>>86115106Snuffy looking a bit boyish in this styleContinuing the sampler exploration, I never realized how weird euler smea dy is.
Not only is there immense variance between schedulers, but also cfgs.
https://litter.catbox.moe/z9bwqp.pngAlso, I'm not sure if I'm going crazy, but I feel like the comfy smea implementation is not only reliant on the seed, but the entire workflow, cause after forgetting the euler comparison again, I redid the xy-plot but the results are completely different.
see:
https://litter.catbox.moe/61s1pv.png>>86120333Yeah, all the cfg++ samplers work really well with it. I did a small comparison at different cfgs and schedulers for the euler a and dpmpp 2s one if you hadn't seen it.
>>86042545>>86116596cute!