- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
The paper proposes KC-3DGS, which augments standard 3D Gaussian Splatting with three wavelet-domain losses. The first aligns rendered and ground-truth wavelet coefficients with scale-dependent weights. The second matches supervised per-band kurtosis statistics. The third matches supervised cross-band covariance statistics. A three-level Daubechies-3 transform is applied to rendered and ground-truth images. Pixel-aligned wavelet alignment is used on training views, while the statistical losses are applied to renders from translated pseudo-cameras after densification. The Gaussian representation and rasterizer are unchanged. The paper argues that these losses improve the distribution of image structure across scales and orientations, reducing artifacts such as floaters, color casts, and oversmoothing.
Critical strengths
-
The paper addresses a relevant sparse-view failure mode with a multi-scale and orientation-aware representation. Wavelet coefficients can expose frequency-specific errors that are not visible from a single aggregate pixel-space objective, making the motivation for the regularizer meaningful.
-
The method is modular and preserves the underlying 3DGS representation and rasterizer. The separation between training-view coefficient alignment and pseudo-view statistical supervision gives the method a clear structure and makes its intended integration point easy to understand.
Critical weaknesses
-
The theoretical justification for the L1 refinement claim would benefit from clarification. Orthogonality directly guarantees L2 isometry, whereas the stated lemma concerns L1 norms; the argument provided does not appear to establish the general inequality and equality condition as written. The discussion also treats boundary effects as negligible, while the implementation uses zero padding.
-
Kurtosis matching captures one aspect of the subband distributions, but does not by itself fully characterize image structure. Since it constrains standardized fourth-order statistics rather than band energy, phase, edge location, or spatial arrangement, images with different structures may still have similar kurtosis values.
-
The supervised cross-band covariance loss matches only a scalar Frobenius norm. It loses subband-pair identities and correlation signs, so different orientation-specific covariance structures can have the same loss.
-
The pseudo-view target may be inconsistent with the rendered viewpoint. A translated pseudo-camera uses statistics from the anchor-view ground truth, but the paper does not establish stability under parallax, occlusion, disocclusion, or view-dependent appearance.
-
All three losses start after densification has ended, so they cannot add Gaussian primitives to represent missing regions. They can still adjust the existing set, but their capacity to recover structures that are wholly absent is limited.
-
The plug-and-play claim is restricted by color parameterization. Shared color networks couple many Gaussians to one pseudo-view gradient, and the reported DNGaussian result shows that the losses do not transfer directly.
-
The supplementary material has several visibly underfilled pages and uneven whitespace around figures and tables, which makes its layout look less polished. This is a minor presentation issue.
Questions for authors
-
The key lemma claims a general L1 inequality for any orthogonal transform. Since orthogonality guarantees L2 isometry, can the authors provide a valid L1 derivation or revise the lemma for the implemented zero-padded db3 transform?
-
What justifies using anchor-view ground-truth statistics for a translated pseudo-camera under parallax, occlusion, disocclusion, or view-dependent appearance?
-
Since CBC matches only a scalar covariance norm, is the intended claim limited to aggregate dependence, or should it preserve individual orientation-specific correlations?
7 comments
W4 / Q2. What justifies using anchor-view ground-truth statistics for a translated pseudo-camera?
We measured how far the anchor view's ground-truth statistics are from the pseudo view's true statistics. To do this, we trained a 3DGS model on all views to estimate the pseudo view's image, and checked the trained model's statistics against real held-out views where true ground truth exists.
The anchor target is approximate, and the trained model's statistics at the pseudo poses stay near the anchor's. In the held-out real views test, the trained model's statistics are closer to the true ground truth than 3DGS's. Varying the pseudo-camera offset from 0.1 to 0.4 grows the estimated target error 1.983x (kurtosis) and 1.309x (cross-band):
| Setting | scenes | ΔPSNR (wins) | ΔSSIM (wins) | ΔLPIPS (wins) | ΔDreamSim (wins) |
|---|---|---|---|---|---|
| KC-3DGS (jitter 0.2) | 9 | +0.23 (8/9) | +0.0128 (9/9) | -0.0076 (9/9) | -0.0104 (8/9) |
| KC-3DGS, jitter 0.1 | 9 | +0.20 (8/9) | +0.0127 (9/9) | -0.0075 (9/9) | -0.0089 (8/9) |
| KC-3DGS, jitter 0.4 | 9 | +0.26 (8/9) | +0.0131 (9/9) | -0.0069 (9/9) | -0.0083 (7/9) |
These measurements do not separate parallax, occlusion, or view-dependent appearance. The target is an approximation that is robust in practice.
W2 / W3 / Q3. Kurtosis and CBC are not full structure descriptors. Is the intended claim limited to aggregate dependence?
Location, phase, edge position and energy are carried by L_wav, which aligns coefficients pixel-wise on training views. L_KC and L_CBC constrain only aggregate distribution shape and aggregate cross-band dependence. The CBC claim is restricted to aggregate dependence. We do not claim it preserves orientation-specific correlations or their signs.
W1 / Q1. The L1 lemma and the zero-padded db3 transform
We thank the reviewer for this question; the concern is correct. Orthogonality guarantees only an L2 isometry, and no L1 ordering holds in either direction. For db3 with J=3, a detail basis function has wavelet L1=1 and pixel L1 between 3.4 and 14.1, which violates the stated inequality. A single-pixel error reverses the ordering: wavelet L1=4.2 and pixel L1=1. The equality condition fails for the same reason.
We have therefore taken the second option and revised the lemma in L2 for the implemented transform.
Zero-padded db3. With zero extension, the rendered error Δ is a finitely supported signal on Z2, and our transform computes its orthonormal DWT while keeping every nonzero coefficient. Hence W⊤W=I holds exactly, without neglecting boundaries. W is an isometry onto its range but is not surjective, since it produces more coefficients than pixels. We verified this numerically: the relative Parseval error is shown in Table below. The revised appendix states this explicitly in place of the earlier "negligible boundary" assumption.
The relative Parseval error ∥WΔ∥22−∥Δ∥22/∥Δ∥22 is measured at each dataset's training resolution. This residual comes from storing the db3 filter coefficients in single precision; with double-precision coefficients it falls to 3×10−16. This confirms that zero extension keeps all coefficients and W⊤W=I holds exactly.
Revised Lemma 1 (and Lemma 3 in Appendix B). Let e(Δ) collect the squared L2 norms of all 3J+1 subbands, including the approximation band. Then:
The proof uses only W⊤W=I, and its counterexamples lie in the image interior, so it holds exactly for the implemented zero-padded db3 transform. The lemma states what subband supervision adds: it distinguishes errors that pixel L2 treats identically. We have removed all claims about L1 ordering and "strictly stronger" supervision, and the revised appendix includes the L1 counterexamples above.
These corrections do not affect our experimental conclusions. The ablations (Tables 4 and 10) show that the perceptual gain comes mainly from the pseudo-view kurtosis and covariance terms, not from the wavelet term this lemma concerns.