FlexVSR: Continuous-Scale Super-Resolution for Streaming Video Generation
Organizations: Philo AI · City University of Hong Kong
Abstract
Real-time streaming video generation has gained rapid momentum in live content creation and interactive media. Deploying such systems in practice requires flexible scaling factors to accommodate diverse source resolutions, target displays, and computational budgets. However, extending fixed-scale generative video super-resolution (VSR) to varying target scales poses two challenges: downsampling from a larger fixed scale incurs redundant cost, while re-optimizing the diffusion transformer (DiT) is costly and alters the shared generative prior. To address both, we propose FlexVSR, a decoupled framework for continuous-scale streaming generative VSR with supervision at only a few scales. Our key insight is that naively resampling fixed-scale latents onto a new target grid induces scale-dependent input mismatch, which can be effectively addressed through parameter-efficient, scale-aware adaptation of the latent interface without updating the DiT. To this end, FlexVSR introduces ReLSA (Refined Latent Scale Adaptation), a lightweight latent adapter that constructs and refines target-grid latents before they enter the DiT, enabling a shared frozen VSR prior to generalize across continuous scales with only 0.274M trainable parameters and an adaptation cost of 7.3 H100 GPU-hours, while introducing no additional temporal dependencies and preserving the native causal streaming protocol. Extensive experiments demonstrate competitive or superior reconstruction and temporal quality, lower latency than most native-scale streaming baselines and offline methods, and effective generalization to unseen continuous scales, filling the continuous-scale gap in real-time streaming generative VSR.
Figures & tables
| Dataset | Type | Method (venue) | Avg. Rank | ||||||
|---|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | ||||
| LiteVSR N [ 5 ] (ICML 2026) | 26.79 | 0.1943 | 0.01038 | 25.97 | 0.2200 | 0.01187 | 2.83 | ||
| Offline VSR methods | LiteVSR D (ICML 2026) | 23.06 | 0.3576 | 0.01470 | 23.35 | 0.3392 | 0.01434 | 9.67 | |
| LUVE N [ 47 ] (ICML 2026) | 16.64 | 0.4398 | 0.02030 | 16.32 | 0.4540 | 0.02222 | 12.00 | ||
| VEnhancer N [ 16 ] (arXiv 2024) | 19.90 | 0.3274 | 0.01736 | 20.15 | 0.3254 | 0.01798 | 10.50 | ||
| FlashVSR-T N [ 49 ] (CVPR 2026) | 26.51 | 0.1723 | 0.01440 | 25.72 | 0.1851 | 0.01421 | 4.33 | ||
| Method | ||||||
|---|---|---|---|---|---|---|
| FPS | FPS | |||||
| FlashVSR-T N | 10.335 | 6.408 | 2.50 | 5.678 | 3.588 | 4.46 |
| FlashVSR-T D | 2.353 | 1.870 | 8.55 | 1.890 | 1.411 | 11.34 |
| FlashVSR-F N | 18.994 | 14.105 | 1.13 | 9.564 | 6.188 | 2.59 |
| FlashVSR-F D | 3.402 | 2.563 | 6.24 | 2.948 | 2.138 | 7.48 |
| InfVSR N | 97.299 | 89.151 | 0.18 | 57.139 | 52.347 | 0.31 |
| Metric | Method | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Actual | 1.20/1.20 | 1.25/1.25 | 1.29/1.30 | 1.40/1.40 | 1.50/1.50 | 1.61/1.60 | 1.69/1.69 | 1.74/1.74 | 1.78/1.79 | 2.00/2.00 | |
