374d5eaf-797f-4db7-a159-8ccf2b7d60bd

**POSITIVE PROMPT** Plate for shot S1.31 — @Loft, derived from the **P-05 glass macro plate**. Vertical 9:16, 1440x2560. Macro on **@Monitor1's matte anti-glare surface**, square to the screen plane, framed on the waveform band with **her diffuse reflection floating soft over the glass** — the same optical family as S1.21, the trace and the suggestion of her sharing one plane. Single plane plus the reflection layer. Lens character 85mm macro at 0.30 m, camera at seated eye height 118 cm, square to the matte screen, shallow plane of focus with the surface sharp and the extreme frame edge softening. Surface: real anti-glare coating granularity under raking light, the micro-texture that scatters rather than mirrors, the thumb smudge at the lower left, a faint spray of dried speech droplets in the lower third, fine dust at the panel edge. @Monitor1 matte, W2, thin bezel barely intruding at the extreme frame edge. Reflection: **diffuse only, capped opacity, heavy blur** — an unresolved suggestion of a face and shoulder, no features, no eyes, no legible geometry, never a mirror image. Position consistent with her seated square to the screen, hair loose on the left. **Motionless.** Lighting: **generate the panel unlit and neutral.** The key colour changes three times across this shot — green on, off, on — and every one of those transitions is a comp-stage change, never baked into the plate. No green in the generated image, no blue-white bias, no secondary source, no sodium. No glow bloom, no halation, no light wrap. **Generate the screen dark and empty.** Leave the waveform band and **the extreme frame edge** clean and unpopulated as comp receiver plates. No waveform, no trace, no type, no letters, no numbers, no cursor, no UI elements in the generated image — the four-second waveform is a fixed asset, built once, composited **pixel-identical on every repetition**, and the pulsing draft window at the frame edge is likewise composited. Static. Locked off. Nothing moves. **NEGATIVE PROMPT** glossy screen, mirror reflection, sharp reflection, legible reflected face, reflected eyes, recognisable person in reflection, resolved portrait, high-opacity reflection, second reflection, reflection of a different room, moving reflection, mirror, glass sheen, specular highlight, screen glare, green light, green emission, green cast, blue-white light, warm light, amber light, sodium light, coloured spill, baked-in key colour, glow, bloom, halation, light wrap, neon, lens flare, waveform, audio trace, spectrum bars, equaliser, oscilloscope, visualiser, varying waveform, text, type, letters, words, numbers, timestamps, cursor, UI elements, interface, window, panel, icons, menu bar, buttons, generated screen content, legible anything, watermark, subtitle, deep focus, sharp background, everything in focus, separated planes, clear foreground and background, wider shot, tighter shot, bezel dominant, second monitor, dead pixel, room visible, desk visible, keyboard visible, mug visible, hands, face, person in frame, body, camera movement, push in, drift, handheld, dolly, tilt, angled to screen, off-square, perspective distortion, pristine panel, new monitor, clean screen, no smudges, no dust, moved smudge, crushed blacks, high contrast, oversaturated, HDR, sharpening halo, grain, film emulation, vignette, colour grade, stylisation, cinematic look, moody, atmospheric, beautiful, aesthetic, curated, clean, tidy, horizontal framing, AI artefacts, morphing edges

Generate a single, hyper-realistic character reference image (strictly no collages, text, watermarks, or borders). The image must feature a cinematic, three-quarter portrait of the described character, maintaining a sharp, tack-focused emphasis on the face with natural skin textures, pore details, and realistic micro-expressions. The lighting should be cinematic and directional (e.g., subtle split or Rembrandt lighting) to give depth, blending seamlessly with the environment specified in the character description. The character must have a natural, living presence, looking slightly off-camera or directly at the lens with a subtle, context-aware expression. Ensure the full head, neck, and upper shoulders are completely unobstructed. This image serves as a high-fidelity identity blueprint for temporal-consistent video generation.