Super-resolution ×2 — SeedVR2 vs Real-ESRGAN (Set14, RTX 5090)

Timings & VRAM

Variant	CFG	Steps	Load	Avg gen / img	Peak VRAM	PSNR ↑	SSIM ↑	LPIPS ↓	DISTS ↓	NIQE ↓	MUSIQ ↑	CLIP-IQA ↑	Success
Bicubic x2 (baseline)	—	—	—	0.0s	—	28.08	0.8492	0.1877	0.1268	5.42	52.27	0.5958	14 / 14
Ground truth (original HR)	—	—	—	—	—	—	—	—	—	4.90°	66.96°	0.7513°	14 / 14
Real-ESRGAN x2 (fl16)	—	1	4.0s	0.1s	0.25 GB	25.42	0.7608	0.1575	0.1139	4.66	66.95	0.6364	14 / 14
SeedVR2 3B (bf16, 1-step)	1.0	1	7.6s	0.7s	8.11 GB	21.61	0.6090	0.2309	0.1550	4.37	71.18	0.7583	14 / 14
SeedVR2 7B (bf16, 1-step)	1.0	1	9.2s	0.9s	17.70 GB	22.76	0.6559	0.1779	0.1330	4.14	71.55	0.7677	14 / 14
SeedVR2 7B-sharp (bf16, 1-step)	1.0	1	9.0s	0.9s	17.70 GB	23.23	0.6864	0.1591	0.1197	3.86	71.20	0.7661	14 / 14

Shaded columns are no-reference — they judge “does this look like a real photograph?” with no ground truth. They have no winner: the ° ground-truth row is the calibration line, and out-scoring it means inventing texture prettier than the original. The unshaded columns are full-reference and measure distance to the ground truth. A model can win one axis and lose the other — that is the perception–distortion tradeoff, not a contradiction.

Drag the slider to compare A vs B; switch either dropdown to pick any variant. The number next to each dropdown is that variant's generation time for this example. Click the input thumbnail to view it full size.