Fix an image with GPT Image 2

Correct the anatomical, design, and physical errors in

[ A @image] while keeping it in its original painted concept-art
style.

FIX (correct these, and paint the corrections in the same style as
the surrounding image):

Anatomy — hands, fingers, limbs, joints, and proportions; correct
count, foreshortening, and anything physically wrong. Fix facial
features, eyes, and symmetry where broken.

Design logic — clothing, armor, straps, buckles, and accessories
must connect and make sense; paired items (shoes, gloves,
pauldrons, earrings) must match and mirror correctly.

Physicality — how garments drape, sit, and fold on the body;
contact points, weight, occlusion, and how objects attach.
Everything must obey gravity and physical sense.

PRESERVE EXACTLY (do not alter):

Rendering style — the concept-art look:
visible brushwork, soft edges, stylized (non-photographic)
rendering. No photoreal skin texture or pores.

Color palette and grade

Lighting, rim/backlight, glow, bloom, and atmospheric mood

Character identity, face, costume design, silhouette, and pose

Composition, framing, depth-of-field, and painterly background
falloff

Treat this as a correction pass, not a re-render — fix only what is
Anatomically or physically broken, and blend every fix seamlessly into the existing illustration.