Blackberry-anon back. Messing around with blueberry-anon's recipe again, I've swapped out NAI for AnythingV3 in the recipe.

I'm reposting the previous recipes here, and adding the new ones.

Blueberry:
>NAI + SD1.5 Weighted Sum @ 0.25 -> NAI-v1.5-0.25
>NAI-v1.5-0.25 + F222 + SD1.5 Difference @ 1.0 -> berry-lite
>berry-lite + r34_e4 Weighted Sum @ 0.15 -> blueberrymix

Blackberry:
>NAI + SD1.4 Weighted Sum @ 0.25 -> NAI-v1.4-0.25
>NAI-v1.4-0.25 + F111 + SD1.4 Difference @ 1.0 -> berry-lite
>berry-lite + r34_e4 Weighted Sum @ 0.15 -> blackberrymix

Raspberry:
>AnyV3 + SD1.5 Weighted Sum @ 0.25 -> AnyV3-v1.5-0.25
>AnyV3-v1.5-0.25 + F222 + SD1.5 Difference @ 1.0 -> berry-lite
>berry-lite + r34_e4 Weighted Sum @ 0.15 -> raspberrymix

Strawberry:
>AnyV3 + SD1.4 Weighted Sum @ 0.25 -> AnyV3-v1.4-0.25
>AnyV3-v1.4-0.25 + F111 + SD1.4 Difference @ 1.0 -> berry-lite
>berry-lite + r34_e4 Weighted Sum @ 0.15 -> strawberrymix

(Note: berrymix-v2 in my plot is just the normal berrymix recipe from here https://rentry.org/LFTBL#berrymix but swapping F111 for F222 and SD1.4 for SD1.5)

I am really grateful to blueberry-anon for this recipe, because it works an absolute treat swapping out any ingredients as long as each of them is good on their own. I think the key is that very particular 0.15 ratio of R34-e4. Going by what he said, which is that 0.05 in either direction turns the merge into an abomination, it must have been time consuming to find the exact right interpolation ratio.