Satellite Image Synthesis on Git-Dense 3.5k
18.2FIDTerraDiT-Ω-XL
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TerraDiT-Ω-XLCondition=T + Ω + L, Zero-shot=true2026.06 | 18.2 | 16.21 | 0.3455 | — | — | |
| TerraDiT-Ω-XLCondition=T + B + L, Zero-shot=true2026.06 | 19.46 | 16.28 | 0.3594 | — | — | |
| TerraDiT-Ω-XLCondition=T + P + L, Zero-shot=true2026.06 | 20.57 | 17.44 | 0.3876 | — | — | |
| TerraDiT-Σ-XLCondition=T + P + L, Zero-shot=true2026.06 | 27.49 | 17.95 | 0.4692 | — | — | |
| TerraDiT-α-XLCondition=T, Zero-shot=true2026.06 | 29.76 | 17.94 | 0.4853 | — | — | |
| Text2Earth*Condition=T, Zero-shot=true2026.06 | 40.23 | 17.37 | 0.4871 | — | — | |
| VectorSynthCondition=T + O, Zero-shot=true2026.06 | 72.26 | 25.87 | 0.4505 | — | — | |
| GeoSynth-OSMCondition=T + O, Zero-shot=true2026.06 | 82.32 | 20.78 | 0.44 | — | — | |
| GeoSynthCondition=T, Zero-shot=true2026.06 | 83.9 | 22.26 | 0.5149 | — | — | |
| PixArt-Σ-XLCondition=T, Zero-shot=true2026.06 | 84.18 | 27.13 | 0.5515 | — | — | |
| CRS-DiffCondition=T, Zero-shot=true2026.06 | 89.98 | 17.54 | 0.5141 | — | — | |
| PixArt-α-XLCondition=T, Zero-shot=true2026.06 | 91.22 | 25.72 | 0.5463 | — | — | |
| MHN-VQGANCondition=T, Zero-shot=true2026.06 | 93.92 | 52.62 | 0.4869 | — | — | |
| GLIGENCondition=T + B, Zero-shot=true2026.06 | 96.28 | 24.94 | 0.5256 | — | — | |
| DiffusionSatCondition=T + L, Zero-shot=true2026.06 | 97.04 | 25.51 | 0.5585 | — | — | |
| InstanceDiffusionCondition=T + B, Zero-shot=true2026.06 | 112.35 | 29.93 | 0.5979 | — | — | |
| SD 3Condition=T, Zero-shot=true2026.06 | 113.16 | 23.43 | 0.5752 | — | — | |
| GeoRSSDCondition=T, Zero-shot=true2026.06 | 127.95 | 29.58 | 0.5527 | — | — | |
| AeroGenCondition=T + B, Zero-shot=true2026.06 | 132.84 | 27.45 | 0.5333 | — | — | |
| SDXLCondition=T, Zero-shot=true2026.06 | 187.45 | 27.03 | 0.6076 | — | — | |
| MHN-VQVAECondition=T, Zero-shot=true2026.06 | 224.27 | 28 | 0.6084 | — | — | |
| AeroGenCondition=T + B, Training Set=DIOR/-R, Images=45K, Params=907M2026.06 | — | — | — | 0.1526 | 0.2095 | |
| CRS-DiffCondition=T, Training Set=RSICD/fMoW/MillionAID, Images=1.1M, Params=860M2026.06 | — | — | — | 0.1138 | 0.2854 | |
| DiffusionSatCondition=T + L, Training Set=fMoW/Satlas/SpaceNet, Images=2M, Params=880M2026.06 | — | — | — | 0.0983 | 0.2803 | |
| GeoRSSDCondition=T, Training Set=RS5M, Images=5M, Params=866M2026.06 | — | — | — | 0.1049 | 0.2798 | |
| GeoSynthCondition=T, Training Set=Internal, Images=45K, Params=866M2026.06 | — | — | — | 0.1178 | 0.2649 | |
| GeoSynth-OSMCondition=T + O, Training Set=Internal, Images=45K, Params=1.23B2026.06 | — | — | — | 0.145 | 0.2705 | |
| GLIGENCondition=T + B, Training Set=LVIS/GoldG/O365/SBU/CC3M, Images=7M, Params=1.07B2026.06 | — | — | — | 0.0949 | 0.2649 | |
| InstanceDiffusionCondition=T + B, Training Set=Internal, Images=5M, Params=1.23B2026.06 | — | — | — | 0.0772 | 0.2249 | |
| MHN-VQGANCondition=T, Training Set=RSICD, Images=9K, Params=116M2026.06 | — | — | — | 0.1032 | 0.2478 | |
| MHN-VQVAECondition=T, Training Set=RSICD, Images=9K, Params=29M2026.06 | — | — | — | 0.1144 | 0.2173 | |
| PixArt-α-XLCondition=T, Training Set=ImageNet/SAM/Internal, Images=25M, Params=611M2026.06 | — | — | — | 0.0669 | 0.2657 | |
| PixArt-Σ-XLCondition=T, Training Set=Internal, Images=33M, Params=611M2026.06 | — | — | — | 0.0912 | 0.2631 | |
| SD 3Condition=T, Training Set=Internal, Images=N/A, Params=2B2026.06 | — | — | — | 0.1012 | 0.2639 | |
| SDXLCondition=T, Training Set=Internal, Images=N/A, Params=2.6B2026.06 | — | — | — | 0.1004 | 0.249 | |
| TerraDiT-α-XLCondition=T, Training Set=Git-10M, Images=2M, Params=836M2026.06 | — | — | — | 0.1275 | 0.2661 | |
| TerraDiT-Σ-XLCondition=T + P + L, Training Set=Git-10M/OSM, Images=2M, Params=1.12B2026.06 | — | — | — | 0.1329 | 0.2671 | |
| TerraDiT-Ω-XLCondition=T + P + L, Training Set=Git-10M/OSM, Images=2M, Params=1.18B2026.06 | — | — | — | 0.1503 | 0.2787 | |
| TerraDiT-Ω-XLCondition=T + B + L, Training Set=Git-10M/OSM, Images=2M, Params=1.18B2026.06 | — | — | — | 0.1656 | 0.2813 | |
| TerraDiT-Ω-XLCondition=T + Ω + L, Training Set=Git-10M/OSM, Images=2M, Params=1.18B2026.06 | — | — | — | 0.1757 | 0.2821 | |
| Text2Earth*Condition=T, Training Set=Git-10M, Images=10M, Params=867M2026.06 | — | — | — | 0.1349 | 0.2914 | |
| VectorSynthCondition=T + O, Training Set=OSM-Polygon, Images=20K, Params=1.23B2026.06 | — | — | — | 0.1111 | 0.2842 |