


cinematic ultra realistic portrait, 8k sharp details. A 25-year-old East Asian woman as a Dunhuang flying apsara (feitian) mural dancer, elegant oval face with classical Tang-dynasty beauty features, long black hair in an elaborate high bun adorned with gold hairpins and pearl tassels, subtle classical makeup with red lips, skin with visible fine texture and pores, no plastic smoothing, serene compassionate gaze slightly upward. She wears flowing silk robes in layered crimson, amber and teal, wide sleeves billowing as if dancing mid-air, intricate gold-embroidered cloud patterns on the hem, a long translucent silk ribbon (pibo) swirling around her arms. Body arched gracefully in a classical dance pose, one arm raised holding the ribbon, the other extended downward, bare feet pointed. Foreground: faint drifting incense smoke and floating petals; midground: the dancer suspended above a Mogao cave interior floor; background: ancient mural-covered grotto walls with faded thousand-buddha motifs and a stone stupa silhouette. Warm volumetric light streaming from a cave opening on the left, 4000K golden glow mixed with cool ambient blue from the murals, dust motes glowing in light shafts. Slightly low-angle medium shot, 50mm lens, shallow depth of field, rule of thirds. Rich mineral-pigment color grading, cinnabar red, malachite green, lapis blue, subtle film grain, Eastmancolor vintage feel. Sacred, ethereal, timeless mood. Photorealistic style fused with classical Chinese mural aesthetics, sharp focus, high detail. No oil painting, no anime, no 3D render, no deformed hands, no extra fingers, no plastic skin, no modern clothing.
This image was generated with GPT Image 2.5 by OpenAI. The prompt on this page is the exact text used to create it.
The example was generated with GPT Image 2.5 using: 2:3 · Cinematic · Fantasy · Portrait. Reusing these settings gives the closest match.
Copy the prompt as-is first, then replace only the subject while keeping the style, lighting, and composition keywords. Generate a few variations — outputs differ between runs.
Yes. Prompts work best as starting points — adjust descriptors, aspect ratio, or style tags to make the output your own.