LoCon trained on @dragonsitoo3o's (deleted) artstyle for Illustrious models
Images without a model tagged are done with cocoStyleNoobaiXL_v70, a model which is no longer on CivitAI
General usage Notes:
For more specific issues check the "about this version" section of the specific model.
Thus far my most successful permanent positive/negative combo is:
masterpiece, best quality, good quality lowres, worst quality, low quality, oldest, artist name, signature, patreonThis likely changes depending on settings and merges, but I've not needed to modify them very much.
Strength depends on merge lora compatibility and what you're trying to do, generally:
Base Illustrious: Not recommended
Compatible merges: 1
Poorly Compatible merges: ~.6
This is not my finest training and I find that I need to do a lot of cherry picking, but I like this style so ¯\_(ツ)_/¯.
Issues of particular note:
The eyes tend to be inconsistent or heavily artifacted
Fine details have more of an AI/Unnatural appearance than normal
The closer you prompt the bg to the artist's images (simple/lined/polka dot backgrounds, borders) the more likely you are to get little doodads around the subject (hearts, motion lines, notice lines) or outlines
Occasionally may add an anklet or a choker to the character.
The sketch mannequin on top of the character is part of the style.
General Workflow for Sample Images (on ComfyUI):
Sampler from images is done with nothing special
No special scheduler is used
Conditioning is split (A1111 BREAK) into 3 parts, generally BG and camera/Subject Face and attributes/Subject body and clothing
Initial resolution is selected from the SDXL whitepaper appendix I: https://arxiv.org/pdf/2307.01952
Image is upscaled by 1.5x using Lanczos resampling
Second sample pass is the same, but at around 15 steps at .4 denoise
A facedetailer is used after the upscale
Description
Initial version