- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This work proposes FS-Gstream to address flicker and background degradation in two-stage streaming Gaussian reconstruction. DAMM estimates a frame-specific threshold from a KDE of log-transformed Gaussian translation magnitudes, and MGO uses the resulting masks to regulate motion application, parameter optimization, densification and pruning. Dynamic geometry and appearance are refined, while static geometry is protected and selected static appearance parameters remain adaptable.
Critical strengths
- This work proposes a plug-and-play module, which can be applied to some sotas directly, making it widely applicable.
- Table 1 shows lower Ewarp and higher PSNR for all twelve backbone–dataset pairs. These improvements should be credited.
- The ablations experiments are substantial, supporting a genuine quality and stability contribution beyond changing initialization and low-level Gaussian operations.
Critical weaknesses
W1. Missing comparisons with the closest alternatives is my main concern.
Table 1 mainly compares three backbones with and without FS-GStream, alongside IGS. These results demonstrate useful gains, but do not establish an advantage over existing selective-update and anti-flicker methods. In particular, CPOStream (AAAI 2026) and Yun et al.'s observation-restoration method (SIGGRAPH 2025) address closely related objectives and are cited but not evaluated. Comparisons under matched protocols, including a shared backbone where applicable, are needed to assess reconstruction quality, temporal consistency and computational cost. The existing ablations cannot substitute for these comparisons.
W2. The reliability of the translation-based separation is not established in important boundary cases.
The paper already supplies an EMA fallback and acknowledges dependence on the backbone's motion estimates. What remains unclear is its cold-start behavior, fallback frequency and parameterization, degenerate-KDE handling, and motion fidelity in these regimes. More specifically, Section 3.3 assumes that the leftmost prominent low-translation mode represents static displacement noise and that a subsequent valley provides a meaningful boundary. Slow true motion overlapping noise, poorly separated distributions, and changes with negligible center translation are not directly examined. Eq. (5) gates rotation using translation alone, while Eq. (6) freezes geometry for static-classified primitives; an anisotropic Gaussian can require an orientation change without center displacement.
W3. Tables 3–4 compare fixed thresholds, but do not isolate DAMM's benefit over another adaptive selector under the same MGO. Runtime attribution is also unclear: Table 7 shows increased runtime when the module is added under identical settings, so Table 1's speedups cannot be attributed to DAMM/MGO alone.
Questions for authors
Q1. Under matched streaming protocols, how does FS-GStream compare with CPOStream and, on a shared compatible backbone, Yun et al.'s observation-restoration method in reconstruction quality, temporal consistency and computational cost?
Q2. Does translation-based masking preserve slow motion and rotation with negligible center displacement? Please provide a targeted example and clarify threshold initialization and EMA fallback behavior when no valid peak/valley is detected.
Q3. With MGO fixed, how do alternative adaptive or flow-based masks affect final reconstruction and temporal consistency, including selector cost? Please separate module overhead from FastGS implementation effects and assess longer-term stability and dynamic-region fidelity.
5 comments
Response to W2 and Q2 (Slow Motion, Rotation, and EMA Fallback)
Slow motion and rotation. DAMM uses the displacement of each Gaussian’s center. An object rotating around a fixed center can still move its Gaussian centers, allowing DAMM to detect the motion when these displacements are distinguishable from static noise. However, very slow motion or rotation expressed mainly through orientation changes with negligible Gaussian-center displacement can be missed. We will explicitly discuss this failure mode as a limitation.
Threshold initialization and EMA fallback. Before EMA initialization, if a valid peak is found but no valid valley is detected, we use , where is the selected peak’s left HWHM (half width at half maximum) in log space. If no left half-maximum crossing is found, the search stops at the lower KDE evaluation boundary. If no peak passes the prominence criterion before EMA initialization, we instead set and treat all Gaussians with nonzero displacement as dynamic. The factor of 2 is based on median valley offsets of 2.21/1.89/2.26 HWHM for ReCon-GS/HiCoM/3DGStream, with leave-one-dataset-out estimates of 1.75–2.50. Neither of these initialization branches was used in the evaluated runs.
The first valid valley initializes the EMA in displacement units, using . Each new valid threshold receives weight 0.9 and the previous EMA receives weight 0.1. Once initialized, frames without a valid valley reuse the stored EMA without updating it. Fallback rates for ReCon-GS, HiCoM, and 3DGStream were 11.5%, 0.7%, and 25.5%, with maximum consecutive durations of 13, 3, and 26 frames, respectively.
