Table 5

Inconsistency and intrinsic difficulty of different diffusion-based generative AI models. Classification error rate and obtained with EfficientNetv2 small.

Testing TrainingFLUX.l-schnellStable Diffusion 3SD XL TurboWuerstchenKandinsky 2.2
FLUX.l-schnell0.91%4.83%23.18%9.09%5.43%
Stable Diffusion 35.63%1.51%16.69%4.60%6.27%
SD XL Turbo18.73%31.76%1.29%23.30%24.05%
Wuerstchen16.12%18.87%22.21%0.61%16.69%
Kandinsky 2.29.13%9.57%21.78%6.40%1.26%

or Create an Account

Close Modal
Close Modal