TheArtofMathew Style

TheArtofMathew is an artist who likes to draw solo pin-ups of girls in a bimbo-ish style aka(big breasts, wide hips, and a wasp-like waist). This LoRA aims to emulate that style. I really liked his style so I wanted make it a LoRA and here it is.

This is my first time making a LoRA so I forgot to remove the activation token which was the most used tag being 1girl. There's six versions here from half way trained to fully trained for you to mix in with different weights in your own generations.

Essentially if your looking for just the body shape the low version are the ones you want and if you want more of the art style go for the higher versions.

Trained on "animefull-final-pruned-fp16" at 768 resolution though generating at (512 x 768) portrait it was completely fine as shown below.
Aside from that here is a sampling of testing the model out it on different models at strength 1, same seed, prompt, with VAE:"kl-f8-anime2" but you can use other vaes like vae-ft-mse-840000-ema-pruned, etc. just fine.
Sampler I used was Euler a and DPM++ 2M Karras didn't really check the others so uh warning there.
Anything4.5-pruned:

Abyssorangmix3:

Based64mix:

Yesmix_v16:

revAnimated_v122: OooH it works with more realistic models :pog:

If you generate in a square you'll honestly get some pretty sick outcomes like this: (768x768) each section is individual not together.

Sample of prompt used for the images:
Positive:
lips, makeup, breasts, huge breasts, wide hips, thick thighs, (((best quality)), ((masterpiece)), (mature female), photorealistic, photorealistic face, cinematic light, (skindentation:1), cute, feminine, (perfect face:1.1), (perfect eyes, detailed eyes), perfect hands, curvy:1.5, (narrow waist:1.2), big eyes), (1girl:1.4, solo, solo focus, looking at viewer, dynamic pose, smile), <lora:GoodHands-vanilla:0.8>, <lora:ArtofMathewStyle-10:1>
Negative:
(worst quality, low quality:1.4),EasyNegative , bad-hands-5,((((blurry background)))),multiple views, comic, (signature:1.3),(panels:1.4),(signature, watermark:1.3),forehead mark, (greyscale,monochrome:1.3), (nipples:1.2), (plump:1.2, fat:1.2), areolae, crowd,

I recommend using ADetailer/ or anything to bring in more detail to really bring out the eyes and lips else you'll get some less desirable faces, and If you want you can leave out your outfit prompts cause the LoRA makes some really good ones by itself or it was the anything4.5-pruned model that I was mainly using for generation.
Generated without ADetailer to show how eyes come out:

P.S I was using "loraeyesv1" and "tachi-e" full body LoRA for these which was the reason the body and background is a little iffy.

Neat outfits huh? Here's some good ones I found while generating:

Anyway info dump complete now go, go have fun with this resource.
I hope you enjoy it as much as I was pleased with its outcome.
Happy Generating!!!