HybridGS: High-Efficiency Gaussian Splatting Data Compression using Dual-Channel Sparse Representation and Point Cloud Encoder

Geert Van der Auwera; Le Yang; Qi Yang; Zhu Li

REVIEW 3 major objections 6 minor 2 cited by

HybridGS: High-Efficiency Gaussian Splatting Data Compression using Dual-Channel Sparse Representation and Point Cloud Encoder

T0 review · 3 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read HybridGS compresses 3D Gaussian Splatting scenes into standard point-cloud bitstreams with roughly two seconds of coding time, at reconstruction quality comparable to state-of-the-art generative compression methods like HAC.

desk verdict HybridGS is a useful, reproducible integration of quantization-aware 3DGS generation with GPCC, but the headline 0-2s encoding claim is not supported because it excludes the 70,000-epoch training stage that baselines include in their reported times. read the letter →

arxiv 2505.01938 v1 pith:EOT2UHS6 submitted 2025-05-03 cs.CV eess.IV

Qi Yang , Le Yang , Geert Van Der Auwera , Zhu Li This is my paper · ORCID

classification cs.CVeess.IV

keywords 3DGaussianSplattingcompressionpointcloudGPCCdual-channelsparserepresentationquantization-awaretrainingratecontrolnovelviewsynthesisprimitivepruning

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

HybridGS addresses the practical bottleneck of 3D Gaussian Splatting (3DGS) compression: today's best methods embed scene data in bespoke neural formats and take from tens of seconds to over a minute to encode. The paper argues this trade-off is unnecessary. It builds a two-stage pipeline that first generates an explicit, compact 3DGS representation with integer positions and quantized low-dimensional latent features, then compresses that file with a standard point cloud codec (GPCC). The reported result is reconstruction quality close to state-of-the-art generative compression while encoding and decoding typically finish in 0 to 2 seconds, producing standardized bitstreams. The claim matters because it points to a route where 3DGS streaming can inherit mature, standard codec infrastructure instead of waiting on bespoke neural decoders.

What carries the argument

The load-bearing mechanism is the dual-channel sparse representation paired with a standard downstream encoder. 'Dual channel' denotes two kinds of sparsity imposed during generation: attribute sparsity, in which compressible features (color and rotation) are replaced by low-dimensional latent codes decoded by a one-hidden-layer MLP — a trained analogue of PCA — and precision sparsity, in which all attributes are quantized to a chosen bit depth inside the training loop using the Robust Quantizer, whose perturbation-injected affine transform at quantization and ridge-regression reconstruction at de-quantization keep quantization differentiable. For positions, the Learnable Quantizer-based Method (LQM) translates and rescales the scene into the integer range of a chosen bit depth, decomposes each integer coordinate as an inner product of a fixed basis vector and a learnable coding vector over $\{-1,0,1\}$, and enforces primitive uniqueness with progressive pruning, so positions render directly without de-quantization. The output is an explicit integer file whose per-primitive bit cost $3(\mathrm{BD}_p+\mathrm{BD}_s)+k_c\mathrm{BD}_c+\mathrm{BD}_o+k_r\mathrm{BD}_r$ is known in closed form; GPCC then encodes geometry by octree and attributes by RAHT, and that same formula is what makes rate control possible — prune primitives to hit a target count, or reduce attribute bit depth uniformly.

What would settle it

Run both pipelines end-to-end on the same hardware and scenes: HybridGS in full (70,000-epoch generation plus GPCC coding) and a state-of-the-art generative method such as HAC, measuring total wall-clock time to a compressed bitstream and the resulting PSNR. If the generation training is counted, the speed advantage reported in Table 2 (0.66 to 1.67 seconds versus 17 to 132 seconds) shrinks to roughly the training-time ratio, which directly decides whether the speed claim describes the whole compression system or only its last stage.

