Visual brief: Ad-hoc Brief, Best-shot viral reel ADHD 11pm brain (remix of #1)

Source verdict: approved/2026-05-24/11pm-brain-reel.md Source draft: drafts/2026-05-24/11pm-brain-reel.md Format: reel (12-15s, single continuous shot, hook-and-caption, bipartite static overlay) Generated: 2026-05-25T02:00:00Z

Visual register summary

A single-shot macro of an eye in a darkened room, warm bedside-lamp key from camera-left, with a cool screen-glow blue reflection slowly intensifying inside the iris over 12-15 seconds. The visual register mirrors the brand's #1 reel (2.44M views, macro close-up eye dark moody) and stays inside its proven first-frame template. Palette is dominated by deep brown skin tones, warm amber key, and a single accent of dusty cool blue inside the iris reflection. Typography overlay is bipartite, static, persistent full duration, set in the brand display serif on the warm side and a clean sans on the cool side for legibility against two different skin and iris zones.

Production-ready units

Beat 1: 0-15s (single continuous shot, bipartite static text, subtle moving visual)

• Text overlay (verbatim from approved draft, BIPARTITE, both lines visible from frame 1, STATIC, no animation, persistent full 12-15s):

  • Top line: For 19 years researchers studied why ADHD brains finally come alive at 11pm
  • Bottom line: and the cause was never motivation

• Layout (9:16 canvas, 1080 x 1920):

  • Safe zones respected: avoid bottom 320px (Instagram caption + sound sticker overlap) and right-edge 140px (action rail: like / comment / share / save).
  • Top line: anchored in upper third. Top margin 180px from the top edge. Left margin 80px, right margin 220px (clears action rail). Width budget 780px. Text block height roughly 280-340px depending on wrap (expect 3 lines at the recommended size). Center-aligned within its block, left edge of block aligned at 80px.
  • Bottom line: anchored in lower-middle third, above the IG caption zone. Bottom edge sits at y = 1480px from top (440px clear of bottom edge). Same horizontal anchoring as top (80 / 220px margins, 780px width). Expect 1-2 lines. Block height roughly 140-180px.
  • Vertical breathing room between the two text blocks: minimum 480px of visual canvas where the eye is the focal point uninterrupted. This is where the iris and pupil sit and where the dilation and glow read clearly.
  • No slide-number tag, no logo, no watermark (reel native, brand handle is supplied by the IG account, not baked into the frame).

• Background (the shot itself, treated as the live background under the overlay):

  • Deep desaturated brown-black behind the subject. No texture overlay added in post. Background is naturally rendered through the shallow depth-of-field of the macro shot. Effective background hex reference range: #2A1F18 (charcoal) fading to near #1A1310 in the far out-of-focus zones. No pure black #000000 (off-brand per protocol section 0).

• Typography:

  • Top line: Recoleta Semi-Bold, ~64-72pt at native 1080 x 1920. Sentence case (matches the approved draft). Color: warm off-white #FAF3E2. Tracking -10. Leading 1.15. Subtle text shadow: 0 0 24px at 35% opacity in #1A1310 to lift the type off the skin-tone zone without reading as a designed drop shadow. No outline, no stroke, no glow.
  • Bottom line: Inter Medium (NOT Recoleta, intentionally subordinate), ~42-48pt at native 1080 x 1920. Sentence case. Color: warm off-white #FAF3E2 at 92% opacity (one tonal step below the top line). Tracking 0. Leading 1.25. Same subtle shadow spec as the top line.
  • Why two typefaces here: the bipartite hierarchy is mechanical, not decorative. The serif top line reads as headline and parses first; the sans bottom line reads as supporting clause and parses second. Both still legible under 1.5s scan per the brief's "bottom subordinate to top but parses <1.5s" requirement. This is within the protocol's "maximum two typefaces per asset" cap.

