arXiv preprint arXiv:2512.15508 , year=

    Off The Grid: Detection of Primitives for Feed-Forward 3D Gaussian Splatting , author=. arXiv preprint arXiv:2512.15508 , year=

  48. [56]

    arXiv preprint arXiv:2604.07053 , year=

    AnchorSplat: Feed-Forward 3D Gaussian SplattingWith 3D Geometric Priors , author=. arXiv preprint arXiv:2604.07053 , year=

  49. [57]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Structure-from-motion revisited , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  50. [58]

    Abril and Robert Plant

    Patricia S. Abril and Robert Plant. The patent holder's dilemma: Buy, sell, or troll?. Communications of the ACM. 2007. doi:10.1145/1188913.1188915

  51. [59]

    Deciding equivalances among conjunctive aggregate queries

    Sarah Cohen and Werner Nutt and Yehoshua Sagic. Deciding equivalances among conjunctive aggregate queries. 2007. doi:10.1145/1219092.1219093

  52. [60]

    Special issue: Digital Libraries. 1996

  53. [61]

    Understanding Policy-Based Networking

    David Kosiur. Understanding Policy-Based Networking. 2001

  54. [64]

    The title of book two. 2008. doi:10.1007/3-540-09237-4

  55. [65]

    Asad Z. Spector. Achieving application requirements. Distributed Systems. 1990. doi:10.1145/90417.90738

  56. [66]

    Douglass and David Harel and Mark B

    Bruce P. Douglass and David Harel and Mark B. Trakhtenbrot. Statecarts in use: structured analysis and object-orientation. Lectures on Embedded Systems. 1998. doi:10.1007/3-540-65193-4_29

  57. [67]

    Donald E. Knuth. The Art of Computer Programming, Vol. 1: Fundamental Algorithms (3rd. ed.). 1997

  58. [68]

    Donald E. Knuth. The Art of Computer Programming. 1998

  59. [69]

    Structured Variational Inference Procedures and their Realizations (as incol)

    Dan Geiger and Christopher Meek. Structured Variational Inference Procedures and their Realizations (as incol). Proceedings of Tenth International Workshop on Artificial Intelligence and Statistics, The Barbados

  60. [70]

    Stan W. Smith. An experiment in bibliographic mark-up: Parsing metadata for XML export. Proceedings of the 3rd. annual workshop on Librarians and Computers. 2010. doi:99.9999/woot07-S422

  61. [71]

    Catch me, if you can: Evading network signatures with web-based polymorphic worms

    Matthew Van Gundy and Davide Balzarotti and Giovanni Vigna. Catch me, if you can: Evading network signatures with web-based polymorphic worms. Proceedings of the first USENIX workshop on Offensive Technologies. 2007

  62. [72]

    Catch me, if you can: Evading network signatures with web-based polymorphic worms

    Matthew Van Gundy and Davide Balzarotti and Giovanni Vigna. Catch me, if you can: Evading network signatures with web-based polymorphic worms. Proceedings of the first USENIX workshop on Offensive Technologies. 2008

  63. [73]

    Catch me, if you can: Evading network signatures with web-based polymorphic worms

    Matthew Van Gundy and Davide Balzarotti and Giovanni Vigna. Catch me, if you can: Evading network signatures with web-based polymorphic worms. Proceedings of the first USENIX workshop on Offensive Technologies. 2009

  64. [74]

    Predicate Path expressions

    Sten Andler. Predicate Path expressions. Proceedings of the 6th. ACM SIGACT-SIGPLAN symposium on Principles of Programming Languages. 1979. doi:10.1145/567752.567774

  65. [75]

    LOGICS of Programs: AXIOMATICS and DESCRIPTIVE POWER

    David Harel. LOGICS of Programs: AXIOMATICS and DESCRIPTIVE POWER. 1978

  66. [76]

    Anisi , title =

    David A. Anisi , title =

  67. [77]

    Clarkson

    Kenneth L. Clarkson. Algorithms for Closest-Point Problems (Computational Geometry). 1985

  68. [78]

    Introduction to Bayesian Statistics

    Harry Thornburg. Introduction to Bayesian Statistics. 2001

  69. [79]

    CLIFFORD: a Maple 11 Package for Clifford Algebra Computations, version 11

    Rafal Ablamowicz and Bertfried Fauser. CLIFFORD: a Maple 11 Package for Clifford Algebra Computations, version 11. 2007

  70. [80]

    Stats and Analysis

    Poker-Edge.Com. Stats and Analysis. 2006

  71. [81]

    A more perfect union

    Barack Obama. A more perfect union. 2008

  72. [82]

    The fountain of youth

    Joseph Scientist. The fountain of youth. 2009

  73. [83]

    Solder man

    Dave Novak. Solder man. ACM SIGGRAPH 2003 Video Review on Animation theater Program: Part I - Vol. 145 (July 27--27, 2003). 2003. doi:99.9999/woot07-S422

