>>105596505
Comparison is difficult because different models/finetunes/merges react to different samplers in different ways depending on how they were trained. It's always best to test it yourself with an xy plot. It'll take awhile on bigger models like flux/chroma, but it's worth it imo.