- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper proposes ATSplat, a feed-forward 3D Gaussian Splatting framework that restores scene-adaptive capacity allocation, overcoming the limitations of existing pixel-aligned methods whose primitive counts and placements depend on input image grids rather than scene complexity. By initializing sparse 3D anchor tokens and progressively expanding them in under-reconstructed regions via an Adaptive Token Expansion (ATE) module supervised by rendering error reduction, ATSplat decouples primitive placement from input pixel grids. ATSplat achieves competitive rendering quality while using up to 61x fewer Gaussians than dense pixel-aligned methods on the RealEstate10K and DL3DV benchmarks.
Critical strengths
-
Scene-Adaptive Capacity Allocation in Feed-Forward 3DGS: The paper successfully transfers the "sparse-to-adaptive" principle of optimization-based 3DGS into a feed-forward paradigm by dynamically allocating Gaussians based on rendering error reduction rather than rigid pixel grids.
-
Compression and Performance Trade-off: ATSplat achieves equivalent or superior performance on RealEstate10K and DL3DV datasets with a dramatically smaller number of Gaussian primitives (up to 61x reduction).
Critical weaknesses
-
Marginal Performance Gain of ATE over Simple Heuristics (Table 6): As shown in Table 6, using simple heuristic selections like Random (28.70 PSNR) or Farthest Point Sampling (28.65 PSNR) already achieves surprisingly strong performance—outperforming many existing feed-forward baselines—while trailing only slightly behind the proposed error reduction supervision (29.23 PSNR). This raises concerns about whether the complex, learned ATE module provides a proportional benefit given its added training and inference overhead.
-
Limited Scalability to Dense Multi-View Inputs: The evaluation is restricted to relatively sparse view settings (up to 16 views). In dense multi-view scenarios (e.g., 32+ views), overlapping captures generate massive redundancy. Without an explicit token pruning mechanism to remove redundant tokens during decoding, progressive token expansion combined with dense inputs leads to severe memory and computational bottlenecks, aggravated by the quadratic complexity of global multi-view attention.
Questions for authors
-
Justification of ATE vs. Simple Heuristics: Given that random or FPS token selection already yields competitive PSNR, could the authors elaborate on the exact justification for the learned ATE module? Does the performance gap widen significantly under extremely tight Gaussian budgets (e.g., < 10K Gaussians)?
-
Handling Redundancy in Dense Multi-View Settings: How does ATSplat handle massive information redundancy in dense multi-view captures (e.g., 32+ views)? Since initial tokens scale linearly with input views and the framework relies solely on progressive expansion without token pruning, does scaling to dense views cause an unmanageable token explosion and memory bottleneck? Furthermore, how does ATSplat's scalability and efficiency compare against dense feed-forward baselines optimized for multi-view inputs, including Zpressor (Wang et al., NeurIPS 2026) and MVP (Kang et al., CVPR 2026)?
-
GPU Memory Footprint and Computational Complexity: While inference runtime and Gaussian counts are reported, peak GPU memory (VRAM) consumption is omitted. Could the authors provide VRAM footprint analyses under high-resolution and multi-view setups to evaluate real-world deployment feasibility?
-
Generalization of Threshold : Can the inference threshold (e.g., 0.1 vs. 0.01) be applied consistently across different datasets or new domains (indoor vs. outdoor), or does it need to be manually re-tuned based on scene complexity or scale?
8 comments
Response to Reviewer WRnw (2/3)
Scalability and Efficiency
[...] progressive token expansion combined with dense inputs leads to severe memory and computational bottlenecks, aggravated by the quadratic complexity of global multi-view attention. [...] does scaling to dense views cause an unmanageable token explosion and memory bottleneck? Furthermore, how does ATSplat's scalability and efficiency compare against dense feed-forward baselines optimized for multi-view inputs, including Zpressor and MVP?