Watch

Extended reading notes

Core claim

The central claim is that 3DGS compression does not have to sacrifice speed for compactness. Because 3DGS is surjective — many distinct primitive sets render to nearly identical views — the generation stage can be steered toward a representation that a conventional codec handles well, and HybridGS does exactly that in two steps. First, a dual-channel sparse representation supervises the primitive position field and the attribute bit depth during training: color and rotation are reconstructed from low-dimensional latent codes through a lightweight one-hidden-layer decoder, a trainable analogue of PCA; attributes are quantized during training with a robust quantizer; and positions are converted to unique integers through a learnable basis-vector and coding-vector decomposition, with progressive pruning and deduplication governing the primitive count. Second, the resulting explicit file is encoded by GPCC, with positions in octree mode and attributes via RAHT. On five benchmark scenes the paper reports PSNR within a small margin of HAC and CompGS(MM) while cutting coding time from tens of seconds to well under two seconds. Because HybridGS deliberately omits modules that would improve rendering quality, its declared quality ceiling is vanilla 3DGS itself, which the authors treat as a feature: compression loss stays interpretable.

Load-bearing premise

The 0 to 2 second encoding claim counts only the final GPCC point-cloud coding pass; the roughly 70,000-epoch training run that produces the compact representation is not included in that time, even though the competing methods' published encoding times include their entire compression process.

Editorial extensions

If this is right

Coding time for 3DGS scenes drops from the tens of seconds to minutes reported for HAC, CompGS(MM), and HGSC down to roughly 0 to 2 seconds, bringing interactive streaming of preprocessed Gaussian scenes into reach.
Because the output is a standard point-cloud bitstream, 3DGS compression can ride on existing codec deployments and hardware rather than requiring a custom neural decoder at the receiver.
Rate control becomes predictable: target bitrates are reached by pruning primitives or lowering feature bit depth against a known per-primitive bit budget, whereas RD-trained generative methods only discover their achieved rate after optimization.
The deliberate absence of quality-improving modules caps HybridGS at vanilla 3DGS quality, and the paper reports that pruning can occasionally even raise PSNR slightly by removing redundant primitives — evidence that primitive count is not a strict quality ceiling.

Reading between the lines

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

Read as an end-to-end system, HybridGS suits offline asset preparation: the 70,000-epoch training that creates each compact scene happens once, and the 0 to 2 second figure covers only the coding pass, so the practical niche is fast streaming or delivery of already-processed scenes, not live encoding.
The paper's own numbers show GPCC's lossless ratio on 3DGS data (about 1.3 to 1.5 times) is far below its 3 to 4 times on dense point clouds; codec-side work tuned to 3DGS's locally dense, globally sparse geometry could roughly double the compression at no quality cost.
Because position uniqueness is enforced during generation, the explicit integer files are also ready-made inputs for learning-based point cloud codecs, whose destructive deduplication of duplicated points would otherwise be a source of loss.
The reported divergence between training-view and test-view PSNR under pruning suggests that compression-quality studies on 3DGS should report both; a method could look better on held-out views than it actually is on the views it was trained on.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Colleague,

The thing to know: HybridGS is a genuine integration, not a new theoretical result. It combines quantization-aware 3DGS generation (low-rank latent features, robust quantizer, integer unique positions) with GPCC encoding, ships code, and reports rate control. That is worth a look. But the headline speed claim—encoding/decoding “typically between 0s and 2s”—is not supported by the measurements. Section 4.2.2 times only the GPCC tail on an already-trained representation and excludes data I/O; the 70,000-epoch generation/training step that produces the compact 3DGS is never timed. Baseline numbers like HAC’s 85s include full compression optimization. Counting the training stage would put HybridGS far outside the 0-2s range. The decoding number also leaves out the dequantization and MLP decoding time that the paper reports separately.

What is genuinely useful: the LQM for integer unique positions, the two-stage rate control, and the clean demonstration that a standard point cloud codec can be applied after quantization-aware generation. The ablation material (uniqueness, pruning, outlier removal) is reasonably thorough, and the code being public helps reproducibility. The paper’s own limitation statement admits that compression efficiency is below end-to-end RD-optimized methods, which is consistent with Table 1: on playroom, HAC reaches 30.84 dB at 6.86 MB while HybridGS gets 29.89 dB at 12.15 MB. “Comparable” overstates it.

Soft spots besides timing: only five scenes in the main table, RD curves on two scenes, and the rate-control formula uses L=1.3 as a single fitted constant, disclosed as preliminary. None of these are disqualifying by themselves. The citation pattern looks fine—relevant MPEG documents and recent compression work are cited.

