- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 3
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
The paper studies temporal artifacts in a one-step DMD video super-resolution student. It analyzes how a 50-step teacher's attention becomes more localized along motion correspondences at low noise, then proposes TRD-VSR: low-noise correction distillation plus relational structure alignment to HR-video feature relations. The extra teacher and HR supervision are training-only, preserving one-step inference. Component ablations on UDM10, SPMCS, and YouHQ40 show improved tOF and Ewarp over the starting student, and the method obtains the best tLPIPS and DOVER among listed methods on all three benchmarks.
Critical strengths
- The method keeps one denoiser evaluation at inference and uses a matched one-step starting student for LNCD and RSA ablations. On UDM10, their combination reduces tOF from 1.384 to 0.870 and Ewarp from 0.877 to 0.821, with the same trends appearing on the other two main benchmarks.
- The authors look beyond a single temporal score: tOF, tLPIPS, tLP and Ewarp are considered, and the low-noise attention intervention gives some diagnostic support. The additional MVSR4x benchmark and measured inference time/memory are useful checks.
Critical weaknesses
- The attention-based explanation remains suggestive rather than decisive. Figure 2 shows an association between denoising step and attention locality, Appendix B.2 tests only 10 UDM10 clips and blocks neighborhoods across all 30 teacher layers. Blocking within-frame neighborhoods also harms temporal metrics. A matched intervention at high noise or a less disruptive control is needed to isolate late cross-frame refinement as the cause.
- The perceptual/temporal gains come with notable reference-fidelity tradeoffs: UDM10 PSNR is 22.91 versus DOVE's 25.36 and YouHQ40 PSNR is 21.24 versus DOVE's 23.21. The claim of generally better temporal consistency needs this mixed evidence stated more prominently.
Questions for authors
- Can the authors compare low-noise correction to an equal-compute high-noise correction and to continued training with the same extra 5K updates, to isolate the benefit of the proposed mechanism?
- What explains the much worse MVSR4x Ewarp despite improved tLPIPS, and do human observers prefer the resulting motion consistency under real degradations?
5 comments
Response to Reviewer ye1G [Critical Weaknesses 2]
Critical Weaknesses 2: Reference fidelity and temporal consistency
Reviewer concern: The perceptual/temporal gains come with notable reference-fidelity tradeoffs.
Fidelity and temporal consistency.
Table 2 shows that our temporal gains hold at comparable reference fidelity. FlashVSR [1] is the closest method to ours in PSNR: the two methods differ by at most 0.15 dB on each of the three datasets. Our method achieves higher SSIM on all three and performs better in eight of the nine temporal comparisons. These results show that we can improve temporal consistency while maintaining similar fidelity.
Why our PSNR and SSIM are lower than those of some methods.
The comparison with FlashVSR shows that our temporal gains do not require a substantial loss of fidelity. The lower PSNR and SSIM relative to some other methods are consistent with the established perception–distortion trade-off [2]. VSR is an ill-posed inverse problem with multiple plausible solutions. Minimizing mean squared error to maximize PSNR tends to favor smooth outputs that average over plausible reconstructions. Distortion metrics can therefore penalize visually plausible details that differ from the particular ground-truth reference [3]. Lower PSNR or SSIM thus need not imply lower perceptual quality. Like Upscale-A-Video, STAR, DLoRAL, and FlashVSR [1, 4–6], we aim to recover realistic, detailed videos. Our work further addresses the temporal instability of those generated details. Table 2 illustrates this distinction: DOVE [7], SeedVR2 [8], and PS-SR [9] achieve higher PSNR than our method, while our method scores higher on the no-reference perceptual quality metrics.
Overall results.
In Table 2, our method ranks first in 7 of the 9 temporal comparisons and 8 of the 9 no-reference perceptual quality comparisons, while maintaining reference fidelity comparable to FlashVSR. These results support our central contribution of improving both perceptual quality and temporal consistency at competitive reference fidelity.
References
[1] Zhuang et al. FlashVSR: Towards Real-time Diffusion-Based Streaming Video Super Resolution. CVPR, 2026.
[2] Blau and Michaeli. The Perception-Distortion Tradeoff. CVPR, 2018.