ATSplat starts from sparse anchor tokens initialized from coarse patches, which correspond to patches in our high-resolution setups, and the encoder performs global multi-view attention on these coarse tokens. While the token count increases as selected tokens are expanded across stages, progressive expansion reduces decoder computation since the full set of tokens is processed only at the last block. We report the average number of tokens at each stage on DL3DV at using our 16-view model with 16 and 32 input views, along with the number of tokens for the and patches commonly used in the baselines.
| Views | Method | Initial | Stage 1 | Stage 2 | Stage 3 |
|---|---|---|---|---|---|
| 16 | patches | 130.6K | 130.6K | 130.6K | 130.6K |
| patches | 32.6K | 32.6K | 32.6K | 32.6K | |
| Ours () | 8.2K | 10.5K | 13.5K | 17.1K | |
| Ours () | 8.2K | 14.1K | 25.2K | 44.8K | |
| 32 | patches | 261.1K | 261.1K | 261.1K | 261.1K |
| patches | 65.3K | 65.3K | 65.3K | 65.3K | |
| Ours () | 16.3K | 21.2K | 27.5K | 35.4K | |
| Ours () | 16.3K | 28.0K | 49.9K | 89.1K |
As shown in the table, ATSplat maintains a moderate token count even with dense input views, thanks to its progressive expansion formulation starting from sparse anchor tokens. At , it processes fewer tokens at every stage than a model using dense patch tokens throughout the decoder.
While MVP focuses on efficient encoding of dense multi-view inputs and reconstructs faster than ATSplat, ATSplat focuses on scene-adaptive Gaussian allocation, using up to fewer Gaussians (Tab. 4). We also highlight that ATSplat is not limited to a specific multi-view encoder design and can incorporate efficient designs such as the hierarchical attention of MVP.
The revised manuscript discusses scalability in Sec. 5 and reports per-stage token counts in Appendix E.1.
Redundancy in Dense Views
In dense multi-view scenarios (e.g., 32+ views), overlapping captures generate massive redundancy. [...] How does ATSplat handle massive information redundancy in dense multi-view captures (e.g., 32+ views)?
We agree that redundancy becomes more apparent with dense multi-view inputs. While ATSplat does not explicitly prune redundant tokens during decoding, as noted in our limitations, such redundancy can be alleviated by a simple voxel-based subsampling strategy without retraining, since our initial anchor tokens are defined in 3D space. Specifically, after the sparse anchor initialization, we keep one anchor token in each voxel of size 0.05 and discard the rest, passing the remaining tokens to the decoder. We demonstrate this capability on DL3DV at under the ZPressor setup with 12, 24, and 36 input views, and compare with ZPressor.
| Views | Method | PSNR | SSIM | LPIPS | #GS |
|---|---|---|---|---|---|
| 12 | ZPressor | 24.28 | 0.819 | 0.146 | 688K |
| Ours | 29.98 | 0.915 | 0.094 | 133K | |
| Ours + voxel subsampling | 29.43 | 0.905 | 0.104 | 115K | |
| 24 | ZPressor | 24.24 | 0.818 | 0.147 | 688K |
| Ours | 30.48 | 0.923 | 0.089 | 256K | |
| Ours + voxel subsampling | 29.91 | 0.913 | 0.098 | 198K | |
| 36 | ZPressor | 23.87 | 0.813 | 0.150 | 688K |
| Ours | 29.89 | 0.915 | 0.101 | 369K | |
| Ours + voxel subsampling | 29.59 | 0.908 | 0.106 | 261K |
With more input views, voxel subsampling removes more anchor tokens (14.6%, 24.9%, and 32.0%), which reflects the growing redundancy, while PSNR stays within 0.6 dB of the model without subsampling. At 36 views, it reduces the number of anchor tokens from 16,128 to 10,962 (-32.0%) and the number of Gaussians from 369K to 261K (-29.3%) while showing the smallest PSNR drop of 0.30 dB. We also note that ZPressor compresses the inputs into a fixed budget of 6 anchor views regardless of the number of input views or scene complexity, resulting in a higher compression ratio with more input views. In contrast, ATSplat adaptively allocates Gaussians by selectively expanding tokens in under-reconstructed regions, offering dynamic Gaussian budgets based on scene complexity. The dense-view redundancy analysis is now included in Appendix E.3.
Response to Reviewer WRnw (3/3)
GPU Memory Footprint
While inference runtime and Gaussian counts are reported, peak GPU memory (VRAM) consumption is omitted. Could the authors provide VRAM footprint analyses under high-resolution and multi-view setups to evaluate real-world deployment feasibility?
To evaluate real-world deployment feasibility, we report peak GPU memory (GB) averaged over the test scenes on DL3DV at using our 16-view model with 16, 32, and 64 input views, and compare with the baselines.
| Method | 16 views | 32 views | 64 views |
|---|---|---|---|
| iLRM | 3.91 | 6.74 | 12.40 |
| MVP | 4.86 | 8.33 | 15.27 |
| ReSplat-init | 16.29 | 31.61 | 62.25 |
| Ours () | 4.12 | 6.88 | 12.61 |
| Ours () | 4.28 | 7.38 | 20.06 |
As shown in the table, ATSplat () maintains moderate GPU memory usage up to 64 input views, with peak memory comparable to or lower than that of dense pixel-aligned baselines. This memory efficiency originates from our sparse-to-adaptive formulation, which begins with a coarse scene scaffold and progressively expands and refines tokens in under-reconstructed regions. Peak GPU memory results are now reported in Appendix E.1.
Generalization of Threshold
Can the inference threshold (e.g., 0.1 vs. 0.01) be applied consistently across different datasets or new domains (indoor vs. outdoor), or does it need to be manually re-tuned based on scene complexity or scale?
The expansion score is supervised by the reduction in 2D rendering error from token expansion, making a threshold on the expected improvement from expansion rather than a dataset- or scene-specific hyperparameter. As shown in the experiments throughout the manuscript, generally works well without re-tuning across different datasets, numbers of input views, and resolutions. The threshold controls the trade-off between rendering quality and the number of Gaussians, and a lower threshold such as yields higher rendering quality at the cost of a larger Gaussian budget. The discussion of threshold generalization now appears in Appendix B.
Response to Reviewer WRnw (1/3)
We thank the reviewer for the thoughtful review. We are encouraged that the reviewer finds ATSplat successfully transfers the sparse-to-adaptive principle to feed-forward 3DGS while achieving equivalent or superior performance with far fewer Gaussians. We first clarify the benefit of learned token selection over simple heuristics, then address questions on scalability, efficiency, and generalization.
ATE vs. Simple Heuristics
Apart from the selection strategy, the Random and FPS variants share the rest of ATSplat's key components, including progressive expansion and refinement and the flexible 3D anchor design. As requested, we evaluate the selection strategies at a tighter budget of about 7K Gaussians on RealEstate10K with 2 input views at 256×256, which corresponds to τ=0.2 for our model.
Starting from 4K Gaussians without expansion, heuristic selection strategies improve PSNR by about 0.6 dB, while our learned selection roughly doubles the gain to 1.29 dB. While the heuristic selections also achieve competitive results thanks to these shared components, our learned expansion scores consistently outperform the heuristic variants, with the gap becoming larger under the tighter budget (about 0.7 dB higher PSNR). At about 7K Gaussians, our learned selection reaches 28.62 dB PSNR, close to the 28.70 and 28.65 dB achieved by Random and FPS with 13-15K Gaussians, showing more effective use of a limited Gaussian budget. We also note that our learned selection offers dynamic Gaussian budgets based on scene complexity, whereas heuristic selections can only produce a fixed number of Gaussians regardless of the complexity of each scene.
Overhead. We would like to emphasize that ATSplat predicts the expansion score using a lightweight MLP at each decoder block. The three MLPs have fewer than 1.8M parameters in total, and their inference time stays below 5 ms even at 540×960 resolution with 16 input views, which is negligible compared to the total reconstruction time of 0.243 s. During training, score supervision accounts for about 9% of the training step time.
We have included the tighter-budget results in Appendix D.5 and the overhead analysis in Appendix E.2.