Bottom line: the paper deserves a serious referee, but not in current form. The speed comparison needs to be redone end-to-end, with training time counted or explicitly excluded and clearly stated as generation time rather than encoding time. If the authors fix the measurement boundary and soften the abstract, this is a solid technical report for the MPEG/3DGS streaming crowd. I would not cite the 0-2s claim, but I would cite the LQM and rate-control pieces.

Referee Report

3 major / 6 minor

Summary. The paper proposes HybridGS, a two-stage 3D Gaussian Splatting (3DGS) compression framework. In the first stage, a dual-channel sparse representation is trained for 70,000 epochs, producing an explicit, compact 3DGS point cloud with quantized integer positions and low-dimensional latent features for color and rotation. In the second stage, the resulting point cloud is compressed with the standard GPCC codec, using octree coding for positions and RAHT for attributes, producing a standard bitstream. The paper claims that this design provides reconstruction quality comparable to state-of-the-art generative 3DGS compression methods while being much faster, with encoding and decoding 'typically ranging between 0s and 2s'. It also introduces two rate-control mechanisms based on primitive pruning and bit-depth reduction, supported by a formula that estimates bitrate using a fixed lossless compression ratio L=1.3.

Significance. If the speed and quality claims were supported, the work would be valuable for practical 3DGS streaming and standardization, since it would demonstrate that standard point cloud codecs can be used on compact, quantization-aware 3DGS representations. The paper includes several positive elements: the code is publicly available; the experiments cover multiple datasets and scenes; the appendix provides ablations for position uniqueness, outlier removal, and different quantizers; and the underlying idea of coupling quantization-aware training with a standard codec is coherent and worth pursuing. However, the central speed claim is not established by the reported measurements, and the reconstruction-quality comparison is overstated relative to the actual tables. These issues affect the paper's headline contributions and cannot be resolved by minor edits.

major comments (3)

[Abstract and Section 4.2.2, Table 2] The claim that HybridGS encoding/decoding 'typically ranges between 0s and 2s' is not supported by the experiments. Section 4.2.2 states that the reported times are 'CPU computation time for HybridGS with GPCC' and that 'Data I/O time is excluded', so Table 2 measures only the GPCC tail of the pipeline after the 70,000-epoch 3DGS training stage (Section 4.1). For a compression system, encoding is the full path from source input to bitstream; the comparison methods' times, such as HAC's 85.03s and CompGS(MM)'s 36.29s, include their full end-to-end optimization. If the 70,000-epoch training is counted, HybridGS's encoding time is not 'between 0s and 2s' and is not 'evidently' faster than the baselines. Because the speed advantage is the paper's headline contribution, this is a load-bearing flaw.
[Section 4.2.1 and Table 1] The abstract and Section 1 state that HybridGS 'provides comparable reconstruction performance against state-of-the-art methods', but the quantitative results do not support this. On 'playroom', HAC (λ=0.0005) achieves 30.84 dB PSNR at 6.86 MB, whereas HybridGS (kc=6, HR) achieves 29.89 dB at 16.08 MB; on 'bicycle', HAC (λ=0.004) achieves 24.81 dB at 26.99 MB, whereas HybridGS (kc=6, HR) achieves 24.10 dB at 30.21 MB. In these and most other rows, HybridGS is worse on both rate and distortion than HAC. The paper's own Limitations section concedes that 'the optimal compression efficiency of HybridGS is lower than end-to-end generation compression methods using RD loss as supervision.' The 'comparable' wording is therefore misleading and should be revised to a more precise statement of the trade-off.
[Table 2 and Section 4.1] The coding-time comparison in Table 2 is not apples-to-apples even beyond the measurement-boundary issue. HybridGS timings were measured on the authors' machine ('Intel Core i9-14900HX, NVIDIA RTX 4090 Laptop', Section A.1), while the comparison times for HAC, CompGS(MM), and HGSC are taken from their original papers or official code (Section 4.1), which may involve different hardware, software stacks, and optimization levels. A fair speed comparison requires either re-running the baselines on the same machine with the same measurement protocol, or clearly stating the hardware configuration of each reported number. As presented, the speed advantage is confounded by both unequal scope and unequal platform.