[3] Ledig et al. Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network. CVPR, 2017.
[4] Zhou et al. Upscale-A-Video: Temporal-Consistent Diffusion Model for Real-World Video Super-Resolution. CVPR, 2024.
[5] Xie et al. STAR: Spatial-Temporal Augmentation with Text-to-Video Models for Real-World Video Super-Resolution. ICCV, 2025.
[6] Sun et al. One-Step Diffusion for Detail-Rich and Temporally Consistent Video Super-Resolution. NeurIPS, 2025.
[7] Chen et al. DOVE: Efficient One-Step Diffusion Model for Real-World Video Super-Resolution. NeurIPS, 2025.
[8] Wang et al. SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training. ICLR, 2026.
[9] Wu et al. PS-SR: Pseudo-Single-Step Video Super-Resolution via Speculative Diffusion. CVPR, 2026.
Response to Reviewer ye1G [Questions 1]
Q1. Equal-compute teacher correction and continued DMD training
Can the authors compare low-noise correction to an equal-compute high-noise correction and to continued training with the same extra 5K updates, to isolate the benefit of the proposed mechanism?
We conducted the following controlled comparisons to test whether the gains come from low-noise correction, additional training, or standard teacher distillation.
Shared settings.
All post-training variants start from the same one-step DMD VSR student checkpoint and use the same training data, optimizer, learning rate, and 5K additional generator updates. For the comparison across correction noise levels, we also fix the number of teacher evaluations and vary only the starting noise level in Eq. (8). We evaluate all variants on UDM10, SPMCS, and YouHQ40. The baseline is the DMD checkpoint before the additional 5K updates.
| Experiment | Setting | Purpose |
|---|---|---|
| Ours | Apply low-noise correction at and train with the objective in Eqs. (8)–(12). | Proposed method. |
| (a) Continued DMD training | Continue training with the original DMD objective in Eq. (5). | Test whether 5K additional updates alone explain the gains. |
| (b) Higher-noise correction | Keep the same correction procedure and training objective as ours, changing only the starting noise level to or . | Test whether the gains depend on performing correction at low noise. |
| (c) Standard teacher distillation | Starting from pure Gaussian noise, the teacher generates a target over multiple denoising steps conditioned on the LR input. The student learns this target through endpoint regression in Eq. (9). | Test whether standard teacher distillation yields comparable gains. |
UDM10 (n = 10)
| Variant | tOF ↓ | ↓ | tLPIPS ↓ | CLIPIQA ↑ | MUSIQ ↑ | DOVER ↑ |
|---|---|---|---|---|---|---|
| Baseline | 1.384 | 0.877 | 0.1125 | 0.566 | 64.4 | 0.591 |
| Ours () | 0.870 | 0.821 | 0.1040 | 0.572 | 66.4 | 0.621 |
| Higher-noise correction () | 0.975 | 0.919 | 0.1132 | 0.582 | 66.5 | 0.594 |
| Standard distillation () | 1.056 | 0.592 | 0.1446 | 0.502 | 58.0 | 0.448 |
| DMD + 5K updates | 1.263 | 0.901 | 0.1134 | 0.542 | 63.9 | 0.602 |
| Higher-noise correction () | 0.935 | 0.951 | 0.1128 | 0.576 | 66.3 | 0.601 |
SPMCS (n = 30)
| Variant | tOF ↓ | ↓ | tLPIPS ↓ | CLIPIQA ↑ | MUSIQ ↑ | DOVER ↑ |
|---|---|---|---|---|---|---|
| Baseline | 0.372 | 0.791 | 0.0573 | 0.574 | 71.0 | 0.552 |
| Ours () | 0.333 | 0.785 | 0.0576 | 0.574 | 70.3 | 0.555 |
| Higher-noise correction () | 0.361 | 1.131 | 0.0624 | 0.581 | 71.2 | 0.520 |
| Standard distillation () | 0.377 | 0.718 | 0.0812 | 0.501 | 66.1 | 0.371 |
| Higher-noise correction () | 0.364 | 1.053 | 0.0610 | 0.583 | 71.4 | 0.533 |
| DMD + 5K updates | 0.359 | 0.828 | 0.0598 | 0.565 | 70.6 | 0.543 |
YouHQ40 (n = 40)
| Variant | tOF ↓ | ↓ | tLPIPS ↓ | CLIPIQA ↑ | MUSIQ ↑ | DOVER ↑ |
|---|---|---|---|---|---|---|
| Baseline | 1.810 | 1.602 | 0.1157 | 0.560 | 65.5 | 0.709 |
| Ours () | 1.149 | 1.466 | 0.1076 | 0.558 | 66.4 | 0.716 |
| Higher-noise correction () | 1.375 | 1.734 | 0.1128 | 0.560 | 67.3 | 0.719 |
| Standard distillation () | 1.609 | 0.749 | 0.1251 | 0.437 | 57.8 | 0.635 |
| Higher-noise correction () | 1.127 | 1.658 | 0.1118 | 0.561 | 66.9 | 0.723 |
| DMD + 5K updates | 1.708 | 1.516 | 0.1193 | 0.546 | 67.2 | 0.707 |
Results.
