RemasterizeXL merged model based on RemasterXL+1.2.1 and animagineXL 3.1, can generate semi-realistic and non-realistic style images.
Thanks to cagliostrolab (Cagliostro Research Lab) (huggingface.co)
Animagine XL V3.1 - v3.1 | Stable Diffusion Checkpoint | Civitai model to generate anime images and characters.
I realized there's no option for different licenses for different versions in civitAI, so I renewed the RemasterXL+1.2.1(1.2.2 needs different license). And the prompting method is different from animagineXL 3.0 due to 3.1's renewal prompting method.
License
Please read Animagine XL V3.1 - v3.1 | Stable Diffusion Checkpoint | Civitai
This model follows same license of AnimagineXL
Prompting tips
Please read Animagine XL V3.1 - v3.1 | Stable Diffusion Checkpoint | Civitai
I highly recommend 'realistic' in positivie prompt if you want semi-realisitic images.
absurdres, best quality, masterpiece, very aesthetic makes better images, but you have to note that they may generate nsfw images. So please input nsfw in negative, if you don't want them. Also, general in positive may be helpful to prevent nsfw things.
Other tips
Consider comfyUI, If the model generates noise
Sample images parameter
Steps: 30, Sampler: Euler a, CFG scale: 9, Seed: 2044721703, Size: 1024x1536, Model hash: a4398235c9, Model: RemasterizeXL, VAE hash: 63aeecb90f, VAE: sdxl_vae.safetensors, ENSD: 31337, Version: v1.8.0
Recipes
remaster+1.2.1 x (1-alpha) + animagine-xl-3.1 x alpha(0.4,0.0,0.0,0.3,0.4,0.4,0.4,0.4,0.4,0.0,0.4,0.0,0.4,0.4,0.4,0.4,0.4,0.3,0.0,0.0,0)_cosineA