<pre><code>Khadija Al-Harthi is an original fictional adult Omani character LoRA for Krea 2. She is traditional dhow sailmaker and modern marine-canvas engineer, built around a distinct human, professional, and visual lane rather than generic glamour.

Her identity is rich warm-brown skin with natural texture, long elegant oval face, high broad cheekbones, deep dark-brown almond eyes, strong gently arched brows, refined aquiline nose, full defined lips, one tiny beauty mark near the anatomical right corner of her mouth, and thick black hair gathered into a heavy low braid. Her body anchor is adult 5-foot-9 statuesque strong hourglass with broad capable shoulders, powerful arms and forearms, a very full natural bust with realistic weight, defined waist, substantial hips, long powerful thighs, and strong calves.

Her defining professional styling begins with fully opaque fitted deep-teal high-collar work tunic with rolled sleeves, high-waisted sand wide-leg trousers, dark leather sailmaker waist apron, one reinforced palm guard on her right hand, and closed brown marine-work boots, and her core environment begins with empty modern dhow sail loft beside a bright coastal window with one large folded white sail panel, long cutting tables, organized waxed thread, brass grommets, and no boat crew, tourist, costume set, readable text, logo, brand, open sea hazard, or fantasy pirate imagery. The validated adapter also supports modern fashion, lifestyle, movement, formal, night, profile, and out-of-role scenes without losing the woman.

The approved twenty-frame trained gallery covers close identity, full-body role, waist-up, seated, front and rear three-quarter, exact profile, walking, dynamic movement, rain-lit night, cool light, studio editorial, and occupation-free lifestyle compositions. Seven frames preserve professional and role-adjacent behavior; thirteen neutral, lifestyle, lighting, and out-of-dataset frames prove flexibility.

**Trigger:** `khadija_al_harthi_vn`

**Recommended strength:** `1.0`

**Useful range:** `1.0 tested; other strengths unverified`

For identity prompting, include:

`fictional adult Omani woman age 35, rich warm-brown skin with natural texture, long elegant oval face, high broad cheekbones, deep dark-brown almond eyes, strong gently arched brows, refined aquiline nose, full defined lips, one tiny beauty mark near the anatomical right corner of her mouth, and thick black hair gathered into a heavy low braid, adult 5-foot-9 statuesque strong hourglass with broad capable shoulders, powerful arms and forearms, a very full natural bust with realistic weight, defined waist, substantial hips, long powerful thighs, and strong calves`

For sailmaking scenes, use one seam sample, folded sail panel, needle, or thread action and avoid crew, pirate, or costume framing.

Keep exactly one adult subject, one readable action, concrete camera distance, countable props, explicit ownership/contact, and complete required anatomy. Avoid crowds, duplicate or reflected people, overloaded hands, vague abstract actions, branded elements, readable text, real-person likenesses, franchise references, transparent clothing, and childlike styling.

The exact epoch-10 adapter passed six unseen scenes covering close identity, full-body role fidelity, exact profile, dynamic professional action, cool lighting, and out-of-dataset flexibility. The twenty-frame trained gallery then passed direct owner review for identity, clearly adult age/body, anatomy, role/lifestyle separation, scene intent, diversity, duplicate control, and gallery quality.

Khadija Al-Harthi is fictional, clearly adult, and not based on an actual person.</code></pre><p></p>

Khadija Al-Harthi | The Wind Stitcher | Character LoRA

<pre><code>Initial Krea 2 release.

The fifteen-image cloud-training package used one concept role anchor, three neutral diagnostics, six occupation-free lifestyle frames, and five professional frames. Krea cloud training produced ten checkpoints;
the exact epoch-10 adapter passed the established six-scene unseen validation. This launch gallery contains twenty new trained-adapter renders.</code></pre>

Files with SHA256: 38d6f8f6805afffc18feaa7584d1b37841cc366a7ae3de49decc53c02a543a6c