This model was trained on TheSabu's artstyle/s. There is two versions of this LoRA, A1 and A2. I think setting the weight of the LoRA to 0.6-1.2 works fine, they're pretty flexible.<ul><li>A1 was trained both on their cartoony &amp; realistic art.</li><li>A2 was trained on only the cartoony style.</li></ul>Both have tokens to activate them. (Not necessary but helpful)<ul><li>A1= "SabuCartoon" &amp; "SabuRealism"</li><li>A2= "SabuCartoon"</li></ul>I personally like both of them, but A1 and A2 do have their differences. A1 is usually more flexible with the style. But both can be tough to tame at times.I also would suggest using some helping tags like the following depending on which style you're going for when using A1, these are some tag templates that worked well for me. (put them at the start of your prompt, you can substract from them or add to them if you want, i also recommend changing the weights depending on how your output images look if they look bad or inaccurate)<ul><li>SabuRealism: (SabuRealism:1.5), (realistic:1.3), (realism, hyperrealism:1.3), (oil painting, painterly:1.3), (soft lighting, soft shading:1.2)</li><li>SabuCartoon: (SabuCartoon:1.5), (cartoon:1.5), (sketch:1.3), (lineart:1.3), (cellshading:1.1), (painterly:1.1)</li></ul><ul><li>Random Sketch Style Event: </li></ul>I'd also advise putting (monochrome, greyscale:1.3) in the negativesSample images were made using <a rel="ugc" href="https://civarchive.com/models/4451/abyssorangemix2-hardcore">AbyssOrangeMix2_Hard</a>

Files with SHA256: cacad70a9a75226739f9a5cca005e0fd41b3719ace30e075d3b0d31d2e66fd79