physical violence

weapon violence

wide hips

revealing clothes

thick thighs

downblouse

convenient censoring

huge breasts

pg-13

corpses

suggestive

oral invitation

pg13

sexy

sexual situations

male nudity

disturbing

male swimwear or underwear

female swimwear or underwear

partial nudity

undressed

female nudity

breasts out

exposed female nipple

breast out

lingerie

male underwear

hair over breasts

female swimwear

gigantic breasts

no panties

graphic violence or gore

covered nipples

huge butt

strapless leotard

sitting on face

emaciated bodies

one breast out

nsfw

female underwear

nude

graphic male nudity

adult toys

illustrated explicit nudity

nudity

graphic female nudity

hentai

futanari

porn

sexual intent

genitals

peeing

vore

oral

sexual activity

anal

blowjob

dildo riding

incest

hanging

hate symbols

nazi party

white supremacy

diapers

scat

self injury

hate speech

urine

extremist

child on child

latex clothing

swimwear

bukkake

fellatio

cumshot

implied fellatio

eat_cum

cumdrip

cum in pussy

cum on face

after fellatio

cum on hair

cum on body

cum on tongue

cum on hands

cum in mouth

triple fellatio

autofellatio

fucked silly

cum on pussy

<h2 id="heading-4120">Model details:</h2>This model was created to generate images of an OC of mine inspired from Rosen Garten.For more detail about the version please read the "version detail" section.

This is the 2nd version of the model.It overal keep its strenghts, but it now generate a consistent character as intended."Monochrome" and "grayscale" are affected by the lora, and they will "often" produce images with colored eyes and/or hairs (change the strenght to fix of those tags or use the "BREAK" keyword to control the output more consistently).The new versions all work better with colored images, and they all are good in postprocessing in the in-paint section (i particularly like the output of the in-painting on face at 0.65 strenght without changing anything on the prompt and affecting only selection.There are 5 versions:<ul><li>Soft, it's the version that less impact on the style of the images, so it is the actual recomended one if you plan to mix styles.</li><li>Mid, it's a version in the midpoint between soft and full. It impacts the style, but not too much (among those three versions is the one that required less, in-painting on my generations)</li><li>Hard, it's fine and reliable.</li><li>Full, It's a version that impact on the style of the image and that generate often "sharp" faces, like intended.</li><li>Complete, If you can't choose between the 3 previous ones, you can pick thise one. It's pretty versatile, but it requires you to play with the lora strenght.</li></ul>N.B: I in-painted only faces to increanse some details, but the results were pretty ok even without doing it.I suggest a strenght between 0.4-0.9 (it is the suggested value for each of those models), around 0.7 the results are almost always the one intended.This version has only 1 trigger word:<pre><code>Lynn</code></pre>It is not required to add: "1girl, solo", but it recomended to achieve better results.There are still keywords that are affected by the lora:<pre><code>monochrome, grayscale, holding, hair bow, bow, sword, ponytail, ribbon, hair ribbon, portrait, smile, open mouth, upper torso</code></pre>In case of issues with one of those tags you could fix it by changing the strenght of the keyword or placing a "BREAK" between the keyword and the lora in the prompt. (I noted great differences in the output using the last mentioned method, with the same seeds)(I did also some test with clipskip and it looks like it works also in clipskip 3, but ofc it's not recomended)

Lynn[V2_Complete]

Lynn[V2_Full]

Lynn[V2_Hard]

Lynn[V2_Mid]

Lynn[V2_Soft]

This version was trained using 30 images in the style of Rosen Garten, using just "monochrome" images, so it works best in generate comic-like art.Overall this version [1] is a semi-failure because it doesn't work as intended, but it can work to obtain specific results.IT NEEDS some negative embeddings like "easyNegativeV2" and some prompt:<pre><code>Lynn, monochrome</code></pre>Those 2 keywords are the ones that have the most impact on the image.Other prompts that could affect the generations are:<pre><code>hair bow, ponytail, bow, hair ribbon, ribbon, smile, holding, ponytail, sword, open mouth, portrait</code></pre>This LoRa works with every model and it's awesome in in-painting (View examples images).Experiment with it, it can give some weird outputs but also some satisfying ones, all the training images used for the realization of the [V2] were generated using this Lora (the V2 will be uploaded after some testing, the 24/10/2023).

Lynn[V1]

Lynn

00006-2329924753.png