Note
It is also available on pixai for you to try.
No quality tag & negative prompt required. Of course, quality tags would improve results.
If noise occurs, reduce the CFG scale gradually or use Hi-Res to eliminate the noise.
The order of the tags should be the same as illustrious-XL and should be fine.
Rating tags have probably lost their functionality. This is not my fault.
The corresponding LoRAs remain the same, those that were available can be used, and those that are not cannot be used.Sample images are produced with the following settings.
Sampling method : Euler a / Schedule type : Simple / CFG Scale : 5 / Sampling steps : 20
ByteDance Hyper-SDXL - 8 Steps This allows for fast generation with the following settings.
Sampling method : Euler a / CFG Scale : 1 / Sampling steps : 10
merge recipe
illusob-xl-merge-base-model = Illustrious XL personal merge v3.0_noob-1.0based
+ (( Community illustrious type Model - noobaiXLNAIXL_epsilonPred05Version )
x 1,1,1,1,1,1,1,1,1,1,0,0,0,0,1,1,1,1,1,1 )
Difference LoRA A = Community pony type Model - ponyDiffusionV6XL_v6StartWithThisOne
Difference LoRA B = Community animagine type Model - animagineXLV31_v31
This model = illusob-xl-merge-base-model
+ Difference LoRA A ( 0,0,0,0,0,0.4,0.4,0.4,0.2,0,0,0 )
+ Difference LoRA B ( 0,0,0,0,0,0.4,0.4,0.4,0.2,0,0,0 )
+ self-made illustrious LoRA A ( 0,0,0,0,0,0.9,0.9,0.9,0.5,0,0,0 )
+ self-made illustrious LoRA B ( 0,0,0,0,0,0.3,0.3,0.3,0.3,0,0,0 )
+ self-made Pony LoRA A ( 0,1,1,1,1,1.2,1.2,1.2,1,0,0,0 )








