Testing AI generated content.

Using absolutereality_v1 model in stable diffusion and Latent Couple extension for the two different characters.

Prompt data of the first image:

parameters: masterpiece, best quality, beach, sea
AND two women looking at each other, midriff, small blue bikini, (massive breasts:1.8), blonde hair, hot, sexy, (cleavage:1.2)
AND two women looking at each other, midriff, small red bikini, (flat chest:1.4), brunette hair, short hair
Negative prompt: deformed hands, ugly, trees, palms, grass, rocks, mountains, hills, chairs, couches, pier, buildings, naked, earrings, dark room, Afro, saggy, glasses, morbid, extra fingers, poorly drawn hands, mutation, blurry, extra limbs, missing arms, mutated hands, long neck, mutilated, mutilated hands, poorly drawn face, deformed, cloned face, malformed limbs, missing legs, too many fingers, worst quality, low quality, monochrome, overexposure, watermark, text, bad anatomy, blue hair AND
deformed hands AND
deformed hands, breasts
Steps: 25, Sampler: Euler a, CFG scale: 11, Seed: 4267062712, Face restoration: CodeFormer, Size: 512x512, Model hash: bfea7e18e2, Model: absolutereality_v1, Version: v1.3.2

Sequences like this need consistency to feel right. Not sure what the solution is (I don't dabble personally) but I do think there's a solution out there already since I've seen sequences feel like it's the same character changing outfits.

Granted it's weird to me that hand-drawn sets which are essentially 1 primary image modified ~10 times end up getting rated so much higher than this set. Be consistent, people: if you dislike sets of near-identical images, rate all of them low (even the hand-drawn ones).

Base +6