Brought this issue up with the diffusers team and maybe someone here wants to pick it up https://github.com/huggingface/diffusers/issues/9802 Basically, only the panorama technique is implemented in the diffusers pipeline, not the multiprompt generation (and maybe not the controlnet?)...
Maybe a better technique has been developed and perhaps that is why these hasn't been implemented their (this one seemed pretty simple and useful though!)
Brought this issue up with the diffusers team and maybe someone here wants to pick it up https://github.com/huggingface/diffusers/issues/9802 Basically, only the panorama technique is implemented in the diffusers pipeline, not the multiprompt generation (and maybe not the controlnet?)...
Also, here is the existing capability, built into diffusers, listed on the huggingface page https://huggingface.co/docs/diffusers/v0.15.0/en/api/pipelines/stable_diffusion/panorama
Maybe a better technique has been developed and perhaps that is why these hasn't been implemented their (this one seemed pretty simple and useful though!)