Introduction
WillowbarkXL is a merged SDXL model created with the aim of generating "2.5D" style (semi-realistic) anime art.
How to use
Sampler: DPM++ 2M Karras or DPM++ SDE Karras
Resolution: At least 1024 x 1024
CFG scale: Between 6 and 8. Try a higher CFG if you are having trouble with certain scenes.
Prompt: Put realistic, as the first term, then describe the scene you want
If you are looking to create a specific character (i.e. from Blue Archive), the first term must be the character's name, followed by realistic.
For instance,
Ichinose Asuna \(blue_archive\), realistic, lying on bed on her back, looking at viewer, halo over her head
Negatives: worst quality, low quality
Clip skip: 1 or 2
Recipe
Unet merge
Model A: AnimagineXL_V3
Model B: colossusprojectXLSFW_v5.3
1,1,1,1,1,1,1,0,0,1,1,1,0,0,0,0,1,1,1,1,1,1,1,1,1
base_alpha: 0
Note: It has come to my attention that I have accidentally used the merge option for SD 1.x, rather than SDXL. This means I have inadvertently merged the layers I had not originally intended. If you are looking to replicate my model, please use the merge-block-weighted extension as I have done.