can you adjust the default setting .8 is too high for the first one as the low setting set that as middle

for the low just set to - 1

works with other models... like I have it up with flux and it seemed to work with Pony too, the trained tags are weak and the model only has 8 instances of the most common tags in the set so it is not super strong in terms of the influence on CLIP. Clip does not appear to comprehend the word "cosmicstyle" and the word never appears on its own as a tag so it is split across tokens and probably registers as nonsense to CLIP. Just generating with "cosmicstyle" in the positive with nothing else and no negative on a few different SDXL models just yields patterns of abstract nothing so the term has no deeper meaning. I don't see a big difference turning it off and on in unrelated prompts and subjects. It basically looks like a soft image/fog/blurry modifier IMO. I tried a medium tag, "cosmicstyle coffee machine" with five occurrences in the trained tags and just got latte art and some broken spoons or cups like the model is saying it is confused by something. Am I missing something here?