The team of heroines has been defeated by the evil tentacle monster and fucked into submission! Now they're captured in his fleshy lair and repeatedly impregnated and birth tentacle monsters!

This LoRA is based on art of that scenarios, primarily by numeko, allenes, and blackcony, though no one artist is more that 25% of the training data.

Portrait works for 2 characters, for 3+ I recommend using landscape.

(I don't condone violence against real women nor treating real women as walking incubators, this is and my other LoRAs are just hentai/porn.)

MOSS_a1: Trained on MOSS-Image Base, but works on sft. Main trigger is "tentacle breeding facility". Optional trigger is "embedded in a meatwall".

Trained on manual-captions + curated auto-tags, so things like "[x] women in the foreground and [y] women in the background" should work. Expect a style impact because I forgot to add artstyle descriptions to the captions.

Krea2 a1: Experiment with training on Krea2. Trigger phrase is "in a tentacle breeding facility", can be used within a larger sentence. If you want the characters in a row you have to specify "in a row". Also impacts huge/giant breasts, pregnant, and ahegao.

A bit prone to drenching everything in slime/cum.

Also seems to work for improving tentacle pits even without the main trigger.

Anima v1: Trained on Anima preview 3, same training data as the previous versions. Picked the epoch that I think best balanced concept adherence and style impact. Works on Anima Base v1, but needs a few extra negatives to remove the censor blur that was on most of the training data.

Main trigger: ttclbrdfclt, tentacles, uncensored

Optional: meatwall, pregnant, cum bath

Negative: censored, mosaic censor

IL_v1_D4: Massively reduced file size. Slight adjustment to trigger words.

IL_v1: Has as little style impact as I could manage while keeping good concept adherence.

Main trigger: ttclbrdfclt, tentacles

Optional triggers: meatwall (embeds the characters in fleshy walls), restrained, pregnant, multiple girls (sometimes adds background characters)

Noob_v1: Same training data and triggers as the Illustrious version, just trained against NoobAI vPred 1.0.