I used NovelAI's anime model. No specific prompt. Recursively used img2img with different prompts until I liked what I saw. That specific image you two are asking about used this web image as the base (nsfw)(
https://gelbooru.com/index.php?page=pos ... id=1968314). I manually flipped it 180 as a last step because the AI simply cannot do upside-down at all.
However, I find you can also prompt a non-vore base image yourself and do the same process. For example, I prompted this:
(((masterpiece))),best quality, illustration, full_body, (detailed face), {messy hair}, 1girl, {touhou_project}, (beautiful detailed girl), {beautiful detailed glowing eyes}, nude girl lying on pink fuzz with arms behind back, full body, on back, huge black aura
nsfw:
...and eventually img2img'd it into this:
nsfw:
Last iteration prompt was:
(((masterpiece))),best quality, illustration, straight body, (detailed face), {messy hair}, squeezed 1girl, restrained, (fetal position), (looking up), covered in sweat, looking up, detailed tentacles tunnel, nude, nipples, medium breasts, scared, open_mouth, arms_behind_back, arm_up, inside_tight_hole, dripping liquid, detailed_tentacles, (beautiful bright eyes), wet, partially_submerged, water slime, liquid dripping from above, crossed_legs,
So the AI is really flexible. The AI doesn't really understand "vore" or "internal". Instead, I find that variations of "tentacles" and "slime" are good for getting an organic appearance.