Ishani Perera is a fictional adult Sri Lankan character LoRA for Krea 2.

**Role:** luxury sleeper-train conductor and service director.

Core identity anchor:

rich golden-brown skin, soft oval face, high cheekbones, dark-amber almond eyes, strong gently curved black brows, long straight nose, full lips, one deep dimple on the anatomical left cheek, long glossy black hair arranged in one low braid secured with an  oxblood ribbon.

Core body anchor:

slender 5-foot-8 adult pear-shaped frame, narrow shoulders, long torso, narrow waist, gently rounded hips, long legs and a natural medium bust.

Core wardrobe lane:

original oxblood tailored conductor jacket over an opaque cream blouse, high-waisted navy midi skirt, slim brass belt, dark low heels and one unbranded round brass pin.

The approved upload package supports strong single-subject compositions with the trained identity, role, and silhouette across scene-diverse editorial environments.

Supported scene families:

- identity anchor portraits and full-body references;
- role-grounded action frames;
- controlled indoor and outdoor editorial environments;
- seated, standing, walking, and three-quarter body compositions.

Wardrobe support:

- the trained signature lane centered on `original oxblood tailored conductor jacket over an opaque cream blouse, high-waisted navy midi skirt, slim brass belt, dark low heels and one unbranded round brass pin` with safe, opaque styling continuity.

Prop and hand guidance:

- keep props role-consistent, one primary prop family per frame;
- keep both hands visible whenever the shot allows;
- avoid overloaded hands, mirrored duplicates, and multi-subject staging.

Composition guidance:

- strongest reliability is single-subject waist-up, three-quarter, and full-body framing;
- specify camera distance and subject action explicitly for stable pose/control.

**Trigger:** `ishani_perera_vn`

**Recommended strength:** `1.0`

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

For best consistency, keep prompts explicit about one subject, camera distance, role action, concrete environment nouns, and visible hand ownership of props.

This model is a fictional adult character and is not based on an actual person.