minor comments (6)

[Section 3.2.2 and Table 4] The rate-control formula in Eq. (8) uses a fixed lossless compression ratio L=1.3, but Table 4 shows rate errors such as a target of 10 MB for 'train' producing 8.59 MB, and the text attributes this to point density influencing L. The fixed-L assumption should be presented with an explicit caveat or replaced by a density-dependent estimate, since the authors already note that 'the denser the point cloud, the higher the compression ratio'.
[Section 4.2.2] The statement 'Data I/O time is excluded' appears only in the experimental section. Since the abstract's 'between 0s and 2s' claim is central to the paper's framing, the measurement scope should be stated clearly wherever the speed claim is made, otherwise readers will reasonably interpret the numbers as end-to-end encoding time.
[Table 2] Table 2 reports coding times for only 'bicycle' and 'room', although Table 1 includes results for five scenes. Adding timing results for the other scenes would strengthen the generality of the speed claim.
[Table 3] Table 3's header 'Total Size' is ambiguous because it includes pre-GPCC sizes in parentheses and the bitstream size as the outer value. Clarifying that 'Total Size' is the GPCC output size and the parenthetical values are the explicit 3DGS file sizes would improve readability.
[Appendix A.6, Table 7] The dataset name 'drjoshson' is a typo and should read 'drjohnson'.
[Section 4.1] The sentence selecting 'the samples in the 50,000 and 70,000 epochs as High and Low (i.e., pruning 47% and 75% primitives)' would benefit from a brief explanation of why these two epochs were chosen, especially for reproducibility of the reported HR/LR points.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the central claims rest on external baselines and disclosed measurements; the speed comparison asymmetry is a measurement-boundary issue, not a circular derivation.

full rationale

HybridGS's derivation chain does not reduce any claimed result to its inputs by construction. The compact-representation generation (LQM, robust quantization, latent-feature decoding) is trained against a rendering loss, and the output sizes and PSNRs are measured comparisons against external baselines (HAC, CompGS, GGSC, HGSC). The appendix honestly states a limitation: the optimal compression efficiency of HybridGS is lower than end-to-end generation compression methods using RD loss. The abstract's 0-2s encoding claim is an empirical measurement-boundary question, because Section 4.2.2 records only the GPCC CPU time and excludes data I/O and the 70,000-epoch generation stage, while baseline times include their full compression optimization; this asymmetry is a correctness or measurement-scope concern, not a circular reduction, since no equation or parameter is defined in terms of the claimed outcome. The only fitted constant, the lossless compression ratio L=1.3, is explicitly disclosed as a preliminary setting and is checked against measured real rates (Table 4), so it is not a hidden prediction. Self-citations (GGSC, point-cloud quality metrics) serve as prior-work baselines and metrics, and no load-bearing argument is justified solely by a self-citation chain. Consequently, no circular step can be exhibited and the correct finding is no significant circularity.

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

The method carries no new physics or invented entities. The load-bearing assumptions are domain assumptions about 3DGS surjectivity and feature compressibility, plus hand-chosen hyperparameters. The rate-control prediction depends on an empirical compression-ratio estimate L=1.3.

free parameters (6)

Latent feature dimensions kc, kr = kc=3 or 6, kr=2
Chosen by hand; controls rate-quality tradeoff for color and rotation features (Section 4.1).
Bit depth BD = 16 for position and all attributes
Set to 16 based on prior quantization analysis (Zaghetto et al.); Table 5 sweeps 12-16.
RQ regularization lambda = 1e-2
Regularization factor in de-quantization ridge regression (Section 3.1.1).
Lossless compression ratio estimate L = 1.3
Used in rate control equations (8)-(9), estimated from preliminary experiments; Table 4 shows rate errors due to density-dependence.
Training schedule (T, Td, Tp, Tu, intervals) = T=70000, Td=15000, Tp=36000, Tu=66000, prune interval 2500, prune 0.1%/step
Hand-chosen schedule for progressive pruning and uniqueness; affects rate and quality.
Scaling and scaling-lr adjustment = 0.2*log(k)*scaling_lr(t1)
Heuristic learning-rate scaling after scene rescaling in LQM (Section A.1).

