
backroomstyle, sinorealmale, backroomstyle, a man floats on his back in a shallow pool in the vast white-tiled poolrooms, seen from high above. Camera: a high snapshot looking down from a tiled ledge; the man is small, floating in the middle of a long turquoise pool surrounded by white tiles and pillars. Setting: the poolrooms of the backrooms: a silent, endless indoor pool hall made entirely of small square white tiles: floor, walls, low ceiling, pillars, archways and stepped edges of shallow pools of perfectly clear still turquoise water. It repeats into the distance in every direction. Light: soft, even, sourceless white light; the water throws faint moving turquoise caustics on the tiles; very quiet and dreamlike. There is exactly one person in the frame and no one else. What is happening: He has given up trying to find the way out and simply floats on his back with his arms spread, eyes closed, alone in the silent tiled hall. Person 1 is a Japanese man aged 22, with wet black hair. He has a slim, natural adult build and smooth light warm East Asian skin tone. Position in the frame: small, in the middle of the pool. Posture: floating on his back. Arms and legs: arms spread on the water; legs relaxed. Clothing: navy swim briefs. The clothes fit naturally, with real fabric folds, wrinkles at the elbows and knees, visible seams and plain fabric with no printed text or logos. Expression: eyes closed, calm. He has exactly two arms, two hands with five fingers each, two legs and two feet, all clearly his own. Text: there is no text anywhere in the image except what is described above: no captions, no logos, no brand names, no readable writing on clothes, posters, books, packaging or signs, and no gibberish letters. Image quality: it looks like the original backrooms photograph: a single amateur snapshot from an old cheap digital camera, slightly overexposed, low saturation, a faint yellow-green colour cast, soft JPEG compression and grain, flat on-camera flash or flat overhead light, imperfect slightly tilted framing. Real-world scale and straight walls; the emptiness and the repetition are what make it unsettling. Nothing is dramatic, it is just wrong. Anatomy and body check: every person in the picture is one complete, coherent human body with exactly one head, one neck, one torso, two arms, two hands with five fingers each, two legs and two feet, all attached to their own body in natural adult proportions. No limb is duplicated, missing, merged, stretched, floating or attached to the wrong person. Where two bodies touch or overlap, the boundary between them is clear and it is always obvious which arm, hand, leg and foot belongs to whom. Hands hold objects with a natural grip and fingers wrap around what they hold; elbows and knees bend only in natural directions; feet rest flat under the body's weight. Bodies keep believable weight and balance: a seated person's hips are on the seat, a lying person's back is on the surface beneath them, a kneeling person's knees are on the floor. Faces are ordinary, natural, attractive adult faces of people aged 21 to 24, each clearly different from the others in bone structure, hair and features; men are clean-shaven with no beard or stubble.
Parameters used to generate this content
worst quality, bad anatomy, watermark, jpeg artifact, deformed hands, extra fingers, extra limbs, extra feet, extra legs, fused bodies, duplicate person, blurry, lowres, watermark, childlike body, old man, nude, penis, phone body, device frame, bezel, notch, misspelled text, garbled text, extra arms, extra hands, extra legs, extra feet, third arm, third leg, merged bodies, fused limbs, disembodied limb, floating hand, missing limb, twisted limbs, backwards knee, duplicated person, cloned person, twins, extra person, looking at camera, posing, cropped head, garbled letters, beard, stubble, studio lighting, professional photo, watermark, blurred face, censored, mosaic, detached head, twisted neck, cartoon, 3d render, gore, blood, horror makeup
The node graph used to generate this content — pan and zoom to explore, or download it to run in ComfyUI.
AI models used to generate this content