Same model but different quality

[hoary viper]

same prompt on the serverless stable diffusion

[deep yarrow]

are you running both in fooocus?

Both looks the same for me, whats the issue of the quality ?

[hoary viper]

yeah my bad

it's not fooocus

for the serverless

i was using the stable diffusion xl

so now i switch to the "SD Realistic Vision"

got way better results but idk why i get those white border

btw how to run foocus then without the ui and stuff

[deep yarrow]

Try to use negative prompt: borders

and use the right resolution ( width, height )

[deep yarrow]

Not sure if they support it, check their github

[hoary viper]

i checked i think community made some few thing for run pod

because with the actual version of "SD realistic vision" got multiple results like this

[deep yarrow]

Hmm yeah i see imperfections in the face

[hoary viper]

hope you see it lmao

[deep yarrow]

the eyes haha yeah

[hoary viper]

but if i use pods it wont be serverless??

https://github.com/davefojtik/RunPod-Fooocus-API/blob/Standalone/docs/standalone-guide.md i saw this what does he mean in "use 3wad/runpod-fooocus-api:0.4.0.6-standalone"

[deep yarrow]

Im not sure either, never used foocus but you can try it

[deep yarrow]

Thats the image name, and tag

You can check the template tag, create template > set to serverless

[deep yarrow]

yes like that, but i recommend you to create a template first

so you can just select it from there

[hoary viper]

ok i do this

[deep yarrow]

Yep ofc, template isnt serverless endpoints

[hoary viper]

ok i manage to run it

[hoary viper]

yes you can make templates

in serverless

[deep yarrow]

sure try to test it too