assumptions (6)

domain assumption 3DGS is surjective: different primitive sets can render perceptually close views.
Used to justify generating a compact 3DGS as a proxy for the original; stated in Section 1 from MPEG consensus.
domain assumption Color and rotation features are low-rank compressible; scaling is not.
PCA on two scenes (bicycle, dance) in Section A.2 guides the choice to use latent codes for color and rotation only.
ad hoc to paper GPCC lossless compression ratio is roughly stable for a given point density (L=1.3 for 3DGS).
Rate control uses this to predict bitrate; Section 3.2.2 notes densities affect L and reports errors.
domain assumption Integer 3DGS positions can be rendered without dequantization by correspondingly scaling and translating the camera.
LQM in Section 3.1.2 and A.3 relies on scene and camera co-transform preserving rendered images.
standard math Balanced ternary decomposition via basis vector [2^(N-2),...,1] and coding vector in {-1,0,1} can represent any integer position in range.
Used in primitive position decomposition (Section A.4); a standard positional representation, unproved in paper.
standard math Straight-through estimator (STE) and robust quantizer (RQ) provide usable gradients for quantization-aware training.
Lifted from Ye et al. 2024 and Bengio et al. 2013; relied on for training stability.

how reviews work

0 comments

Cite this review

Pith. "Pith review of HybridGS: High-Efficiency Gaussian Splatting Data Compression using Dual-Channel Sparse Representation and Point Cloud Encoder." pith.science (2026). https://pith.science/paper/EOT2UHS6

@misc{pith2026250501938,
  author       = {Pith},
  title        = {Pith review of: HybridGS: High-Efficiency Gaussian Splatting Data Compression using Dual-Channel Sparse Representation and Point Cloud Encoder},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EOT2UHS6}},
  note         = {Machine review of arXiv:2505.01938}
}

read the original abstract

Most existing 3D Gaussian Splatting (3DGS) compression schemes focus on producing compact 3DGS representation via implicit data embedding. They have long coding times and highly customized data format, making it difficult for widespread deployment. This paper presents a new 3DGS compression framework called HybridGS, which takes advantage of both compact generation and standardized point cloud data encoding. HybridGS first generates compact and explicit 3DGS data. A dual-channel sparse representation is introduced to supervise the primitive position and feature bit depth. It then utilizes a canonical point cloud encoder to perform further data compression and form standard output bitstreams. A simple and effective rate control scheme is proposed to pivot the interpretable data compression scheme. At the current stage, HybridGS does not include any modules aimed at improving 3DGS quality during generation. But experiment results show that it still provides comparable reconstruction performance against state-of-the-art methods, with evidently higher encoding and decoding speed. The code is publicly available at https://github.com/Qi-Yangsjtu/HybridGS.

Figures

Figures reproduced from arXiv: 2505.01938 by the authors.

**Figure 1.** Existing generative compression frameworks and the proposed HybridGS. However, due to the use of the explicit data format and primitive densification strategy, 3DGS has a huge data volume, which is challenging for storage and transmission. 3DGS compression has attracted considerable attention from both industry and academia, which is also the focus of this paper. In the recent 148-th Moving Picture Expert Group (MP… view at source ↗

**Figure 2.** Framework of HybridGS. Current (DC) components, 45 channels for color Spherical Harmonic (SH) coefficients, 1 channel for opacity, 3 channels for scaling, and 4 channels for rotation. Excluding opacity, PCA can be used to partition the remaining 55 features into compressible and compression-vulnerable ones. The concentration of the feature variance in a limited number of principal vectors suggests that a certain f… view at source ↗

**Figure 3.** Robust Quantization of 3DGS features. For the latent representations of color and rotation, opacity, and scaling, we initialize a quantizer for each attribute. Using RQ as an example, the training process is shown in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (12 more)

**Figure 4.** Figure 4: Scheme of LQM. For uniqueness and pruning, a na¨ıve approach is to adopt one time node after 3DGS converges to a stable good performance, where uniqueness and pruning are performed, and then fix the primitive position until the optimization of other attributes pivotin…

