List your company

Sketch-to-image on Sketchy (test)

0.74FIDLOTS
0.67041.14021.612.0798Feb 20, 2026

Evaluation results

MethodLinks
LOTSTraining Protocol=Proposed (Current), Conditioning (Visual/Textual)=Local/Local2026.020.740.660.8260.7060.70.691
IP-AdapterTraining Protocol=Fine-tuned, Conditioning (Visual/Textual)=Global/Global2026.020.750.6210.7990.7510.590.637
LOTS*Training Protocol=Proposed (Prior), Conditioning (Visual/Textual)=Local/Local2026.020.790.6510.8180.7090.6920.651
ControlNetTraining Protocol=Fine-tuned, Conditioning (Visual/Textual)=Global/Global2026.020.820.6550.8120.760.6520.583
ControlNetTraining Protocol=Zero-shot, Conditioning (Visual/Textual)=Global/Global2026.020.960.6330.8010.7090.5790.623
Multi-ControlNetTraining Protocol=Fine-tuned, Conditioning (Visual/Textual)=Local/Global2026.020.960.6340.7990.7220.610.546
Multi-ControlNetTraining Protocol=Zero-shot, Conditioning (Visual/Textual)=Local/Global2026.020.980.6240.7860.6870.5280.638
SDTraining Protocol=Zero-shot, Conditioning (Visual/Textual)=None/Global2026.021.030.6130.7530.6940.430.652
AnyControlTraining Protocol=Zero-shot, Conditioning (Visual/Textual)=Local/Global2026.021.060.6080.7880.6880.5540.495
SDXLTraining Protocol=Zero-shot, Conditioning (Visual/Textual)=None/Global2026.021.090.5670.7570.7810.5470.661
GLIGENTraining Protocol=Zero-shot, Conditioning (Visual/Textual)=None/Local2026.021.10.5440.7090.2820.2130.594
GLIGENTraining Protocol=Fine-tuned, Conditioning (Visual/Textual)=None/Local2026.021.120.570.7340.330.2920.511
T2I-AdapterTraining Protocol=Fine-tuned, Conditioning (Visual/Textual)=Global/Global2026.021.250.5710.7530.7260.5360.597
Multi-T2I-AdapterTraining Protocol=Fine-tuned, Conditioning (Visual/Textual)=Local/Global2026.021.360.5610.7410.7540.4870.592
T2I-AdapterTraining Protocol=Zero-shot, Conditioning (Visual/Textual)=Global/Global2026.022.030.5430.7260.6440.5060.503
Multi-T2I-AdapterTraining Protocol=Zero-shot, Conditioning (Visual/Textual)=Local/Global2026.022.250.520.7090.5270.3710.489
IP-AdapterTraining Protocol=Zero-shot, Conditioning (Visual/Textual)=Global/Global2026.022.480.540.7060.4320.3260.71