On Repeat
Places for listening
A white cord on a walk, silver cups while waiting, a dial within reach at home. Objects for listening find their place between clothes and gestures.
A walk beside a blue wall. A white earphone cord falls against a black cardigan; her gaze leads beyond the frame.
At the window, a hand rests beneath her cheek. A phone, a small cup, and the blue window frame draw the scene closer.
An orange platform bench. Silver headphones, a dark jacket, and trainers set side by side form a waiting pose.
The final frame brings us to the fingertips. Between a green knit and a cream speaker, a hand rests on the dial.
From a street to a window, through a platform and into a room. Separate works arranged for this issue, rather than a record of one actual day.
Read the production notes+
Bringing four images together
The original 2D prompts specify full-body framing, clothes, and light, alongside continuous earphone cables and naturally scaled props. The instructions attend to the balance of the whole scene.
The photographic window image uses a 2D scene and the approved red-chair face as distinct references. The speaker image is a revision: its edit prompt asks for gently lifted shadow midtones on the face and exposed hands. We selected the file linked to the publication record, rather than the first result.
The order and scene commentary were written for this edition. The recorded generation tool is built-in image_gen; the model version was not exposed.
Original prompts remain in English. Reusing them does not guarantee an identical result.
01 · EarPods / 2D
Use case: illustration-story. Create ONE vertical 2:3 high-quality 2D anime fashion editorial image, not a collage. Reference image is ONLY RURI identity and linework, NOT outfit, pose, room or framing. Preserve SAME adult RURI delicate grey eyes, long straight black hair, blunt bangs, fine braids, white cross hairpins, refined small nose/mouth, naturally narrow sloping shoulders and natural adult proportions. Pure drawn anime face/body AND painterly animated environment, elegant fine linework, tactile painted fabric, sophisticated limited color palette. No photoreal skin, 3D doll, captions, poster titles or watermark. Product design should be recognizable, physically plausible and real-sized. Do not exaggerate product to show it. Natural anatomy, coherent head/torso orientation, correctly attached hands. Full body street fashion frame. RURI walking left to right past a cobalt blue painted wall in warm afternoon sun, head upright in clear side profile looking ahead. Black fine-knit cardigan, ivory wide trousers, black loafers. Small white Apple EarPods in ears, NO silicone tips, white cables naturally descending over black cardigan into a single cable connected to smartphone in her lowered hand. Correct cable continuity. Spacious composition with feet visible, no giant props.02 · EarPods / Photoreal
Image 1 is the TARGET composition/wardrobe/props reference. Image 2 is the PRIMARY APPROVED fictional adult RURI PHOTOREAL FACE identity, not its clothing or pose. Produce ONE vertical photoreal image. Preserve the exact same adult woman as image2: grey eyes at realistic human size, delicate nose and natural lips, natural skin texture, long black hair/blunt bangs/thin braids/white crossed hairpins, narrow softly sloping shoulders and believable adult body. NO anime eyes, illustration outlines, doll, CGI or beauty filter. Do NOT transfer image2 zebra knit, red bag, red chair or hand-on-cheek pose unless already in target. Transform image1 into a sophisticated real cinematic fashion photograph. Preserve target framing, camera angle, pose, expression, head direction, every wardrobe item, bag, shoes, audio object, background geometry and lighting/mood. Real-world fabric, skin, leather, metal, architecture. Products plausible in size and contact; sound prop shape from target. Keep feet visible when visible originally. Preserve accessories but realistic detail. No text overlay or watermark. Faithful photographic reinterpretation of this exact scene, not a new scene.03 · Sony / 2D
Use case: illustration-story. Create ONE vertical 2:3 high-quality 2D anime fashion editorial image, not a collage. Reference image is ONLY RURI identity and linework, NOT outfit, pose, room or framing. Preserve SAME adult RURI delicate grey eyes, long straight black hair, blunt bangs, fine braids, white cross hairpins, refined small nose/mouth, naturally narrow sloping shoulders and natural adult proportions. Pure drawn anime face/body AND painterly animated environment, elegant fine linework, tactile painted fabric, sophisticated limited color palette. No photoreal skin, 3D doll, captions, poster titles or watermark. Product design should be recognizable, physically plausible and real-sized. Do not exaggerate product to show it. Natural anatomy, coherent head/torso orientation, correctly attached hands. Full body RURI sitting naturally on burnt orange bench at quiet architectural train platform. Both feet visible resting on ground, one leg slightly forward. She wears platinum silver Sony WH-1000XM6 headphones correctly over ears, padded rounded headband over crown, simple oval matte earcups and hinged yokes, no boom microphone. Dark navy short jacket over cream tee, charcoal straight trousers, silver-grey sneakers. Face level in three-quarter view looking left at arriving light, NOT head tilt. Warm orange versus blue-gray palette, generous negative space.04 · Marshall / Initial generation
Image 1 is the TARGET composition/wardrobe/props reference. Image 2 is the PRIMARY APPROVED fictional adult RURI PHOTOREAL FACE identity, not its clothing or pose. Produce ONE vertical photoreal image. Preserve the exact same adult woman as image2: grey eyes at realistic human size, delicate nose and natural lips, natural skin texture, long black hair/blunt bangs/thin braids/white crossed hairpins, narrow softly sloping shoulders and believable adult body. NO anime eyes, illustration outlines, doll, CGI or beauty filter. Do NOT transfer image2 zebra knit, red bag, red chair or hand-on-cheek pose unless already in target. Transform image1 into a sophisticated real cinematic fashion photograph. Preserve target framing, camera angle, pose, expression, head direction, every wardrobe item, bag, shoes, audio object, background geometry and lighting/mood. Real-world fabric, skin, leather, metal, architecture. Products plausible in size and contact; sound prop shape from target. Keep feet visible when visible originally. Preserve accessories but realistic detail. No text overlay or watermark. Faithful photographic reinterpretation of this exact scene, not a new scene.04 · Marshall / Published tone revision
Use case: lighting-weather. Image1 is the edit target, an approved realistic photograph of fictional adult RURI. Image2 is ONLY a reference for clear luminous skin midtones, NOT identity, pose, clothing or background changes. Make a VERY SUBTLE localized photographic lighting/color correction to image1: a little soft reflected fill on the face, neck and exposed hands, slightly lifting muddy shadow midtones (approximately one quarter stop; less where already bright), gently reducing grey-brown cast without bleaching skin. Preserve realistic pores, natural skin color variation, facial depth, cast shadows and existing directional light. Preserve EXACT face identity, facial features, eye sizes, expression, gaze, hairstyle, head angle, body, pose, clothes, objects, composition, background, scene exposure, color mood. No facial reconstruction, beauty smoothing, new makeup, whitening, illustration or relighting entire scene. Keep bright sunlit skin highlights unchanged. Result should still have the same believable sunlight/backlight; only a slightly clearer, less murky complexion. Output the full original vertical composition.