We acknowledge that these implementation details were not fully specified in the original manuscript. We will clarify the threshold initialization, EMA update rule, and fallback conditions, and include the fallback statistics in the revised manuscript.
Targeted evaluation of slow motion. We compare our method with the “Only MGO” variant (all-update) from our submitted ablation study, both using ReCon-GS. This all-update variant disables DAMM and optimizes every Gaussian. We report PSNR and LPIPS within the same SAM3 dynamic masks.
We evaluate two separate groupings. First, we separate frames with and without a valid valley using our method’s DAMM logs. Second, we rank each scene’s frames by the mean UniMatch flow magnitude between consecutive GT images within the dynamic mask, divide them into five equally sized groups, and report the slowest and fastest groups. Both methods use identical frame groups. Results are averaged within each scene and then across all 13 scenes with equal scene weights.
As shown below, the slowest group exhibits larger dynamic-region PSNR and LPIPS gaps from all-update than the fastest group. The gaps are also larger on frames without a valid valley than on frames with a valley. These results are consistent with the risk of suppressing genuine slow motion when its displacement overlaps static noise. We will explicitly discuss this limitation.
| Frames | Estimated motion in GT videos (px/frame) | Dynamic PSNR (ours / all-update) | Dynamic LPIPS (ours / all-update) |
|---|---|---|---|
| With a valid valley | 1.46 | 30.41 / 30.47 | 0.117 / 0.107 |
| Without a valid valley | 0.76 | 30.70 / 30.85 | 0.117 / 0.102 |
| Slowest fifth | 0.66 | 30.94 / 31.25 | 0.114 / 0.096 |
| Fastest fifth | 2.34 | 29.88 / 29.95 | 0.123 / 0.116 |
Response to W3 and Q3 (Alternative Masks, Runtime, and Longer-Sequence Evaluation)
We provide a controlled mask comparison with ReCon-GS and MGO fixed, together with separate runtime analysis and a 1,000-frame evaluation.
Alternative masks and selector cost. With ReCon-GS and MGO fixed, we compare DAMM with a UniMatch-based mask using a one-pixel flow threshold and a CPOStream-derived mask. DAMM outperforms the UniMatch-based mask in overall and dynamic-region PSNR, SSIM, and LPIPS on all four datasets. Temporal warping error is lower on SelfCap, the same on VRU and Meet Room, and slightly higher on N3DV.
The CPOStream-derived mask achieves better image quality in several settings but allows a much larger fraction of Gaussians to be optimized, leading to larger payloads. Its tendency toward better image quality but higher temporal warping error is similar to the “Only MGO” variant in our submitted ablation study, which allows all Gaussians to be optimized at each frame. DAMM achieves lower temporal warping error and smaller payloads than the CPOStream-derived mask on all four datasets. The DAMM-based pipeline also has the lowest selector cost and per-frame training time among the evaluated alternatives.
| Dataset | Method | E_warp ↓ | Dynamic PSNR ↑ | Dynamic SSIM ↑ | Dynamic LPIPS ↓ | Overall PSNR ↑ | Overall SSIM ↑ | Overall LPIPS ↓ | Dynamic ratio (%) | Payload (MB/f) ↓ | Train (sec) ↓ | Selector cost (s/frame) ↓ |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SelfCap | MGO + DAMM (Ours) | 0.014 | 25.17 | 0.798 | 0.204 | 25.37 | 0.837 | 0.244 | 13.55 | 6.44 | 4.1 | 0.049 |
| MGO + UniMatch (1-pixel threshold) | 0.015 | 24.50 | 0.777 | 0.210 | 25.34 | 0.834 | 0.247 | 18.93 | 7.72 | 9.3 | 5.635 | |
| MGO + CPOStream-derived mask | 0.032 | 24.45 | 0.815 | 0.184 | 24.62 | 0.836 | 0.239 | 98.14 | 41.29 | 8.6 | 4.593 | |
| VRU | MGO + DAMM (Ours) | 0.025 | 24.43 | 0.848 | 0.114 | 27.22 | 0.889 | 0.181 | 27.17 | 6.42 | 2.3 | 0.047 |
| MGO + UniMatch (1-pixel threshold) | 0.025 | 23.93 | 0.836 | 0.128 | 26.03 | 0.868 | 0.200 | 28.68 | 6.39 | 3.6 | 1.502 | |
| MGO + CPOStream-derived mask | 0.036 | 25.42 | 0.874 | 0.094 | 27.69 | 0.902 | 0.164 | 99.34 | 24.26 | 4.2 | 2.052 | |
