LoCon trained on Noxnox (@Noxintwice)'s artstyle for Illustrious models<h2 id="general-usage-notes:-kdzsuqx3a">General usage Notes:</h2>For more specific issues check the "about this version" section of the specific model.Thus far my most successful permanent positive/negative combo is:<pre><code>masterpiece, best quality, good quality </code></pre><pre><code>lowres, worst quality, low quality, oldest, jpeg artifacts, blurred, artist name, signature</code></pre>This likely changes depending on settings and merges, but I've not needed to modify them.This model has two distinct styles:<ul><li>General style appears to be more like his more recent works</li><li>"Flat color, no lineart" appears to be his older style that uses significantly less linework</li></ul>Strength depends on merge lora compatibility and what you're trying to do, generally:<ul><li>I haven't tested base Illustrious</li><li>On very compatible merges alone I recommend 1</li><li>On poorly compatible merges I don't recommend going over .7</li><li>When mixing in a style .8 or less strength is good (depending on the style used) but this is to your discretion</li></ul>Images marked with "Noobai v.75" are actually V11<h2 id="general-workflow-for-sample-images-(on-comfyui):-kdr873uqq">General Workflow for Sample Images (on ComfyUI):</h2><ul><li>Sampler from images is done with Restarts with segments of [3,2,0.06,0.30], which are values from the paper on Restarts<ul><li>Note: Using restarts tends to smooth out the image more but makes it more coherent, your mileage may vary and it's up to you if you want to use them</li></ul></li><li>No special scheduler is used</li><li>Conditioning is split (A1111 BREAK) into 3 parts, generally BG and camera/Subject Face and attributes/Subject body and clothing</li><li>Initial resolution is selected from the SDXL whitepaper appendix I: <a target="_blank" rel="ugc" href="https://arxiv.org/pdf/2307.01952">https://arxiv.org/pdf/2307.01952</a></li><li>Image is upscaled by 1.5x using Lanczos resampling</li><li>Second sample pass is the same, but at around 15 steps at .3 denoise</li><li>A facedetailer is used after the upscale</li></ul>

Noxnox (Noxintwice) - LyCORIS LoCon - Illustrious

Initial versionSeems to occasionally add sweatFine details tend to artefactBase style tends to look a bit generic on stricter mergesIf using simple backgrounds it will occasionally add borders

NN-MAnon-XLILLUSV8.safetensors

Noxintwice - Style

masterpiece, best quality, good quality  

Files with SHA256: 700d29b29e35471a3dc98202dfc4477c54e1c88e51e931420f887d2228979751