Image Reconstruction on ImageNet 512x512
0.2rFIDWAN2.2 VAE
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| WAN2.2 VAEResolution=5122026.06 | 0.2 | 30.48 | 0.9 | — | |
| DC-AESetting (f/c)=f64c128, Latent Shape=8x8x1282024.10 | 0.22 | 26.15 | 0.71 | 0.08 | |
| DC-AESetting (f/c)=f128c512, Latent Shape=4x4x5122024.10 | 0.23 | 25.73 | 0.7 | 0.084 | |
| RepViTokResolution=5122026.06 | 0.23 | 31 | 0.92 | — | |
| ViTokResolution=5122026.06 | 0.24 | 30.77 | 0.92 | — | |
| IBQDynamic Resolution=supported, Dynamic Tokens=unsupported, Tokens=16x‡2026.04 | 0.4 | — | — | — | |
| Open-MAGVIT-v2Dynamic Resolution=supported, Dynamic Tokens=unsupported, Tokens=16x‡2026.04 | 0.5 | — | — | — | |
| VibeToken-LLDynamic Resolution=supported, Dynamic Tokens=supported, Tokens=32–2562026.04 | 0.51 | — | — | — | |
| VibeToken-SLDynamic Resolution=supported, Dynamic Tokens=supported, Tokens=32–2562026.04 | 0.55 | — | — | — | |
| LLaMAGen-TokDynamic Resolution=supported, Dynamic Tokens=unsupported, Tokens=16x‡2026.04 | 0.7 | — | — | — | |
| DA-HTDynamic Resolution=supported, Dynamic Tokens=unsupported, Tokens=32x‡2026.04 | 0.83 | — | — | — | |
| InstellaDynamic Resolution=supported, Dynamic Tokens=unsupported, Tokens=1282026.04 | 1.32 | — | — | — | |
| TiTok-BDynamic Resolution=unsupported, Dynamic Tokens=unsupported, Tokens=64/128, resolution-specific training=true2026.04 | 1.52 | — | — | — | |
| SD-VAESetting (f/c)=f64c128, Latent Shape=8x8x1282024.10 | 16.84 | 19.49 | 0.48 | 0.282 | |
| SD-VAESetting (f/c)=f128c512, Latent Shape=4x4x5122024.10 | 100.74 | 15.9 | 0.4 | 0.531 |