  74. [84]

    Interview with Bill Kinder: January 13, 2005

    Newton Lee. Interview with Bill Kinder: January 13, 2005. Comput. Entertain. 2005. doi:10.1145/1057270.1057278

  75. [85]

    The Enabling of Digital Libraries

    Bernard Rous. The Enabling of Digital Libraries. Digital Libraries. 2008

  76. [87]

    (new) Finding minimum congestion spanning trees , journal =

    Werneck, Renato and Setubal, Jo\. (new) Finding minimum congestion spanning trees , journal =. 2000 , issn =. doi:10.1145/351827.384253 , acmid =

  77. [89]

    and Mei, Alessandro , title =

    Conti, Mauro and Di Pietro, Roberto and Mancini, Luigi V. and Mei, Alessandro , title =. Inf. Fusion , volume =. 2009 , issn =. doi:10.1016/j.inffus.2009.01.002 , acmid =

  78. [90]

    and Hutchful, David K

    Li, Cheng-Lun and Buyuktur, Ayse G. and Hutchful, David K. and Sant, Natasha B. and Nainwal, Satyendra K. , title =. CHI '08 extended abstracts on Human factors in computing systems , year =. doi:10.1145/1358628.1358946 , acmid =

  79. [91]

    , title =

    Hollis, Billy S. , title =. 1999 , isbn =

  80. [92]

    Goossens, Michel and Rahtz, S. P. and Moore, Ross and Sutor, Robert S. , title =. 1999 , isbn =

  81. [93]

    and Rosenberg, Arnold L

    Buss, Jonathan F. and Rosenberg, Arnold L. and Knott, Judson D. , title =. 1987 , source =

  82. [94]

    CHI '08: CHI '08 extended abstracts on Human factors in computing systems , year =

    , note =. CHI '08: CHI '08 extended abstracts on Human factors in computing systems , year =

  83. [95]

    Algorithms for Closest-Point Problems (Computational Geometry) , year =

    Clarkson, Kenneth Lee , advisor =. Algorithms for Closest-Point Problems (Computational Geometry) , year =

  84. [96]

    SIGCOMM Comput. Commun. Rev. , year =

  85. [97]

    2004 , isbn =

    IEEE TCSC Executive Committee , booktitle =. 2004 , isbn =. doi:http://dx.doi.org/10.1109/ICWS.2004.64 , acmid =

  86. [98]

    Distributed systems (2nd Ed.) , year =

  87. [99]

    , title =

    Petrie, Charles J. , title =. 1986 , source =

  88. [100]

    Donald E. Knuth. Seminumerical Algorithms. 1981

  89. [101]

    E-commerce and cultural values , year =

    Kong, Wei-Chang , Title =. E-commerce and cultural values , year =

  90. [102]

    E-commerce and cultural values , year =

    Kong, Wei-Chang , type =. E-commerce and cultural values , year =

  91. [103]

    Chapter 9 , booktitle =

    Kong, Wei-Chang , editor =. Chapter 9 , booktitle =. 2002 , address =

  92. [104]

    E-commerce and cultural values , editor =

    Kong, Wei-Chang , title =. E-commerce and cultural values , editor =. 2003 , isbn =

  93. [105]

    E-commerce and cultural values - (InBook-num-in-chap) , chapter =

    Kong, Wei-Chang , editor =. E-commerce and cultural values - (InBook-num-in-chap) , chapter =. 2004 , address =

  94. [106]

    E-commerce and cultural values (Inbook-text-in-chap) , chapter =

    Kong, Wei-Chang , editor =. E-commerce and cultural values (Inbook-text-in-chap) , chapter =. 2005 , address =

  95. [107]

    E-commerce and cultural values (Inbook-num chap) , chapter =

    Kong, Wei-Chang , editor =. E-commerce and cultural values (Inbook-num chap) , chapter =. 2006 , address =

  96. [108]

    Microelectron

    Mehdi Saeedi and Morteza Saheb Zamani and Mehdi Sedighi , title =. Microelectron. J. , volume =. 2010 , pages =

  97. [109]

    Mehdi Saeedi and Morteza Saheb Zamani and Mehdi Sedighi and Zahra Sasanian , title =. J. Emerg. Technol. Comput. Syst. , volume =

  98. [110]

    Kirschmer, Markus and Voight, John , title =. SIAM J. Comput. , issue_date =. 2010 , issn =. doi:https://doi.org/10.1137/080734467 , acmid =

  99. [111]

    Hoare, C. A. R. , title =. Structured programming (incoll) , editor =. 1972 , isbn =

  100. [112]

    History of programming languages I (incoll) , editor =

    Lee, Jan , title =. History of programming languages I (incoll) , editor =. 1981 , isbn =. doi:http://doi.acm.org/10.1145/800025.1198348 , acmid =

  101. [113]

    , title =

    Dijkstra, E. , title =. Classics in software engineering (incoll) , year =

  102. [114]