| FlashVSR-T N | 27.19 | 27.08 | 26.95 | 26.77 | 26.51 | 26.33 | 26.14 | 26.08 | 25.99 | 25.72 | |
| FlashVSR-T D | 23.64 | 23.64 | 23.64 | 23.65 | 23.65 | 23.66 | 23.66 | 23.66 | 23.66 | 23.67 | |
| PSNR | InfVSR N | 24.21 | 24.24 | 24.29 | 24.42 | 24.47 | 24.53 | 24.57 | 24.59 | 24.62 | 24.72 |
| InfVSR D | 24.53 | 24.56 | 24.60 | 24.66 | 24.72 | 24.78 | 24.82 | 24.83 | 24.85 | 24.92 | |
| Ultra Flash N | 25.80 | 25.65 | 25.54 | 25.29 | 25.10 | 24.85 | 24.71 | 24.60 | 24.52 | 24.14 |
| Method | ||||||
|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | |
| FlashVSR-T N | 25.52 | 0.1897 | 0.01404 | 25.44 | 0.1911 | 0.01386 |
| FlashVSR-T D | 23.68 | 0.2207 | 0.01416 | 23.68 | 0.2206 | 0.01416 |
| InfVSR N | 24.79 | 0.2471 | 0.01572 | 24.84 | 0.2462 | 0.01570 |
| InfVSR D | 24.93 | 0.2238 | 0.01446 | 24.94 | 0.2242 | 0.01453 |
| Ultra Flash N | 18.72 | 0.2227 | 0.01420 | 18.63 | 0.2284 | 0.01440 |
| Variant | Settings | (unseen) | (unseen) | |||||||||||||
| Q | R | E2E | SR | PSNR | LPIPS | DISTS | PSNR | LPIPS | DISTS | PSNR | LPIPS | DISTS | PSNR | LPIPS | DISTS | |
| (a) Stage-2 adaptation data ablation (w/o VSR-17K for ReLSA; shared Stage-1 SR prior) | ||||||||||||||||
| FlexVSR (Stage-2 REDS) | 25.99 | 0.1469 | 0.1016 | 26.98 | 0.1174 | 0.0821 | 25.85 | 0.1539 | 0.1061 | 26.35 | 0.1448 | 0.1102 | ||||
| (b) Architecture ablations | ||||||||||||||||
| w/o Query | 25.37 | 0.1713 | 0.1108 | 27.13 | 0.1132 | 0.0773 | 25.32 | 0.1518 | 0.1004 | 26.58 | 0.1320 | 0.0967 | ||||
| w/o Res | 26.40 | 0.1413 | 0.0851 | 26.63 | 0.1266 | 0.0838 | 25.77 | 0.1497 | 0.1046 | 25.93 | 0.1630 | 0.1164 | ||||
Appendix figures & tables17 assets
Supplementary material from the paper’s appendix.
Appendix
| Statistic | Training | Validation |
|---|---|---|
| Clips / source videos | 17,377 / 3,493 | 232 / 51 |
| Video bitrate: median (Mbps) | 15.76 | 17.37 |
| Clip duration: median [ , ] (s) | 15.57 [9.01, 22.99] | 9.01 [9.00, 15.01] |
| Frame rate: median [min, max] (FPS) | 29.97 [20.42, 60] | 29.97 [23.98, 60] |
| source clips | 16,792 (96.6%) | 226 (97.4%) |
| Frames per prepared pair (main model) | 21 | 21 |
| Method | Avg. Rank | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | ||
| Offline VSR methods | ||||||||||
| LiteVSR N | 26.74 | 0.1767 | 0.00792 | 26.39 | 0.2007 | 0.00965 | 25.97 | 0.2200 | 0.01187 | 2.67 |
| LiteVSR D | 23.16 | 0.3411 | 0.01281 | 23.29 | 0.3439 | 0.01351 | 23.35 | 0.3392 | 0.01434 | 9.00 |
| LUVE N | 16.43 | 0.4298 | 0.01785 | 16.38 | 0.4427 | 0.01963 | 16.32 | 0.4540 | 0.02222 | 12.00 |
| VEnhancer N | 19.35 | 0.3176 | 0.01364 | 19.53 | 0.3243 | 0.01560 | 20.15 | 0.3254 | 0.01798 | 10.67 |
| Dataset | Scale | PSNR | SSIM | LPIPS | DISTS | tLPIPS | tOF |
|---|---|---|---|---|---|---|---|
| YouHQ40 | 26.38 | 0.7729 | 0.1421 | 0.0819 | 0.01406 | 0.4734 | |
| YouHQ40 | 25.98 | 0.7551 | 0.1525 | 0.0864 | 0.01476 | 0.4945 | |
| Inter4K-40 | 25.55 | 0.7747 | 0.1220 | 0.0790 | 0.00884 | 0.2093 | |
| Inter4K-40 | 25.56 | 0.7726 | 0.1347 | 0.0823 | 0.00837 | 0.2190 |
| Model | YouHQ40, | YouHQ40, | Inter4K-40, | Inter4K-40, | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | |
| FlexVSR (main) | 27.25 | 0.1122 | 0.01244 | 26.79 | 0.1277 | 0.01242 | 26.17 | 0.1040 | 0.00723 | 26.38 | 0.1184 | 0.00801 |
| FlexVSR-KV | 27.30 | 0.1131 | 0.01178 | 26.84 | 0.1311 | 0.01187 | 26.59 | 0.0964 | 0.00764 | 26.46 | 0.1136 | 0.00838 |
| FlexVSR + TAEHV | 26.38 | 0.1421 | 0.01406 | 25.98 | 0.1525 | 0.01476 | 25.55 | 0.1220 | 0.00884 | 25.56 | 0.1347 | 0.00837 |
| FlexVSR-KV + TAEHV | 26.39 | 0.1397 | 0.01448 | 26.04 | 0.1519 | 0.01517 | 25.91 | 0.1153 | 0.00933 | 25.85 | 0.1303 | 0.00864 |
| Model | ||||||
|---|---|---|---|---|---|---|
| FPS | FPS | |||||
| FlexVSR (main) | 1.957 | 2.387 | 6.70 | 1.885 | 2.388 | 6.70 |
| FlexVSR-KV | 1.911 | 2.287 | 7.00 | 1.893 | 2.263 | 7.07 |
| FlexVSR + TAEHV | 0.389 | 0.674 | 23.73 | 0.385 | 0.668 | 23.95 |
| FlexVSR-KV + TAEHV | 0.407 | 0.562 | 28.44 | 0.403 | 0.558 | 28.69 |
| Method | Mean metric rank | Overall Avg. Rank | Position | ||
|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | |||
| FlashVSR-T D | 2.59 | 2.47 | 2.47 | 2.51 | 3 |
| Ultra Flash N | 2.35 | 2.35 | 2.29 | 2.33 | 2 |
| FlexVSR (Ours) | 1.06 | 1.18 | 1.24 | 1.16 | 1 |
| Method | † | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | |
| FlashVSR-T D | 23.64 | 0.2209 | 0.01430 | 3.00 | 23.64 | 0.2209 | 0.01429 | 3.00 | 23.64 | 0.2208 | 0.01428 | 3.00 |
| Ultra Flash N | 25.80 | 0.1225 | 0.01199 | 2.00 | 25.65 | 0.1258 | 0.01220 | 2.00 | 25.54 | 0.1285 | 0.01234 | 2.00 |
| FlexVSR (Ours) | 27.08 | 0.1116 | 0.01161 | 1.00 | 27.16 | 0.1105 | 0.01170 | 1.00 | 27.22 | 0.1101 | 0.01170 | 1.00 |
| Method | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | |
| FlashVSR-T D | 23.65 | 0.2208 | 0.01427 | 3.00 | 23.65 | 0.2209 | 0.01426 | 3.00 | 23.66 | 0.2207 | 0.01425 | 3.00 |
| Ultra Flash N | 25.29 | 0.1349 | 0.01267 | 2.00 | 25.10 | 0.1405 | 0.01298 | 2.00 | 24.85 | 0.1477 | 0.01328 | 2.00 |
| FlexVSR (Ours) | 27.25 | 0.1109 | 0.01184 | 1.00 | 27.23 | 0.1118 | 0.01257 | 1.00 | 27.11 | 0.1161 | 0.01206 | 1.00 |
| Method | † | † | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | |
| FlashVSR-T D | 23.66 | 0.2207 | 0.01425 | 3.00 | 23.66 | 0.2207 | 0.01423 | 3.00 | 23.66 | 0.2207 | 0.01423 | 3.00 |
| Ultra Flash N | 24.71 | 0.1528 | 0.01341 | 2.00 | 24.60 | 0.1565 | 0.01358 | 2.00 | 24.52 | 0.1596 | 0.01371 | 2.00 |
| FlexVSR (Ours) | 26.99 | 0.1200 | 0.01209 | 1.00 | 26.91 | 0.1226 | 0.01212 | 1.00 | 26.83 | 0.1256 | 0.01215 | 1.00 |
| Method | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | |
| FlashVSR-T D | 23.67 | 0.2206 | 0.01421 | 3.00 | 23.67 | 0.2206 | 0.01419 | 3.00 | 23.69 | 0.2207 | 0.01411 | 2.00 |
| Ultra Flash N | 24.36 | 0.1660 | 0.01388 | 2.00 | 24.14 | 0.1699 | 0.01416 | 2.00 | 18.44 | 0.2466 | 0.01459 | 3.00 |
| FlexVSR (Ours) | 26.62 | 0.1329 | 0.01221 | 1.00 | 26.64 | 0.1378 | 0.01331 | 1.00 | 25.44 | 0.1811 | 0.01310 | 1.00 |
