- Rating
- 3
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper proposes PointSTAR, a coarse-to-fine autoregressive framework for 3D point cloud generation that replaces discrete tokens with continuous latents to improve generation quality and efficiency.
Critical strengths
-
Comprehensive experimental validation The paper provides a fairly complete set of experiments, including comparisons with recent SOTA methods (PointNSP and Quartet, 2026), extensive ablation studies, reconstruction, generation, and downstream tasks like point cloud completion and upsampling. Most results consistently outperform the baselines, which makes the proposed approach reasonably convincing.
-
A simple but effective extension of next-scale autoregressive generation Replacing discrete tokens with continuous latents is a straightforward and well-motivated idea that avoids quantization errors. The paper also introduces a progressively decodable latent hierarchy and multi-directional autoregressive flows. Together, these design choices lead to consistent, though mostly modest, improvements in generation quality.
Critical weaknesses
Modest improvements over recent SOTA baselines.
Despite the extensive experiments, the improvements over PointNSP are generally small, often around 1–3 percentage points in generation metrics. The comparison with recent baselines is also incomplete, as several methods lack MMD and COV results. Given the additional complexity of the continuous latent modeling framework, the paper does not convincingly demonstrate a substantial enough improvement to justify its methodological contribution.
Questions for authors
You acknowledged that runtime comparisons are not fully standardized, but provide no similar justification for the missing generation metrics: do not explicitly explain why MMD and COV are missing for several recent baselines.
Is there any specific reason?
2 comments
W2 / Q1. Why are MMD and COV missing for several baselines? Those methods (e.g., PointNSP, Quartet, FrePoLad, PSF) report only 1-NNA in their papers, and we copied published numbers without estimating the missing metrics ourselves. 1-NNA was preferred because, as discussed in PointFlow (Yang et al., 2019), MMD is insensitive to low-quality samples and COV to sample quality. 1-NNA reflects both fidelity and diversity. Following your suggestion, we evaluated TIGER, FrePoLad and PointNSP ourselves with the same reference sets, normalization and evaluation code as Table 3. No pretrained checkpoints are released for these methods, so we retrained each of them on the same training split with its public code and the code's default hyperparameters, without additional tuning. For FrePoLad, the defaults in the released code differ from the settings described in the paper, so its MMD/COV may be somewhat pessimistic. We have noted this in the revision (Appendix E.1). We then evaluated the samples with the same reference sets, normalization and evaluation code as Table 3. For consistency with Table 3, the 1-NNA column in Table R2 reports the values published in the original papers, MMD and COV come from our retrained models.
Table R2. Completed metrics. MMD-CD ×10³, MMD-EMD ×10², COV and 1-NNA in %. Each cell is CD / EMD.
| Class | Method | MMD ↓ | COV ↑ | 1-NNA (→50)† |
|---|---|---|---|---|
| Airplane | TIGER | 0.225 / 0.361 | 46.17 / 51.11 | 67.21 / 56.26 |
| FrePoLad | 0.319 / 0.597 | 46.91 / 44.26 | 65.25 / 62.10 | |
| PointNSP | 0.285 / 0.422 | 46.45 / 47.76 | 66.98 / 56.05 | |
| PointSTAR | 0.209 / 0.338 | 50.86 / 53.57 | 65.19 / 54.44 | |
| Chair | TIGER | 2.577 / 1.545 | 46.37 / 51.81 | 54.32 / 51.71 |
| FrePoLad | 3.734 / 3.463 | 42.57 / 41.63 | 52.35 / 53.23 | |
| PointNSP | 3.774 / 1.852 | 47.19 / 46.18 | 54.01 / 53.76 | |
| PointSTAR | 2.329 / 1.478 | 50.45 / 53.17 | 52.45 / 51.06 | |
| Car | TIGER | 1.142 / 0.783 | 46.88 / 50.85 | 54.12 / 50.24 |
| FrePoLad | 2.931 / 1.334 | 45.56 / 46.27 | 51.89 / 50.26 | |
| PointNSP | 1.451 / 1.217 | 46.42 / 48.64 | 53.12 / 50.09 | |
| PointSTAR | 0.873 / 0.744 | 51.14 / 53.13 | 50.57 / 50.99 |
† For baselines, 1-NNA values are taken from the original papers, as in Table 3. MMD and COV for TIGER, FrePoLad and PointNSP are computed from our retrained models.
We can see that PointSTAR obtains the best MMD and COV on all 12 entries against these three baselines. Against PointNSP specifically, MMD-CD is 27–40% lower and COV is 3.3–7.0 points higher on every category. Taking the full Table 3 into account, PointSTAR has the best MMD and COV in 11 of 12 entries. The exception is Car COV-EMD, where LION reaches 56.53 vs. our 53.13. Note also that on Car, where PointNSP has the better 1-NNA-EMD (50.09 vs. 50.99), PointSTAR is still clearly better on MMD-EMD (0.744 vs. 1.217) and COV-EMD (53.13 vs. 48.64). Quartet, PSF and not-so-ot-flow also release no checkpoints. Retraining them was not feasible within the rebuttal period, and we will add their MMD/COV in the revision. We have added these rows to Table 3 of the revised paper, marked with † as evaluated by us.
Response to Reviewer 8V6J
Response to Reviewer 8V6J
We thank the reviewer for the positive evaluation of the experiments and the idea.
W1. Modest improvements over PointNSP. While the 1-NNA improvements over PointNSP on Airplane, Chair, and Car are relatively modest (1.5–2.7 points), we highlight three aspects that provide a more comprehensive assessment:
PointNSP retains a slight advantage on Car EMD.
The extra components are also not costly in practice. PointSTAR samples in 1.37 s end-to-end, compared with 3.59 s for PointNSP and 5.45 s for Quartet as reported, and trains in 46 GPU hours including NAVE.