**Figure 5.** Figure 5: RD curve. 4.2.2. CODING TIME One of the advantages of HybridGS is faster encoding and decoding. We use three SOTA methods, i.e., HAC (λ = 0.0005), CompGS(MM) (λ = 0.001) and HGSC, as benchmark techniques. LR and HR coding times of HybridGS with kc = 3, kr = 2 are test…

**Figure 6.** Figure 6: PCA results of “bicycle” and “dance”. We use data from “dance” to illustrate the difference between using PCA and the proposed learnable low-dimensional latent features with lightweight trainable decoder. After obtaining vanilla 3DGS samples, we perform dimensionality …

**Figure 7.** Figure 7: 3DGS translation and scaling. • 3DGS translation: for a 3DGS G ′ with m primitives and the ith primitive position being pci = [xi , yi , zi ]. The Bbox center of G ′ is C = 1 2 [(xi) i max + (xi) i min, (yi) i max + (yi) i min, (zi) i max + (zi) i min] = [δX, δY, δZ]. …

**Figure 8.** Figure 8: Influence of outlier removal. To take full advantage of finite space, an outlier removal is applied before initializing the point cloud to a 3DGS G, i.e., G = I{O(P)}, O(·) is the outlier removal algorithm and I{·} is the initialization function mapping point cloud to …

**Figure 9.** Figure 9: Ablation study on outlier removal and 3DGS translation. i.e., R(pci ) = [< e i x , t i x >, < e i y , t i y >, < e i z , t i z >]. Considering that the primitive positions share the same BD and the same basis vector, using x coordinate as an example, the primitive posi…

**Figure 10.** Figure 10: Iteration vs. PSNR curves of HybridGS with pruning. 16 [PITH_FULL_IMAGE:figures/full_fig_p016_10.png]

**Figure 11.** Figure 11: PSNR vs. bitrate curves of GPCC and PCGC [PITH_FULL_IMAGE:figures/full_fig_p018_11.png]

**Figure 12.** Figure 12: Overall RD curve. Dataset Tank&Temple Deep Blending MipNeRF360 Method PSNR SIZE FPS PSNR SIZE FPS PSNR SIZE FPS 3DGS-30K 23.14 411.00 154 29.41 676.00 137 27.21 734.00 134 HybridGS kc=3, kr=2 HR 22.90 8.85 207 28.51 11.52 201 25.64 15.82 199 LR 22.66 4.27 247 28.32 5.…

**Figure 13.** Figure 13: Snapshots of HybridGS (kc = 3, kr = 2) on Tanks&Temples. Scene Method Training PSNR Testing PSNR Size (MB) playroom GS 37.63 30.03 550.67 HybridGS kc = 3, kr = 2 HR 33.52 29.89 12.15 LR 32.40 29.49 5.88 HybridGS kc = 6, kr = 2 HR 33.81 29.89 16.08 LR 33.11 29.68 7.79 …

**Figure 14.** Figure 14: Snapshots of HybridGS (kc = 3, kr = 2) on Deep Blending. 19 [PITH_FULL_IMAGE:figures/full_fig_p019_14.png]

**Figure 15.** Figure 15: Snapshots of HybridGS (kc = 3, kr = 2) on MipNeRF360. 21 [PITH_FULL_IMAGE:figures/full_fig_p021_15.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

3DGS-VBench: A Comprehensive Video Quality Evaluation Benchmark for 3DGS Compression
eess.IV 2025-08 unverdicted novelty 7.0 of 10

3DGS-VBench is a benchmark of 660 human-rated compressed 3D Gaussian Splatting models across 6 algorithms, with 15 quality metrics evaluated, for training 3DGS video quality assessment models.
$\mathcal{P}^3$: Toward Versatile Embodied Agents
cs.RO 2025-08 unverdicted novelty 4.0 of 10

P^3 combines real-time perception, feedback-free tool use, and priority-based dynamic scheduling into a unified framework for embodied agents.