    , title =

    Wenzel, Elizabeth M. , title =. Multimedia interface design (incoll) , year =. doi:10.1145/146022.146089 , acmid =

  103. [115]

    , title =

    Mumford, E. , title =. Critical issues in information systems research (incoll) , year =

  104. [116]

    and Golden, Donald G

    McCracken, Daniel D. and Golden, Donald G. , title =. 1990 , isbn =

  105. [117]

    The analysis of linear partial differential operators

    H. The analysis of linear partial differential operators. 1985 , PAGES =

  106. [118]

    IEEE", address =

    A. Adya and P. Bahl and J. Padhye and A.Wolman and L. Zhou , title =. Proceedings of the IEEE 1st International Conference on Broadnets Networks (BroadNets'04) , publisher = "IEEE", address = "Los Alamitos, CA", year =

  107. [119]

    I. F. Akyildiz and W. Su and Y. Sankarasubramaniam and E. Cayirci , title =. Comm. ACM , volume = 38, number = "4", year =

  108. [120]

    I. F. Akyildiz and T. Melodia and K. R. Chowdhury , title =. Computer Netw. , volume = 51, number = "4", year =

  109. [121]

    ACM", address =

    P. Bahl and R. Chancre and J. Dungeon , title =. Proceeding of the 10th International Conference on Mobile Computing and Networking (MobiCom'04) , publisher = "ACM", address = "New York, NY", year =

  110. [122]

    8 (Special Issue on Sensor Networks)

    D. Culler and D. Estrin and M. Srivastava , title =. IEEE Comput. , volume = 37, number = "8 (Special Issue on Sensor Networks)", publisher = "IEEE", address = "Los Alamitos, CA", year =

  111. [123]

    Natarajan and M

    A. Natarajan and M. Motani and B. de Silva and K. Yap and K. C. Chua , title =. Network Architectures , editor =. 960935712

  112. [124]

    Tzamaloukas and J

    A. Tzamaloukas and J. J. Garcia-Luna-Aceves , title =

  113. [125]

    Zhou and J

    G. Zhou and J. Lu and C.-Y. Wan and M. D. Yarvis and J. A. Stankovic , title =

  114. [126]

    Mapping Powerlists onto Hypercubes

    Jacob Kornerup. Mapping Powerlists onto Hypercubes. 1994

  115. [127]

    Automatic Parallelization for Distributed-Memory Multiprocessing Systems

    Michael Gerndt. Automatic Parallelization for Distributed-Memory Multiprocessing Systems

  116. [128]

    J. E. Archer, Jr. and R. Conway and F. B. Schneider. User recovery and reversal in interactive systems. ACM Trans. Program. Lang. Syst

  117. [129]

    D. D. Dunlop and V. R. Basili. Generalizing specifications for uniformly implemented loops. ACM Trans. Program. Lang. Syst

  118. [130]

    Heering and P

    J. Heering and P. Klint. Towards monolingual programming environments. ACM Trans. Program. Lang. Syst

  119. [131]

    Donald E. Knuth. The book

  120. [132]

    Korach and D

    E. Korach and D. Rotem and N. Santoro. Distributed algorithms for finding centers and medians in networks. ACM Trans. Program. Lang. Syst

  121. [133]

    : A Document Preparation System

    Leslie Lamport. : A Document Preparation System

  122. [134]

    F. Nielson. Program transformations in a denotational setting. ACM Trans. Program. Lang. Syst

  123. [135]

    Brian K. Reid. A high-level approach to computer document formatting. Proceedings of the 7th Annual Symposium on Principles of Programming Languages

  124. [136]

    and Abdelzaher, Tarek F

    Zhou, Gang and Wu, Yafeng and Yan, Ting and He, Tian and Huang, Chengdu and Stankovic, John A. and Abdelzaher, Tarek F. , title =. ACM Trans. Embed. Comput. Syst. , issue_date =. doi:10.1145/1721695.1721705 , acmid = 1721705, publisher =

  125. [137]

    Institutional members of the Users Group

  126. [138]

    Boris Veytsman , title =

  127. [139]

    and Peterson, Larry L

    Bowman, Mic and Debray, Saumya K. and Peterson, Larry L. , title =. ACM Trans. Program. Lang. Syst. , volume =. 1993 , doi =

  128. [140]

    TUGboat , volume =

    Braams, Johannes , title =. TUGboat , volume =

  129. [141]

    Post Congress Tristesse

    Malcolm Clark. Post Congress Tristesse. TeX90 Conference Proceedings

  130. [142]

    ACM Trans

    Herlihy, Maurice , title =. ACM Trans. Program. Lang. Syst. , volume =. 1993 , doi =

  131. [143]

    Salas and Einar Hille

    S.L. Salas and Einar Hille. Calculus: One and Several Variable. 1978

  132. [144]

    Publication quality tables in

    Simon Fear , month =. Publication quality tables in

  133. [145]

    Using the amsthm Package , organization =

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.