Models Translation (Dolly/Pedestal/Truck) Zooming Rotation (Pan/Tilt/Roll) Static Avg
InOutUpDownRightLeft InOut RightLeftUpDownCWCCW
🏅Qwen2.5-VL-72B (Ours SFT) 86.841.330.550.670.752.6 57.178.5 81.986.679.170.435.043.8 76.662.8
🥈Qwen2.5-VL-32B (Ours SFT) 85.640.129.349.469.651.5 56.077.3 80.785.477.969.233.942.7 75.461.6
🥉Qwen2.5-VL-7B (Ours SFT) 83.938.627.847.867.950.0 54.575.8 79.283.876.367.632.341.0 73.660.0
MegaSAM 73.843.924.229.145.344.2 11.110.2 79.582.273.865.371.575.8 22.050.1
MASt3R 47.521.123.540.238.738.1 42.246.6 66.658.063.240.350.453.5 15.743.1
CUT3R 68.950.424.734.237.027.6 15.921.3 59.165.065.047.560.766.2 15.142.7
VGGSFM 56.628.928.738.248.935.3 21.717.3 60.958.746.643.361.455.5 16.741.3
Qwen2.5-VL-72B 58.016.820.632.536.429.5 31.745.4 64.768.656.436.020.626.1 21.637.7
Qwen2.5-VL-32B 57.616.420.232.136.029.2 31.445.0 64.368.256.035.620.225.7 21.237.3
DUSt3R 58.924.030.718.038.326.9 18.224.6 59.463.832.927.361.057.9 13.137.0
InternVL3-78B 72.018.219.632.533.829.4 26.433.4 47.253.547.840.327.625.0 22.636.8
GPT-4o 66.329.221.138.238.021.9 41.739.3 44.742.143.635.524.028.7 32.036.4
Qwen2.5-VL-7B 56.014.918.730.534.527.6 29.843.4 62.766.754.534.118.824.2 19.835.8
InternVL3-8B 61.215.518.829.030.527.3 29.528.1 41.649.342.036.521.322.3 20.131.5
InternVL2.5-26B 63.611.821.123.627.219.4 21.831.6 42.538.344.943.614.318.2 25.129.8
InternVL2.5-8B 67.912.928.125.923.423.2 18.632.1 37.430.937.636.911.525.3 23.429.5
COLMAP 36.213.111.919.734.130.0 13.914.2 43.946.428.319.142.148.7 7.527.3
LLaVA-Video-7B 54.715.216.519.327.123.6 16.216.9 33.636.826.937.216.121.7 22.125.6
LLaVA-OneVision-7B 46.813.512.616.923.720.2 10.714.4 33.533.616.931.419.320.8 18.822.2
InternVideo2-Chat-8B 69.918.519.317.617.923.4 12.210.4 22.622.717.222.819.616.4 20.222.0
Tarsier-Recap-7B 59.715.125.723.728.821.5 14.415.0 22.827.324.621.615.218.7 30.721.0
mPLUG-Owl3-7B 47.612.913.916.917.318.5 12.910.6 31.426.626.137.010.412.2 17.820.8
InternLMXComposer2.5-7B 49.010.611.410.414.610.6 11.816.5 14.313.914.717.511.718.1 21.816.5