Reference graph

Works this paper leans on

12 extracted references · 2 canonical work pages · cited by 2 Pith papers

[2]

It indicates that there is an inconsistency in the trends of reconstruction quality between training views and testing views

for the low rate of “bicycle”, the testing PSNR reports a stable value, while training PSNR exhibits increased variability. It indicates that there is an inconsistency in the trends of reconstruction quality between training views and testing views. Considering that the final quality of experience is influenced by both training and testing views, we recom...

work page
[7]

Yang, Q., Zhang, Y ., Chen, S., Xu, Y ., Sun, J., and Ma, Z

doi: 10.1109/TPAMI.2020.3047083. Yang, Q., Zhang, Y ., Chen, S., Xu, Y ., Sun, J., and Ma, Z. Mped: Quantifying point cloud distortion based on multiscale potential energy discrepancy.IEEE Trans. Pattern Analysis and Machine Intelligence, 45(5):6037– 6054,

arXiv 2020
[10]

Open3D: A modern library for 3D data processing.arXiv:1801.09847,

Zhou, Q.-Y ., Park, J., and Koltun, V . Open3D: A modern library for 3D data processing.arXiv:1801.09847,

arXiv
[11]

remove statistical outlier

doi: 10.1109/TMM.2024.3407698. 11 Title Suppressed Due to Excessive Size A. Appendix A.1. Model Parameters For LQM, before generating the na¨ıve 3DGS, we first conduct outlier removal to clean the output of COLMAP, facilitating the following 3DGS translation and scaling (see Section A.3). We use the function “remove statistical outlier” from Open3D (Zhou et al.,

arXiv 2024
[2001]

Fast feedforward 3d gaussian splatting compression

Chen, Y ., Wu, Q., Li, M., Lin, W., Harandi, M., and Cai, J. Fast feedforward 3d gaussian splatting compression. arXiv preprint arXiv:2410.08017, 2024a. Chen, Y ., Wu, Q., Lin, W., Harandi, M., and Cai, J. Hac: Hash-grid assisted context for 3d gaussian splatting com- pression. InEuropean Conference on Computer Vision, pp. 422–438. Springer, 2024b. Choy, ...

arXiv
[2016]

Fan, Z., Wang, K., Wen, K., Zhu, Z., Xu, D., Wang, Z., et al

doi: 10.1109/TIP.2016.2575005. Fan, Z., Wang, K., Wen, K., Zhu, Z., Xu, D., Wang, Z., et al. Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps.Advances in neural informa- tion processing systems, 37:140138–140158,

work page arXiv 2016
[2019]

Sullivan, G

doi: 10.1109/JETCAS.2018.2885981. Sullivan, G. J., Ohm, J.-R., Han, W.-J., and Wiegand, T. Overview of the high efficiency video coding (hevc) stan- dard.IEEE Trans. Circuits and Systems for Video Tech- nology, 22(12):1649–1668,

arXiv 2018
[2020]

Navaneet, K., Pourahmadi Meibodi, K., Abbasi Kooh- payegani, S., and Pirsiavash, H

doi: 10.1109/QoMEX48832.2020.9123147. Navaneet, K., Pourahmadi Meibodi, K., Abbasi Kooh- payegani, S., and Pirsiavash, H. Compgs: Smaller and 10 Title Suppressed Due to Excessive Size faster gaussian splatting with vector quantization. InEu- ropean Conference on Computer Vision, pp. 330–349. Springer,

arXiv 2020

Show all 12 references

[2021]

Wang, J., Ding, D., Li, Z., Feng, X., Cao, C., and Ma, Z

doi: 10.1109/DCC50243.2021.00015. Wang, J., Ding, D., Li, Z., Feng, X., Cao, C., and Ma, Z. Sparse tensor-based multiscale representation for point cloud geometry compression.IEEE Trans. Pattern Anal- ysis and Machine Intelligence, 45(7):9055–9071,

2021
[2022]

Estimating or propagating gradients through stochastic neurons for con- ditional computation.arXiv preprint arXiv:1308.3432,

Bengio, Y ., L´eonard, N., and Courville, A. Estimating or propagating gradients through stochastic neurons for con- ditional computation.arXiv preprint arXiv:1308.3432,

arXiv
[2023]

