Original prompt as shared from the source.
TASK: Render the same characters from the source image as a cinematic photograph in a NEW pose, NEW lighting, and NEW framing set up by a real director on a real set. The source provides identity and wardrobe only — everything else is rebuilt from scratch.
LOCKED (must match the source exactly):
- Face shape, nose, eyes, jawline, age, skin tone of each character.
- Haircut: exact length, color, part, texture.
- Beard (if any): exact shape, length, color, greying pattern, moustache style.
- Costume: exact silhouette, fabric type, colors, trim, accessories.
- Headwear: if a character has a veil, hood, coif, cap, or wrap in the source, keep it identically. Do not remove or replace it.
A viewer comparing source and output side by side must instantly recognize each character as the same individual.
UNLOCKED (must be rebuilt — do NOT carry over from source):
- Pose, gesture, body language. The source poses are stiff game animations; the photograph has natural, fluid, candid blocking with weight, micro-expressions, and overlap between characters.
- Camera angle and framing. Cinematographer's choice — over-the-shoulder, low angle, three-quarter, off-centre. NEVER the source's frontal symmetrical stage layout with characters facing camera.
- Lighting. Motivated practical light from the scene's real sources — hearth fire, candles, oil lamps, window shafts, torches, overcast daylight, golden hour — whichever the setting demands. Strong chiaroscuro for interiors, deep falloff into shadow. NEVER flat ambient game lighting.
- Environment. Fully rebuilt as a real practical set with depth, real materials, props, atmosphere, soot, wear. NOT the source's painted backdrop.
- Rendering medium. A real photograph captured by a real camera on set — NOT a "photorealistic version of a CG render."
Read all on-screen text in the source (any language) to determine the event, setting, mood, and action. Discard every UI element: text boxes, portrait frames, banners, icons, borders, captions, name underlines.
Output: a still from a prestige live-action historical series (Game of Thrones, The Last Kingdom, The Northman, The Last Duel). Shot on ARRI Alexa with anamorphic lenses, shallow depth of field, filmic color grade with rich shadows and muted earth palette, subtle film grain, gentle highlight halation.
Material translation — every surface photographically real: skin with pores, fine vellus hair, subsurface scattering, age-appropriate texture; individual hair strands and flyaways; cloth with visible weave and wrinkles; leather with grain and scuffs; oxidized metal with dull patina; soot, wax drips, and wear on interior props.
The output must look NOTHING like the source as an image: different angle, different pose, different light, different atmosphere, different medium. Only the people and what they wear are recognizable.
Hard negatives: video-game graphics, game engine output, Unreal Engine, 3D render, CGI, photorealistic CGI, concept art, illustration, painting, smooth shaders, plastic skin, flat ambient lighting, three-point studio portrait lighting, symmetrical stage blocking, frontal proscenium framing, characters facing camera, characters lined up evenly spaced, UI, text overlays, portrait frames, character redesign, removed headwear, slimmed faces, beautified actors.
