Image Reconstruction on ImageNet 256p
35.9PSNRQwen-Image-VAE-2.0-f16c128
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Qwen-Image-VAE-2.0-f16c128Setting=f16c1282026.05 | 35.9 | — | — | 95.19 | |
| FLUX.2-devSetting=f16c1282026.05 | 34.34 | — | — | 93.58 | |
| ViTok-v2 5BComp.=f16×64, r=122026.05 | 34.2 | 0.74 | 2.49 | 92.4 | |
| ViTok-v2 5BComp.=f16×64, r=12, high DINO loss weight (λ=1000)=true2026.05 | 33.6 | 0.3 | 1.12 | 91.4 | |
| Qwen-ImageSetting=f8c162026.05 | 33.42 | — | — | 91.59 | |
| HunyuanComp.=f8×16, r=122026.05 | 33.3 | 0.67 | — | 91.6 | |
| HunyuanVideoSetting=f8c162026.05 | 33.21 | — | — | 91.43 | |
| FLUX.1-devSetting=f8c162026.05 | 32.84 | — | — | 91.55 | |
| Qwen-Image-VAE-2.0-f16c64Setting=f16c642026.05 | 32.72 | — | — | 90.86 | |
| Cosmos-0.1-CI8x8Setting=f8c162026.05 | 32.33 | — | — | 90.24 | |
| Stepvideo-T2VSetting=f16c642026.05 | 31.54 | — | — | 89.73 | |
| Wan2.2Setting=f16c482026.05 | 31.3 | — | — | 87.84 | |
| Wan2.1Setting=f8c162026.05 | 31.29 | — | — | 88.7 | |
| ViTok-v2 5BComp.=f16×32, r=242026.05 | 31.2 | 1.26 | 2.94 | 86.7 | |
| HunyuanVideo-1.5Setting=f16c322026.05 | 31.18 | — | — | 87.1 | |
| Qwen-Image-VAE-2.0-f32c192Setting=f32c1922026.05 | 31.13 | — | — | 87.85 | |
| FLUX.2Comp.=f8×16, r=122026.05 | 31.1 | 0.15 | — | 88.7 | |
| FLUX.1Comp.=f8×16, r=122026.05 | 31.1 | 0.24 | 1.26 | 88.7 | |
| HunyuanImage-3.0Setting=f16c322026.05 | 31.08 | — | — | 86.55 | |
| VA-VAEComp.=f16×64, r=122026.05 | 30.7 | 0.14 | — | 87 | |
| QwenComp.=f8×16, r=122026.05 | 30.3 | 1.32 | 7.36 | 86 | |
| Qwen-Image-VAE-2.0-f32c128Setting=f32c1282026.05 | 29.69 | — | — | 84.23 | |
| LTX-VideoSetting=f32c1282026.05 | 29.57 | — | — | 83.29 | |
| ATokenComp.=f16×32, r=242026.05 | 28.8 | 0.26 | — | 81.4 | |
| HunyuanImage-2.1Setting=f32c642026.05 | 28.67 | — | — | 81.99 | |
| ViTok-v2 5BComp.=f16×16, r=482026.05 | 28.5 | 1.52 | 3.66 | 79.3 | |
| VA-VAEComp.=f16×32, r=242026.05 | 28 | 0.28 | — | 79 | |
| VAVAESetting=f16c322026.05 | 27.75 | — | — | 79.86 | |
| VTP-LargeSetting=f16c642026.05 | 26.88 | — | — | 77.18 | |
| LTX-2Setting=f32c1282026.05 | 26.06 | — | — | 79.25 | |
| SDXL-VAEComp.=f8×8, r=242026.05 | 26 | 0.68 | — | 83.4 | |
| SD-VAEComp.=f8×4, r=482026.05 | 25.7 | 0.73 | 6.14 | 70.2 | |
| VA-VAEComp.=f16×16, r=482026.05 | 25.3 | 0.55 | — | 69 | |
| Cosmos-0.1-CI16x16Setting=f16c162026.05 | 25.13 | — | — | 70.15 | |
| DC-AE-sanaSetting=f32c322026.05 | 24.82 | — | — | 68.97 | |
| RAE-SigLIP2-BSetting=f16c7682026.05 | 19.71 | — | — | 52.79 | |
| RAE-DINOv2-BSetting=f16c7682026.05 | 19.24 | — | — | 50.25 | |
| RAEComp.=f16×16, r=482026.05 | 18.8 | 0.61 | — | 49.6 |