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.
Trigger words for V1.20:
<lora:BreastFeedingv120:1>(breast sucking, mature female, 2girls, yuri, lactation)
<lora:BreastFeedingv120:1>(breast sucking, nursing handjob, mature female, mature male, penis, testicles, lactation)
You can try "yuri" or "hetero" to reinforce that it should be 2 characters. "1girl, 1boy" etc.
Follow the steps in the Model Version notes to reproduce my images.
Sample images generated from Nothing model.
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.
It was trained on 219 512x512 images scraped from the internet and 20 epochs using animefull_final.
Danbooru tags generated by Waifu Diffusion 1.4 Tagger and fine tuned for specific tags.
Description
Generate using a weight of 0.4-1 like this: <lora:breastFeedingv120:0.8>
How I archived my generations:
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 (12)
can u make it into a lora?
It probably wont work as a Lora... I've started doing LoHas cus Loras are quite bad at multi concepts
@Part_LoRAs lohas locons and locryis completely break stable diffusion on my computer
sorry this isnt a review telling u if the loha has flaws but i hope its great
The successful ones work well, although some images are distorted. I've tried making this pose and your work is great in comparison.
I noticed that you made a lot of lohas. Before that, I always thought that loha is only good at extracting style. Is it better in making concepts and poses? I'm used to making locon now, thanks a lot for sharing some experience if you'd like
Nah, I just do w/e with it. Works for anything
Distorted images are mostly the nursing handjob ones, right?
Cus its REALLY difficult to find good images for it. Most have like almost the entire body of the guy cropped out and thats one of the reasons for the LoRA to struggle with it.
@Part_LoRAs Yes, the part about handjob is relatively easy to get wrong.
Do you mean that loha can generate human body more stably than lora?
@Saya Honestly, your dataset is 99% of the problems in the LoRA
I just prefer doing LoHas and thats it. Theres no better one
There is a problem with your lora. The same tag, model, sampler, and number of steps cannot generate pictures similar to yours above, more similar to blowjob and titjob.
You have the extension installed and working?