FR: request random model by baseline
Opened this issue · 0 comments
Efreak commented
Currently you can specify an entirely random model in a request, but sd15 can't do 1024x1024 without hires-fix and SDXL/cascade can't do lower resolutions or hires-fix.
Please add special cases for each baseline model (sd15, sd2x, SDXL, cascade) for requests that randomize the model within those limits. This way you can set your request to one of the supported baselines to allow random models but ensure you get a good generation. This can be using the baseline names (below) as used in the model reference or ALL SDXL
etc as used in the bridgeData.yaml.
Current baselines are stable diffusion 1
, stable diffusion 2
, stable_diffusion_xl
, and stable_cascade
.