HL's AI Styles - NoobAI-XL | WAI/Illustrious - Tenebriflora-Illustrious
推奨プロンプト
solo, best quality, masterpiece, amazing quality, tfaix, monochrome, sketch, absurdres, black eyes, newest, bags under eyes, unfinished, greyscale, messy lineart, simple background
推奨ネガティブプロンプト
bad anatomy, watermark
推奨パラメータ
samplers
steps
cfg
clip skip
resolution
vae
other models
推奨ハイレゾパラメータ
upscaler
upscale
denoising strength
ヒント
These LoRAs are experimental and may have issues but are worth trying.
More information about each LoRA version will be provided in the 'About this version' section.
Have fun <3
バージョンのハイライト
This model was originally created for private use, I was planning to train other LoRAs based on it, but it turned out to be quite interesting itself, so I uploaded it.
Trained on the images generated by myself. All pictures were with black background, monochrome and greyscale. since I use this model ONLY for sketch style, I will describe how to work with it:
The tags ‘greyscale, monochrome, simple background, black background’ will allow you to get what you see in my examples. You can replace the background with any other color, it worked for me without any problems.
To make the style more ‘sketchy’ I suggest using the following tags ‘sketch, unfinished, messy lineart’ it gives sketchy and little bit more cleaner look.
Eyes. I'll divide the tags into three levels.
First: Add nothing, get slightly thick lashes and small bags under the eyes.
Second: Add ‘black eyes, bags under eyes’ Now this is similar to what you see in my examples.
Third: To stylize the eyes even more, I recommend adding ‘thick eyelashes, (black eyeshadow:0.8), big eyes’.
The weight on any of the above tags can be changed to your liking.
Artists tags work well with styling as well. So try it out and experiment.
For ‘colored style’ I can't say anything yet, as I haven't tried and experimented with it. If any information comes up, I will add it here.
Additional information:
MODEL IS EXPERIMENTAL so don't expect much.
Trained on the WAI-NSFW-illustrious-SDXL v9
Weight is around 0.65-0.8. I use 0.7.
You can try to generate on bigger resolution, so lines can be thinner and cleaner. My current workflow is 1024x1460 x1.25 upscaling. But I still experimenting with that.
This LoRA series (if I get back to it at all) is experimental. The entire training dataset is generated by me personally using artists tags and LoRAs I've created before. I'll be posting the styles I like here. Perhaps. Maybe.
DISCLAIMER.
These LoRAs are experimental and may have issues. However, they are reasonably workable, so you can give them a try.
More information about each LoRA will be in the ‘About this version’ section.
Have fun <3
モデル詳細
モデルタイプ
ベースモデル
モデルバージョン
モデルハッシュ
学習済みワード
ディスカッション
コメントを残すには log in してください。