| Method | † | † | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | |
| FlashVSR-T D | 23.69 | 0.2208 | 0.01406 | 2.00 | 23.69 | 0.2209 | 0.01400 | 2.00 | 23.69 | 0.2214 | 0.01397 | 1.33 |
| Ultra Flash N | 18.38 | 0.2590 | 0.01488 | 3.00 | 18.32 | 0.2702 | 0.01528 | 3.00 | 18.26 | 0.2920 | 0.01529 | 3.00 |
| FlexVSR (Ours) | 25.06 | 0.1997 | 0.01347 | 1.00 | 24.64 | 0.2190 | 0.01383 | 1.00 | 24.17 | 0.2439 | 0.01441 | 1.67 |
| Method | ||||||||
|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | Rank | PSNR | LPIPS | tLPIPS | Rank | |
| FlashVSR-T D | 23.68 | 0.2219 | 0.01394 | 1.33 | 23.65 | 0.2236 | 0.01387 | 1.00 |
| Ultra Flash N | 18.22 | 0.3061 | 0.01573 | 3.00 | 18.17 | 0.3360 | 0.01589 | 2.67 |
| FlexVSR (Ours) | 23.85 | 0.2617 | 0.01499 | 1.67 | 23.08 | 0.3097 | 0.01597 | 2.33 |
| DiT backbone | (unseen) | (unseen) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | PSNR | LPIPS | tLPIPS | |
| Wan2.1-1.3B (main) | 26.42 | 0.1366 | 0.01151 | 27.25 | 0.1122 | 0.01244 | 26.20 | 0.1428 | 0.01048 | 26.79 | 0.1277 | 0.01242 |
| Wan2.1-14B | 26.46 | 0.1326 | 0.01216 | 27.51 | 0.1072 | 0.01181 | 26.27 | 0.1387 | 0.01006 | 27.10 | 0.1219 | 0.01195 |
| MiniMax-H3 | 23.47 | 0.2001 | 0.01429 | 23.86 | 0.1883 | 0.01409 | 22.61 | 0.2231 | 0.01470 | 23.74 | 0.2086 | 0.01487 |
| Method | RGB resize before the model | Subsequent processing |
|---|---|---|
| FlashVSR-T D [ 49 ] | PIL bicubic to ; normalize to in BF16. | Reflect-pad right/bottom to multiples of 128; LQ projection, DiT, and TCDecoder; remove padding. |
| FlashVSR-F D [ 49 ] | Same RGB resize and padding as FlashVSR-T D . | Same target-grid LQ conditioning and DiT; Wan VAE HR decoder replaces TCDecoder. |
| InfVSR D [ 46 ] | Reshape each RGB chunk from to ; FP32 bicubic with antialiasing to . | Restore the temporal layout, then Wan VAE encoding, causal DiT restoration, and Wan VAE decoding. |
| LiteVSR D [ 5 ] | FP32 bilinear resize of the RGB tensor to ; align_corners=False ; normalize to . | Wan2.2 VAE encoding, five-step UniPC restoration with the released adapter, and VAE decoding. |
| Method | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PSNR | SSIM | LPIPS | DISTS | tLPIPS | tOF | PSNR | SSIM | LPIPS | DISTS | tLPIPS | tOF | |
| Offline VSR methods | ||||||||||||
| LiteVSR N | 26.79 | 0.8240 | 0.1943 | 0.1172 | 0.01038 | 0.3742 | 25.97 | 0.7805 | 0.2200 | 0.1279 | 0.01187 | 0.4115 |
| LiteVSR D | 23.06 | 0.6179 | 0.3576 | 0.1784 | 0.01470 | 0.5923 | 23.35 | 0.6334 | 0.3392 | 0.1712 | 0.01434 | 0.5772 |
| LUVE N | 16.64 | 0.4558 | 0.4398 | 0.1571 | 0.02030 | 0.9688 | 16.32 | 0.4302 | 0.4540 | 0.1651 | 0.02222 | 0.9889 |
| VEnhancer N | 19.90 | 0.5406 | 0.3274 | 0.1395 | 0.01736 | 0.7017 | 20.15 | 0.5525 | 0.3254 | 0.1377 | 0.01798 | 0.6844 |
Video Super-Resolution
Video super-resolution reconstructs sharper, higher-resolution video from low-resolution frames while preserving detail and temporal consistency.
See all 45 papers