<h3>She knows the maker, the decade and the patent number. Ask her.</h3>

Mara Klaver, an original fictional character: a twenty-three-year-old
software engineer and collector of police uniforms and vintage memorabilia. Warm, playful,
endlessly curious and a little too pleased with herself, with dark brown skin, long voluminous
black curls and a soft, curvy figure. Her favourite piece in the collection is a Dutch politie
uniform. Trained on Krea 2 Raw from synthetic images of a made-up character.

Trigger: <code>dxqen7 woman</code>
<ul>
 <li>The trigger gives her face, her long black curls and her figure.</li>
 <li>Write out any outfit you want. She was trained in five uniforms (Dutch politie, NYPD,
 LAPD, a sheriff's tan-and-brown with campaign hat, and a steel-gray prison guard shirt) and
 seven everyday looks, all described in plain words, so everything stays promptable.</li>
 <li>Example: <code>dxqen7 woman in a white long-sleeved police shirt with navy epaulettes
 and POLITIE sleeve patches, a navy tie, a black belt and navy trousers</code>.</li>
 <li>Expressions are promptable: neutral, focused, skeptical, concerned, frustrated,
 surprised, laughing, a confident smirk.</li>
 <li>Her curls are long and loose by default; ask for them pulled back explicitly.</li>
 <li>Add "dark brown skin" to your prompt; on plain backgrounds she can render a little
 lighter without it.</li>
 <li>Weight 0.8–1.0 (0.9 is a good default).</li>
 <li>Known quirk: "a woman with dark brown skin and long black curly hair" leans toward her
 even without the trigger, so describe a different person in full when you want someone
 else.</li>
</ul>

The dataset for this LoRA was generated with GPT Image.

All showcase images are made with this LoRA on a Krea 2 realism checkpoint, 8 steps, CFG 1.

If you like my work, leave a comment, like, or whatever! It'll motivate me to make more and share more. Thank you!

Mara Klaver - Playful Programmer and Collector

Rank 32 on Krea 2 Raw, 1600 steps (ai-toolkit, lr 3e-4, sigmoid timesteps), 36 images at
1024: two each in five police and guard uniforms, and three or four each in seven everyday
outfits (an ivory top with a navy skirt, a blue shirt with jeans, a teal wrap blouse, a plum
shirt dress, a cardigan, a hoodie and a trench coat). Shots include front, both profiles, from
behind, seated, full length, a low camera, studio backdrops, home offices, kitchens, workshops
and city streets. Captions name the outfit, pose, expression, setting and light but never her
face, skin or hair, so the trigger carries her identity. Her identity showed up by step 400 and
held in an outfit outside the dataset (a red evening dress) at every checkpoint.

Files with SHA256: 7fe73c695052fc7f7f6e825e10f338eb5c7f8f9a0666ded75ec49bb00f62d471