Yang, Q., Yang, K., Xing, Y ., Xu, Y ., and Li, Z

doi: 10.1109/TPAMI.2022.3213831. Yang, Q., Yang, K., Xing, Y ., Xu, Y ., and Li, Z. A bench- mark for gaussian splatting compression and quality as- sessment study. InProc. ACM Int. Conf. Multimedia in Asia. Association for Computing Machinery,

2022
[2024]

Ye, C., Chu, G., Liu, Y ., Zhang, Y ., Lew, L., and Howard, A

doi: 10.1145/3696409.3700172. Ye, C., Chu, G., Liu, Y ., Zhang, Y ., Lew, L., and Howard, A. Robust training of neural networks at arbitrary precision and sparsity.arXiv preprint arXiv:2409.09245,

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [2]

It indicates that there is an inconsistency in the trends of reconstruction quality between training views and testing views

for the low rate of “bicycle”, the testing PSNR reports a stable value, while training PSNR exhibits increased variability. It indicates that there is an inconsistency in the trends of reconstruction quality between training views and testing views. Considering that the final quality of experience is influenced by both training and testing views, we recom...

work page

[2] [7]

Yang, Q., Zhang, Y ., Chen, S., Xu, Y ., Sun, J., and Ma, Z

doi: 10.1109/TPAMI.2020.3047083. Yang, Q., Zhang, Y ., Chen, S., Xu, Y ., Sun, J., and Ma, Z. Mped: Quantifying point cloud distortion based on multiscale potential energy discrepancy.IEEE Trans. Pattern Analysis and Machine Intelligence, 45(5):6037– 6054,

arXiv 2020

[3] [10]

Open3D: A modern library for 3D data processing.arXiv:1801.09847,

Zhou, Q.-Y ., Park, J., and Koltun, V . Open3D: A modern library for 3D data processing.arXiv:1801.09847,

arXiv

[4] [11]

remove statistical outlier

doi: 10.1109/TMM.2024.3407698. 11 Title Suppressed Due to Excessive Size A. Appendix A.1. Model Parameters For LQM, before generating the na¨ıve 3DGS, we first conduct outlier removal to clean the output of COLMAP, facilitating the following 3DGS translation and scaling (see Section A.3). We use the function “remove statistical outlier” from Open3D (Zhou et al.,

arXiv 2024

[5] [2001]

Fast feedforward 3d gaussian splatting compression

Chen, Y ., Wu, Q., Li, M., Lin, W., Harandi, M., and Cai, J. Fast feedforward 3d gaussian splatting compression. arXiv preprint arXiv:2410.08017, 2024a. Chen, Y ., Wu, Q., Lin, W., Harandi, M., and Cai, J. Hac: Hash-grid assisted context for 3d gaussian splatting com- pression. InEuropean Conference on Computer Vision, pp. 422–438. Springer, 2024b. Choy, ...

arXiv

[6] [2016]

Fan, Z., Wang, K., Wen, K., Zhu, Z., Xu, D., Wang, Z., et al

doi: 10.1109/TIP.2016.2575005. Fan, Z., Wang, K., Wen, K., Zhu, Z., Xu, D., Wang, Z., et al. Lightgaussian: Unbounded 3d gaussian compression with 15x reduction and 200+ fps.Advances in neural informa- tion processing systems, 37:140138–140158,

work page arXiv 2016

[7] [2019]

Sullivan, G

doi: 10.1109/JETCAS.2018.2885981. Sullivan, G. J., Ohm, J.-R., Han, W.-J., and Wiegand, T. Overview of the high efficiency video coding (hevc) stan- dard.IEEE Trans. Circuits and Systems for Video Tech- nology, 22(12):1649–1668,

arXiv 2018

[8] [2020]

Navaneet, K., Pourahmadi Meibodi, K., Abbasi Kooh- payegani, S., and Pirsiavash, H

doi: 10.1109/QoMEX48832.2020.9123147. Navaneet, K., Pourahmadi Meibodi, K., Abbasi Kooh- payegani, S., and Pirsiavash, H. Compgs: Smaller and 10 Title Suppressed Due to Excessive Size faster gaussian splatting with vector quantization. InEu- ropean Conference on Computer Vision, pp. 330–349. Springer,

arXiv 2020