Type | |
Stats | 2,054 3,388 |
Reviews | (353) |
Published | Jun 24, 2023 |
Base Model | |
Hash | AutoV2 E47CB0D271 |
Another semi-realistic merge for generating art and illustrations. Although the mix is more realistic than the anime, it can work in a wide variety of styles. Priority is given to generating nsfw content, with more hentai shapes but more realistic textures and detailed genitals. I tried to make this mix simple and functional. So you can use quite short prompts and negatives. Also works well with many Lora
✧
✧
✧
✧
Example of suggested settings
Prompt
(masterpiece,best quality),HDR,high-definition,[[Intricate Details]],cinematic,photorealistic, then you describe, for example1girl, curvy, sexy pose, fantasy beautiful background
Negative
cartoon, sketches, (worst quality, low quality:1.3), ((monochrome)), ((grayscale)), here you add the tokens that you do not want to see in the image
You can also use negative embedding like bad-hands-5, EasyNegative, bad-artist-anime, bad_prompt_version2, bad-picture-chill-75v, etc.
Recommended samplers DPM++ SDE Karras and DPM++ 2M Karras, number of steps 20+
The CFG scale is usually 7 or 8, but you can lower it if you want the AI to add more creativity or raise it to follow the guide more closely
Size: 600x900, a couple more good choices are 480x768, 480x832, 500x800, 600x960 and
Clip skip: 2 for more art and illustration style
I recommend using highres.fix upscaler: Latent (nearest-exact)
Important: In all examples, I use the adetailer extension
I will be glad to see your art generated on this model (* ^ ω ^)