ABlendedCheese

<!-- Provide a quick summary of what the model is/does. --> ABlendedCheese is the merged diffusion model which intended to produce bit realistic anime style pictures. I have merged ACertainThing, ElysiumAnimeV2, EimisAnimeV1 and some other models (include Gape) in a complicated way.<br /> This may be one of many mediocre models, but I can't bear to throw it away, so I'm uploading it here.<br /> <img src="https://i.imgur.com/3nqGGmM.jpg" />

Samples

<img src="https://i.imgur.com/8PewLN3.jpg" />

Reference Useage

• SD VAE: None(Waif vae is already baked in)
• Clip skip: 1
• Sampling Steps: 20
• Sampling method: DPM++ 2M Karras
• Use hires.fix
• Upscaler:<br /> R-ESRGAIN 4x+Anime6B to keep its taste. (Hires Steps: 10, Denoising strength: 0.50-0.55)<br /> Latent to enrich details. (Hires Steps: 20, Denoising strength: 0.55-0.60)
• Negative prompt start with: nsfw, (worst quality, low quality:1.4)