• Cinematography:

  • Lens: macro, 100mm equivalent on full-frame (or 60mm macro on APS-C / 50mm macro on Micro Four Thirds). Working distance approximately 25-35cm from the eye. If shooting on phone, use a clip-on macro lens; native phone "macro modes" tend to deliver sub-par bokeh and edge sharpness on a subject this small.
  • Aperture: f/2.8 to f/3.5. Shallow depth of field so the iris and pupil are tack sharp, eyelashes fall off into soft focus, and the background is completely out of focus.
  • Framing in 9:16: the eye fills 60-70% of the frame width. Horizontally centered. Vertically: the pupil sits at canvas y = 960 (true center, +/- 80px). The eye's horizontal axis runs through the canvas middle so the upper and lower lash lines bracket the central focal zone. Skin / brow visible above and below the eye, providing the warm-tone area where the top line and bottom line sit over rendered skin rather than mid-eye.
  • Lighting: warm key light from camera-left at roughly 30-45 degrees off the lens axis (suggesting a bedside lamp out of frame). Color temperature warm, 2700-3000K equivalent. Soft source (small softbox or a paper-diffused practical), not a hard direct bulb. No fill on the right side; let the right side of the eye and face fall into shadow.
  • The cool screen-glow reflection: simulated in the iris using a small cool-blue practical (a phone screen face-up at 25-40cm below the subject's eyeline, displaying a static blue-white frame) OR added in post as a soft cool-blue catchlight inside the iris that intensifies over time. Color temperature ~7500-9000K. Hex reference for the cool reflection at peak intensity: #7C95A6 (dusty blue from the brand palette, focus / science accent).
  • Color grade: desaturated warm shadows, slightly cooler highlights inside the iris reflection only. Crush blacks gently (lift the blackest black to ~12/255 to keep the brand "no pure black" rule). Slight film grain at 5-8% intensity, fine grain (not chunky). No vignette added in post beyond what the shallow DOF naturally provides.

• Color accents (palette mapping):

  • Skin / warm zone: rendered through the warm key, not assigned a fixed hex (it's a live subject); falls within the brand warm range, reads as a desaturated terracotta #B86C50 to deep brown #3D2818 family.
  • Cool reflection inside iris: dusty blue #7C95A6, brand accent for focus / science slides. This is the only cool accent in the frame and it carries the visual narrative.
  • Type: warm off-white #FAF3E2 for both lines.
  • No green accent, no other warm accent. The frame is two-color: warm everywhere, cool inside the iris reflection only.

• Motion choreography (CRITICAL, subtle throughout, no cuts):

  • 0-2s: baseline. Pupil at relaxed normal diameter (roughly 4mm equivalent in frame). Warm key visible on the skin and lashes. Iris reflection: nearly invisible, a faint cool tint only.
  • 2-4s: the cool blue-white reflection begins to appear inside the iris, low intensity. The pupil holds. The viewer is still parsing the bipartite text.
  • 4-6s: reflection intensity builds, now clearly readable as a "screen" reflection (phone / laptop) inside the iris. Pupil still holds at baseline.
  • 6-7s: ONE slow, soft blink. Single eyelid closure over roughly 0.8-1.0s (closing slow, briefly held for ~150ms, opening slow). Not a sharp natural blink, deliberately languid. The blink resets the viewer's attention and signals a state shift.
  • 7-10s: pupil starts dilating SLOWLY in response to the still-growing cool light source. From roughly 4mm baseline to roughly 5.5-6mm equivalent. Dilation should be barely perceptible frame-to-frame, fully readable when comparing the 7s frame to the 12s frame.
  • 10-15s: pupil reaches fully dilated state (~6-7mm equivalent). Cool reflection at peak intensity. Warm key holds steady throughout (the lamp does not change; only the screen-glow changes). End frame: dilated pupil, bright cool reflection in iris, warm-lit skin, bipartite text still locked in.
  • The implied story (and the catchy element): something is waking up in this person right now. The motion is felt as recognition, not as drama. Nothing "happens" big. The whole reel is the slow internal arrival of attention.
  • Loopability: the reel does not need to loop cleanly; the dilation builds in one direction. If Andrii wants a soft loop, cross-fade the final 0.5s back to the opening frame in post (acceptable but not required).