| Meet Room | MGO + DAMM (Ours) | 0.006 | 32.24 | 0.942 | 0.077 | 32.51 | 0.963 | 0.161 | 15.28 | 1.51 | 2.1 | 0.045 |
| MGO + UniMatch (1-pixel threshold) | 0.006 | 26.67 | 0.837 | 0.179 | 30.89 | 0.951 | 0.176 | 7.29 | 1.24 | 3.1 | 1.011 | |
| MGO + CPOStream-derived mask | 0.009 | 32.38 | 0.944 | 0.075 | 32.52 | 0.963 | 0.162 | 57.33 | 2.89 | 2.6 | 0.562 | |
| N3DV | MGO + DAMM (Ours) | 0.008 | 33.25 | 0.944 | 0.110 | 31.83 | 0.947 | 0.133 | 5.52 | 2.64 | 3.2 | 0.048 |
| MGO + UniMatch (1-pixel threshold) | 0.007 | 29.76 | 0.883 | 0.165 | 31.05 | 0.939 | 0.141 | 5.26 | 2.65 | 6.6 | 3.537 | |
| MGO + CPOStream-derived mask | 0.013 | 33.75 | 0.950 | 0.102 | 32.06 | 0.950 | 0.130 | 75.71 | 8.32 | 4.5 | 1.410 |
Module overhead and implementation effects. We agree that the runtime gains in Table 1 should not be attributed to DAMM/MGO alone. With the implementation fixed, adding FS-GStream increases runtime in both Table 7 and our additional experiment with official ReCon-GS. The runtime gains in Table 1 therefore reflect the full FastGS-based system.
Longer-sequence evaluation and dynamic-region quality. We evaluate three configurations on the 1,000-frame Corgi sequence: (1) ReCon-GS trained using its official implementation, (2) the same official implementation with FS-GStream added, and (3) our FastGS-based ReCon-GS implementation with FS-GStream. Adding FS-GStream to official ReCon-GS improves overall and dynamic-region PSNR, SSIM, and LPIPS, while reducing temporal warping error and payload. Fig. 6 in Appendix A of the revised manuscript illustrates the corresponding per-frame PSNR trends. FS-GStream maintains higher reconstruction quality in later frames and avoids the pronounced PSNR drops observed in the baseline. These trends support improved reconstruction stability over the extended sequence.
| Method | Implementation | E_warp ↓ | Overall PSNR ↑ | Overall SSIM ↑ | Overall LPIPS ↓ | Train (s/frame) ↓ | Payload (MB/frame) ↓ | Dynamic PSNR ↑ | Dynamic SSIM ↑ | Dynamic LPIPS ↓ |
|---|---|---|---|---|---|---|---|---|---|---|
| ReCon-GS (baseline) | Official | 0.048 | 23.55 | 0.729 | 0.394 | 5.8 | 14.66 | 21.00 | 0.642 | 0.434 |
| ReCon-GS + FS-GStream | Official | 0.025 | 25.00 | 0.780 | 0.300 | 6.8 | 10.60 | 22.86 | 0.725 | 0.301 |
| ReCon-GS + FS-GStream | FastGS-based | 0.022 | 25.73 | 0.792 | 0.279 | 4.7 | 9.18 | 24.29 | 0.781 | 0.228 |
Thank you for your valuable review comments. We will address each of your concerns here. Below, we respond to each comment and will incorporate the corresponding clarifications and revisions into the final manuscript. If there is anything we haven't explained clearly, please don't hesitate to let us know.
Thank you for identifying these closely related methods. We provide direct comparisons with CPOStream and Yun et al.’s observation-restoration method. Since an official CPOStream implementation was unavailable when we conducted the experiments, we reimplemented it using the hyperparameters reported in its paper and the same initial Gaussian set as our method. As Yun et al. provide public code only for their 3DGStream-based implementation, we use their official code and restrict this comparison to 3DGStream as the shared backbone.
Comparison with CPOStream. We compare our CPOStream reproduction with ReCon-GS + FS-GStream. As shown in the table below, our method achieves better overall and dynamic-region PSNR, SSIM, and LPIPS, together with lower temporal warping error, than this reproduction on all four datasets.
Comparison with Yun et al. on 3DGStream. Using 3DGStream as the shared backbone, our method outperforms Yun et al. in all three dynamic-region quality metrics on every dataset. It also achieves lower temporal warping error on SelfCap, VRU, and N3DV, and the same value on Meet Room. Overall image quality is also better except for PSNR and SSIM on Meet Room. Our evaluated pipeline has lower training time and payload across all four datasets.