Is there a way to replicate Fooocus output in Forge UI/A1111 ?

[Uneternal]

Even when I switch off any optimizations (any styles like V2 + the offset lora) the pictures come out completely different from what I get in Forge UI. Is there a way to somehow get the same output?

[mashb1t]

There currently is no such possibility, as they use different backends afaik. Forge uses A1111 and Fooocus uses ldm_patched.

[R4290N]

please help me.
I used Focus, overlaying SD1.5 as a refiner on SDXL models.
How can I repeat this process in Forge?
All I could think of is to put on SDXL model generation 12 steps and enable HiResFix with scale factor =1 and including there SD1.5 model with denoising about 0.75.
The result is somewhat similar to what I want, but I'm still not very satisfied.
please give me your advice.

[mashb1t]

you can enable refiner also in txt2img and choose a SD1.5 model, no hires fix needed

[barepixels]

someone else said "Different implementations of scheduler , latent, etc. No possible to get same results."

[MohammadAboulEla]

could you help me understand hat is ldm_patched?

[mashb1t]

https://github.com/lllyasviel/Fooocus/tree/main/ldm_patched is the library of Fooocus used for the Stable Diffusion image generation pipeline, model management, resource optimizations etc.