Please check out my Ko-Fi page if you are interested in supporting me.
LoRA commissions open! -> Fiverr
This is a LoHa LoRA! You need the extension! Why? They do a lot better with multi-concepts.
I honestly don't know if v1.20 or v2.20 is better, v2 have more tags for control tho.
Trigger words for V1.20:
<lora:CheekBulgev120:1>(cheek bulge, fellatio, penis, 1girl, 1boy)
You can try "deepthroat", "irrumatio", ":>=", "deepthroat", "cum in mouth", "saliva", "pov", "from above", "from side" (or no camera angle at all for some kind of front/side angle).
V2.20 also have on top of V1.20 tags: "head tilt", "finger to cheek:1.4", "penis grab", "lipstick on penis" and tagged for penis sizes but kinda just doesn't work.
I'm so happy with this that I couldn't stop generating with it!
Follow the steps in the Model Version notes to reproduce my images.
Sample images generated from Grapefruit.
All negative embeds should be easy to find. and Prompts should be included in the sample images! Click the little icon in the bottom right to see it.
You might want to add some negatives, for example:
It was trained on 326 512x512 images scraped from the internet and 20 epochs using animefull_final_pruned.
Danbooru tags generated by Waifu Diffusion 1.4 Tagger and fine tuned for specific tags.
Description
Id say use v1 but up to you.
Generate using a weight of 0.4-1 like this: <lora:CheekBulgev220:1>
txt2img: DPM++ SDE Karras with 20 steps,512x512(can go up to 768 for wide or tall), Restore faces on (using CodeFormer weight 1.0 in Settings), Hires-fix on, to 1024x1024 (double original size) with your preferred upscaler and Denoise 0.5, CFG 7. Then send to img2img!
img2img: DPM++ SED Karras with 20 steps, Restore faces on (CodeFormer weight 1), rescale to 1536x1536, CFG 7, Denoise between 0.3-0.7 (based on how much you want to improve the image), SAME PROMPT! Send to extras when happy with img2img/inpaint results.
extras: Scale to x3 (final will be 4608x4608, I suggest going lower, I did x3 just for the samples here) Upscaler 1 - FatalAnime 4x, Upscaler 2 - SwinIR 4x with 0.10 visibility, GFPGAN 0.10 visibility, CodeFormer 0.10 visibility.
FAQ
Comments (8)
Which version of Grapefruit?
4.1 pretty sure.
what are the negative embeddings, they look renamed.
They are, but you can still find them.
trying to copy this image to see if i have the settings right but its no were even close i even copied the data an placed in to a notepad i am missing a thing called fatalanime 4* any help to understand
FatalAnime 4x is just an upscaler. Heres the link: https://de-next.owncube.com/index.php/s/x99pKzS7TNaErrC
Are you sure your LoCon/LyCoris extension is working?
@Part_LoRAs i think so it some what works but no where close to what the image should be she just has a mad face and holding it nothing more the back ground is the same but not what the image should be. thank you by the way
@SamHat Trying to recreate an image is a waste of time anyway.
If it works, it works.
Its just impossible to recreate the same image.
Too many variables! Even different GPUs can affect it making the output different.