language:
- en
tags:
- stable-diffusion
- text-to-image
- image-to-image
- diffusers
license: creativeml-openrail-m
inference: true
Hello, sorry for my lousy english.
After days of trying and retrying hundreds of times, with dozens of different versions, DucHaitenAIart finally released the official version.
Improved image sharpness, more realistic lighting correction, more shooting angles, the only downside is that it's less flexible and less random than beta-v6.0, so I'm still will leave beta-v6.0 for anyone to download.
This model can create NSFW images but since it is not a hentai and porn model, anything really hardcore will be difficult to create. But, To make the model work better with NSFW images, add “hentai, porn, rule 34” to the prompt
Always add to the prompt “masterpiece, best quality, 1girl or 1boy, realistic, anime or cartoon (it's two different styles, but I personally prefer anime), 3D, pixar, (add “pin-up”) ” if you are going to give your character a sexy pose), highly detail eyes, perfect eyes, both eyes are the same, (if you don't want to draw eyes, don't add them), smooth, perfect face, hd, 2k, 4k , 8k, 16k
Add to the prompt: “extremely detailed 8K, high resolution, ultra quality” to further enhance the image quality, but it may weaken the AI's interest in other keywords.
You can add “glare, Iridescent, Global illumination, real hair movement, realistic light, realistic shadow” to the prompt to create a better lighting effect, but the image will then become too realistic, if you don't want to. Please adjust it accordingly.
When using models to create images, most of your results will fall into one of three cases: realistic photo, realistic 3D and 3D anime. Follow the instructions below to adjust the steps and CFG scale for best results.
Sampler: Euler a
realistic photo: 50 steps - CFG scale 22.5 (not sure because not enough data)
realistic 3D: 46 steps - CFG scale 17
3D Anime: 42 steps - CFG scale 12.5
negative prompt I used for the sample image:
lowres, disfigured, ostentatious, ugly, oversaturated, grain, low resolution, disfigured, blurry, bad anatomy, disfigured, poorly drawn face, mutant, mutated, extra limb, ugly, poorly drawn hands, missing limbs, blurred, floating limbs, disjointed limbs, deformed hands, blurred, out of focus, long neck, long body, ugly, disgusting, bad drawing, childish, cut off cropped, distorted, imperfect, surreal, bad hands, text, error, extra digit, fewer digits, cropped , worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name, Lots of hands, extra limbs, extra fingers, conjoined fingers, deformed fingers, old, ugly eyes, imperfect eyes, skewed eyes , unnatural face, stiff face, stiff body, unbalanced body, unnatural body, lacking body, details are not clear, details are sticky, details are low, distorted details, ugly hands, imperfect hands, (mutated hands and fingers:1.5), (long body :1.3), (mutation, poorly drawn :1.2) bad hands, fused ha nd, missing hand, disappearing arms, disappearing thigh, disappearing calf, disappearing legs, ui, missing fingers
Note 1: “realistic, 3D, anime, pixar” is required in the prompt to create beautiful images, unless you want to explore something new.
Note 2: negative prompt is extremely important, it accounts for 50% of the output of AI, so really pay attention to it, or if you are too lazy, you can take my negative prompt to use for most portrait images.
Note 3: all the instructions I wrote above are not absolute, certainly not the best, if you just follow what I wrote, you will not be able to fully explore the capabilities of the DucHaitenAIart model. Explore, be creative, have a variety of styles and share the results together. That's really what I want.
please support me by becoming a patron http://patreon.com/user?u=83557525
Some test:
