Ways to controlling the sampler:

Controlnets

Controlnets allow you to control your final image by allowing you to inject concepts into the prompt. Common usage include, using a input image to generate a depth map, outlines via canny edge detection, or poses, then using the output from the preprocessor to control the prompting. Parameters: Strength Dictates how strongly the controlnet will affect the result Start & End Percent Dictates when in the denoising process in the sampler will the controlnet be active, eg; start_percent: 0.0 & end_percent 0.5 - will result in the controlnet being active for the first 50% of the sampling steps

IPAdaptor

IPAdaptors allow you to use an image as a prompt to control the style or composition of the final image.