Lora model for the Shirt Cut Meme, a risque template showing off a character in various states of clothing. Follows the template decently, but the 'free space' can be a bit spotty. However, it can be influenced with specific tags. The tags highlighted in red are the tags I used to influence the 'free space' box. Feel free to experiment and hopefully share your findings for other tags that can be used.

Example Template: score_9, score_8_up, score_7_up, shirt cut meme, [character description], [clothing], adjusting clothes, blush, cleavage cutout, clothes lift, clothing cutout, covering chest, embarrassed, english text, hand up, looking at viewer, looking away, midriff, navel, nude, nude cover, open clothing, profile, sideboob, simple background, underboob, white background, <lora:ShirtCutMemePonyXL:1>

Adding nipples makes covering chest less effective. Template should be tweaked accordingly with the derivative model quality tags. Highly recommend Adetailer, though it was not used in the example images.

I've also included the watermark @druzsea in one of the example pics in case the A.I. generates it incorrectly as it would be greatly appreciated if credit could still be given to the artist who made this template when images are shared outside of Civitai.

Weight of 0.9 to 1.1 for this one. Recommend 2 other Loras (character and style) max, but you can try messing with the weight if you add more.

Trigger Word is 'shirt cut meme'

I've also started a patreon with a currently ongoing public poll for what my next Lora will be, tips are appreciated, but even voting in the poll is much appreciated. https://www.patreon.com/ArtificalisAnima

Models Tests from the ones I had on hand and observations:

AutismMix: Works almost as well as Base PonyXL

AnimeboysXLV3: Bit finnicky, add multiple views to reinforce, about 20% less effective than autismmix? Tends to mess up the text far more than the other models I've tested.

aCrystalpony: Decent, messes with the boxes a bit and tends to delete the 'free space' box or add an extra one, about 10% less effective than autismmix

ZonkeyV3: Looks more like a realistic painting than actual photo realism, works as well as autismmix though. Struggles a bit with furries, tends to make fur color more natural and intersperses humans in boxes. Recommend increasing furry and color tag weights.