/vt/ - Virtual Youtubers

>>67662293
>>67661009
>>67664012
Yeah, I messed up. I was trying to do this with sdxl encodes at the time of writing that and I somehow thought I was always using those.

It's actually not as simple to use SDXL encode, as the choice (so {a|b|c}) gets selected independently in text_g and text_l. You'd probably be able to hand it in from outside, but the standard CLIPTextEncode is much simpler to use for this, and is also what I used(for the most part. I don't wanna look through all the pics and see if one of them was made during my testing of the XL node)

In Conclusion: Just use the normal textencode, not the xl one, I'm sleep deprived and can't even remember my own workflows anymore.

I think I'm gonna head to bed now.