4-25

Based on the merge of illustrious V2, SPOlora and IterComp were added. The same usage method as V1. Due to the addition of more things, a lower CFG might be required (the test pictures appear more inclined towards cosplay).

Illustrious xl has been lacking good 3D or fantasy art models. I had to try to make my own, with the goal of solving the problem of having no detail in faces when they are too far away, as well as solving the problem of many danbooru tags not responding (noobai's dataset is so much stronger).

It makes styles easy to trigger without destroying existing concepts, of course, after two days of testing, I did find some issues like unexpected penis growth. also, you can pair cfg++ Sampler with sd Perturbed-Attention(PAG) to scale the image up to 1.5 times the size of the original resolution of sdxl.

You can use sdxl's style prompts to generate “3D Model” or “Fantasy Art” and Illustrious' quality control will work, but of course, try different samplers. You can also drag the sample image into the comfyui window to see the workflow in more detail.

Fantasy Art

"prompt": "ethereal fantasy concept art of {prompt} . magnificent, celestial, ethereal, painterly, epic, majestic, magical, fantasy art, cover art, dreamy",

"negative_prompt": "photographic, realistic, realism, 35mm film, dslr, cropped, frame, text, deformed, glitch, noise, noisy, off-center, deformed, cross-eyed, closed eyes, bad anatomy, ugly, disfigured, sloppy, duplicate, mutated, black and white"

3D Model

"prompt": "professional 3d model {prompt} . octane render, highly detailed, volumetric, dramatic lighting",

"negative_prompt": "ugly, deformed, noisy, low poly, blurry, painting"

*With too many prompts it will obsessively keep generating close-up or long-shot images, so you'll need to add composition prompts, as I have limited energy to test out which keywords this is affected by.