Continuing DMD training for 5K updates improves tOF over the initial checkpoint on all three datasets, but worsens tLPIPS on all three and yields higher errors than our method on all three temporal metrics. With the number of teacher evaluations held fixed, raising the starting noise level from 0.15 to either 0.35 or 0.5 increases both and tLPIPS on every dataset. This supports the benefit of performing teacher correction at low noise. Standard distillation from pure noise achieves the lowest , but also the worst tLPIPS and the lowest CLIPIQA, MUSIQ, and DOVER scores on all three datasets, indicating a substantial loss of the perceptual quality achieved by the DMD student.
Response to Reviewer ye1G [Critical Weaknesses 1]
Critical Weaknesses 1: Attention analysis and matched interventions
Reviewer concern: A matched intervention at high noise or a less disruptive control is needed to isolate late cross-frame refinement as the cause.
Matched attention-masking experiments across noise levels.
To address the reviewer's concern about the lack of matched high-noise interventions, we conducted additional experiments across three noise regimes (Table R1), using 30 clips: all 10 UDM10 clips and 10 clips randomly selected with a fixed seed from each of SPMCS and YouHQ40.
For each clip, we first generated an unblocked reference using the teacher's full 50-step denoising trajectory. We then repeated the trajectory from the same initial noise, applying the same cross-frame attention mask for exactly four consecutive steps at three stages: low noise (steps 46–49, σ≤0.30), intermediate noise (steps 42–45, 0.32≤σ≤0.47), and high noise (steps 0–3, σ≈1). The mask prevents each token from attending to nearby spatial positions in neighboring frames. Thus, the masking operation and intervention duration remain fixed, while only the denoising stage varies. Ewarp and tLPIPS are normalized per clip to the corresponding unblocked trajectory and reported as median [IQR]; ΔPSNR and ΔLPIPS are mean per-clip differences.
Table R1. Blocking four steps of the teacher trajectory across three datasets (n=30). Superscripts denote paired Wilcoxon signed-rank tests against the unblocked trajectory: ∗∗p<0.01 for degradation and †p=0.001 for improvement.
Results.
Blocking cross-frame attention causes the largest temporal degradation at low noise (normalized Ewarp: 2.00; tLPIPS: 1.40), smaller degradation at intermediate noise (1.24; 1.20), and no corresponding degradation at high noise (0.90; 1.00). The ordering in Ewarp is consistent across all three datasets. These matched interventions provide stronger evidence that cross-frame attention plays a particularly important role in temporal refinement during late denoising.
Why does masking within-frame neighbors also reduce temporal stability?
At low noise, masking within-frame neighbors worsened both temporal metrics and caused a larger increase in single-frame LPIPS than masking cross-frame neighbors (+0.045 vs. +0.021, Table R1). These results suggest distinct but complementary roles for the two interactions, consistent with the increasingly localized spatiotemporal attention in Fig. 2. Within-frame neighbors provide spatial context for resolving fine textures, while cross-frame neighbors help maintain texture alignment across motion, reducing frame-to-frame shifts that appear as flicker. We will clarify this interpretation in the revised manuscript.