Bad Person

Here's what a bad person looks like, according to AI models.

Stable Diffusion 1.5

prompt: "english text, parody, text focus, solo, no humans, meme, 1boy, simple background, male focus, poster (medium), monochrome, cover, border, realistic, shirt, copyright name, looking at viewer, fake cover, blurry, balanced colors, closed mouth, artist name"

Stable Diffusion XL

prompt: "monochrome, 1boy, greyscale, male focus, solo, gray tint, indoors, traditional media, realistic, 1girl, old, short hair, no humans, standing, old man, closed mouth, bald, facial hair, balanced colors, sketch, sitting, simple background"

Flux.1

prompt: "solo, dark, realistic, upper body, looking at viewer, photorealistic, black hair, 1girl, 1boy, male focus, lips, jacket, long hair, blurry, hood, closed mouth, balanced colors, messy hair, shirt, hood up, horror (theme), blurry background, facial hair, long sleeves, monochrome, brown eyes, hoodie, short hair, black jacket"

Mode

Identifies the prevailing subject portrayal, based on the most characteristic combination of an image's core descriptive labels.

Stable Diffusion XL

prompt: "monochrome, 1boy, greyscale, solo, realistic, 1girl, no humans, closed mouth, balanced colors, sitting, hood, outdoors, closed eyes, robe, 1other, ambiguous gender, hood up, blurry, cloak, grey theme, own hands together, hooded cloak, sand, water, rock, hooded robe, beach, black robe, orb, ocean, shore"

Flux.1

prompt: "solo, dark, realistic, upper body, looking at viewer, photorealistic, black hair, 1girl, 1boy, male focus, lips, jacket, long hair, blurry, closed mouth, messy hair, blurry background, long sleeves, monochrome, greyscale, nose, artist name, asian, depth of field, gray tint, from side, against wall, crossed arms"

Frequency

The most prominent individual labels, reflecting the model's typical descriptive focus for this subject.

Stable Diffusion 1.5

prompt: "english text, parody, solo, meme, 1boy, male focus, border, shirt, balanced colors, 1girl, black border, standing, open mouth, speech bubble, 1other, what, sweater, green shirt, barefoot, horror (theme), frog, monster, skirt, tongue, !, against wall, crack, cracked wall, purple skirt, spoken exclamation mark, stuck"

Stable Diffusion 1.5
Stable Diffusion XL
Flux.1

Composite

Forms a visual mean of the subjects, constructed by aligning facial features across images and then blending their pixels.