Make step size a varying parameter per image based on convergence?

Currently playing around with stable diffusion and it seems step size is a fixed parameter that you set at the beginning, and is used for every single image. 

Is there a way to have it vary per image based on convergence? Some images may converge to a decent final representation much quicker than others, so having it stop at an earlier step would make sense.