A realism-focused companion LoRA for the Z-Image base.

Trained to improve natural human anatomy, expressive body poses, and believable weight distribution.

The model captures subtle cloth behavior — tension, stretch, and natural drape — while preserving stable proportions and clean hand structures.

It also handles multi-character scenes reliably, maintaining spatial coherence and balanced composition.

Designed to produce images that feel physical, grounded, and naturally staged.

Particularly effective for portrait studies, physical poses, and multi-figure compositions.