This would be such a nice feature but... it does not work at all.

[maniac-0s]

I found this via a youtube video. I have read different reddit posts about this. But even when I copy and paste the exact same settings, all I get is a merged person (yes I clicked apply each time, yes the preview shows different regions). After generating like 100s of pictures in 512x512 AND 768x512, trying to tweak the settings, I gave up.

Is there some thing I am missing completely?

https://www.reddit.com/r/StableDiffusion/comments/11i4etk/basic_guide_6_latent_couple_how_to_get_what_you/

https://www.reddit.com/r/StableDiffusion/comments/11hplx1/does_latent_couple_actually_work_with/jax57hg/

https://www.reddit.com/r/StableDiffusion/comments/11hjjgz/composition_using_controlnet_latent_couple_no/

[frank-wirth]

have the same problem: i always get 1 mixed person

[FemBoxbrawl]

It only works on stable diffusion 1.6