• Audio direction (per draft, restated): low ambient room tone, faint distant traffic hum, single soft analog synth pad swelling under the final third (~10-15s). No voiceover. No trending audio. Pad swells in volume by roughly 3-4dB over the last 5 seconds to subtly mirror the visual intensification. Mix sits quiet enough that sound-off scrolling loses nothing; sound-on rewards lingering.

Cross-unit consistency

Single-shot reel, no cross-unit consistency to specify. Bipartite overlay is present and identical for the full 12-15s duration. No transitions, no cuts, no scene changes.

AI generation prompt

Video prompt (model-agnostic, ~150 words)

Extreme macro close-up of a single human eye in a darkened bedroom, the implied moment when someone's attention quietly wakes up late at night.

The camera holds a static macro shot, locked off, no movement, framing the eye to fill roughly two-thirds of the vertical frame.

The eye is open, naturally moist, with visible iris detail and individual eyelashes. The pupil dilates slowly across the duration, from a relaxed baseline to noticeably wider. The subject blinks once, slow and soft, around the middle of the clip.

Background is a completely out-of-focus dark room, no recognisable objects, deep desaturated brown-black bokeh.

Warm amber lamp light grazes the skin and lashes from the left, suggesting a bedside lamp out of frame. Inside the iris, a cool blue-white screen reflection slowly intensifies across the clip, as if a phone or laptop is being held just below the eye. Slightly desaturated, faint film grain, shallow depth of field.

Audio: quiet room tone, faint distant traffic, a single soft analog synth pad swelling in the final third. No voiceover, no music.

No on-screen text. 9:16 vertical, 12 to 15 seconds.

Designer notes

• Creative latitude:

  • Acceptable: small variation in iris color (brand-neutral; the cool reflection is what carries the visual, not the iris pigment), skin tone, eye shape. The shot is anonymized by the macro framing, not by identity. Any subject works as long as the eye reads clearly and the warm-cool light contrast lands.
  • Acceptable: swap blink timing within 6-8s window if the AI model's blink lands more naturally at 6s or 8s than 7s. The "single slow blink in the middle third" is the constraint, not the exact second.
  • Not negotiable: static bipartite overlay from frame 1, full duration, no animation. One shot, no cuts. Warm key + cool iris reflection as the two-color story. Pupil dilation building across the duration. No on-screen text in the AI generation; overlay added in post.

• AI model fallback path (likely needed): macro eye motion with believable slow pupil dilation AND a believable growing screen-reflection inside the iris is at the edge of what current text-to-video models reliably nail. Recommended fallback in this order:

  1. Try Veo 3.1 first (best at coherent micro-motion + native audio). Test 2-3 generations with the prompt above.
  2. If pupil dilation does not read or the iris reflection looks artificial, shoot a real macro eye clip (Andrii or model) with the warm key in place, and do the pupil dilation + intensifying iris reflection in After Effects. Pupil dilation is a single circular mask scaled over the timeline; iris reflection is a soft blue radial gradient on a screen blend mode, opacity ramped 0% to 65% across 12-15s.
  3. If neither path is workable, the next-best AI fallback is to drop the pupil dilation entirely and let only the iris reflection intensify; this still tells the "something is coming online" story, and the pupil dilation becomes a "nice to have" instead of a "required".

• Overlay typography fallback: if Recoleta is not licensed on the editing machine, use Lora Bold for the top line at the same point size (~64-72pt). Inter is broadly available; if unavailable, fall back to SF Pro Text Medium at the same point size for the bottom line.

• Export specs: 9:16, 1080 x 1920, MP4 H.264, 30fps, color profile Rec.709 sRGB, audio AAC stereo 48kHz. Target file size under 100MB. Caption supplied separately from the approved draft, not baked into the video.

• Sound-off readability: verify the bipartite overlay reads cleanly at 50% phone screen brightness in a sunlit room before shipping. The #FAF3E2 warm off-white on the deep brown background should hold contrast at small phone sizes.

• No open questions. Asset is fully specified above.