Analog Horror Vhs Effect

GLOBAL LOCK: vertical 9:16 teaser trailer built around a stack of black VHS tapes on a rough wooden tabletop in a dark room, every cassette marked with the word BURN in bold red block letters, moody analog-horror promo aesthetic, no people visible, no location change, no props beyond the tapes, table, smoke, fire, and overlaid release text, camera remains close and product-focused with a slightly elevated front angle, shallow depth of field, soft smoky atmosphere, warm firelight mixed with cool ambient shadows, rich blacks, muted browns, red title accents, subtle film grain, realistic flame behavior, realistic smoke drift, premium indie-film announcement tone, on-screen typography appears as part of the edit and must remain crisp and centered when present.

TIMECODED SHOT SEGMENTS:
[00:00-00:02] Tight close-up of a messy stack of black VHS tapes resting on a worn wooden tabletop in a dim room. The top tape sits slightly angled over the others, each visible spine or label reading BURN in red. Lighting is low and cinematic, with soft top-left illumination and deep shadow around the edges. No flames yet, just a tense still-life composition with analog texture and faint dust in the air.

[00:02-00:04] Same composition, but the first small flames appear near the lower-right edge of the stack and begin licking upward along the tape corners. The camera stays locked in the same close framing with minimal drift. Warm highlights start reflecting across the plastic shells, increasing contrast between the red lettering and the dark cases.

[00:04-00:06] Fire grows more visible around the lower half of frame. Overlaid title text begins to appear above the tapes: IF ONLY presents, followed by the large central title BURN. The tapes remain the visual anchor while flames crawl around the base. Smoke becomes more noticeable, rising in soft gray plumes behind the text.

[00:06-00:08] The text stack becomes fully readable, with BURN dominant and a smaller line indicating the project is an AI short film. Flames intensify, wrapping the front and right side of the cassette pile. Maintain the same lens feel and angle so the motion comes from the fire and smoke, not from camera movement. This beat should feel like a full title-card reveal.

[00:08-00:10] The fire reaches its peak intensity. Orange flames surge higher in front of and behind the tapes, while smoke thickens and partially veils the upper background. The central tape remains readable. This is the highest-energy section, but the composition still stays clean and poster-like.

[00:10-00:12] The main title treatment fades out and is replaced by a release call to action: OUT NOW with LINK IN BIO beneath it. Flames still move aggressively around the cassette pile, but the text must stay legible and stable against the smoky background. Keep the same stacked-VHS hero shot and moody color grade.

[00:12-00:13.4] Final hold on the burning tapes with OUT NOW / LINK IN BIO still visible. Smoke continues drifting upward and flames flicker around the lower-right and lower-center edges, ending on a polished indie-film promo tableau rather than a chaotic explosion.

NEGATIVE PROMPT: extra objects, extra tapes appearing or disappearing, warped cassette geometry, unreadable labels, random logos, watermark, subtitles, plastic melting into surreal shapes, exaggerated explosion, unrealistic fire simulation, color banding, muddy blacks, blown highlights, temporal jitter, title flicker, soft unreadable typography, camera whip, zoom pulse, lens breathing, cartoon flames, fake smoke loops, low-detail wood grain, AI artifact edges, ghosting, duplicate tape corners, unstable text alignment.

SPEECH PACK:
- Transcript status: no spoken dialogue.
- Audio design intent: low ominous cinematic atmosphere, crackling fire, subtle room tone, light smoke hiss, no vocals, no narration, no singing.
- TAKE_A: pure fire crackle with restrained dark ambient bed.
- TAKE_B: slightly heavier cinematic drone under the same fire and smoke texture.
- TAKE_C: minimal ambient rumble with the clearest fire detail for a more trailer-like finish.

GLOBAL LOCK: Vertical 9:16 photoreal horror object-animation shot set in a dim, grimy storage or workshop space with a vintage CRT television on a low table and a dirty clown doll sitting slumped in shallow floodwater on the floor directly in front of it. The mood is analog-horror and cursed-object cinema: soft practical lighting, oily reflections on wet surfaces, dusty clutter in the blurred background, worn plastic, old electronics, and a dead silent room turning supernatural. The clown doll has a glossy white face with black teardrop eye makeup and a red sinister smile, yellow torso detail, dirty white costume, and limp seated posture. The CRT is beige-gray with knobs and vents on the right side. The horror action centers on the screen activating with eerie blue glow and static, then a long skeletal fleshy arm forcing its way through the broken glass and grabbing the clown doll, dragging it into the television. Keep the movement deliberate, unsettling, and physically plausible within a cursed-TV logic. No speech, no narration, no subtitles.

[00:00-00:02.00] Start on the doll and CRT in a quiet low-angle composition. The clown doll sits in the lower foreground leaning against the front of the television stand, its dirty costume soaked and wrinkled. The CRT is dark or faintly waking, with the room lit by soft practical overhead spill and dim reflections in the water on the floor. The environment should feel like a neglected prop room or flooded basement workshop. Audio is only room hum, water stillness, and faint old-electronics ambience.

[00:02.00-00:04.20] The television screen brightens into a cold blue static glow, casting a subtle cyan wash over the doll’s face and the wet floor. The doll remains motionless, which increases dread. Tiny flicker patterns and screen hiss imply something alive behind the glass. The camera stays fixed, letting the anticipation build. No words, only CRT hiss and electrical tension.

[00:04.20-00:07.20] A jagged hole appears or becomes visible in the screen as a long, pale, emaciated arm pushes through from inside the television. The hand is bony and elongated with tendon-like structure, emerging from black void and shattered static. It reaches down toward the clown doll’s chest or neck area and clamps onto it. Glass edges, static remnants, and physical contact are important: the arm should feel invasive, as if another dimension is tearing through outdated electronics. Audio should include screen crackle, brittle glass stress, and a fleshy dragging sound.

[00:07.20-00:09.50] The arm yanks the doll upward and backward toward the screen opening. The clown’s limp body jerks as it is pulled, one of the most unsettling beats of the clip because the object that looked like a prop suddenly becomes prey. The water surface trembles from the movement. The television remains the fixed portal, its blue-white glow framing the action.

[00:09.50-00:11.07] End with the doll gone and the screen darkened again, leaving only the vintage CRT on the table in the flooded room. The floor in front looks empty except for small debris and reflection. The final beat should feel like a completed abduction rather than an attack interrupted. Audio decays back into empty room tone, faint residual hiss, and eerie silence.

NEGATIVE PROMPT: comedy clown, bright carnival colors, cartoon ghost hand, modern flat-screen TV, clean room, dry floor, obvious CGI glow, exaggerated gore splatter, over-fast attack, text overlays, subtitles, camera shake, smiling cute doll, plastic fake hand, multiple arms, action-movie pacing, demonic face coming out of screen, neon cyberpunk palette, voiceover, human dialogue, flashy editing, broken proportions on the arm.

SHOT PROMPTS:
SHOT_1 [00:00-00:02.00]: eerie low-angle setup with dirty clown doll seated in shallow water before an old CRT in a cluttered dim room.
SHOT_2 [00:02.00-00:04.20]: screen wakes with blue static glow, doll still frozen, tension builds.
SHOT_3 [00:04.20-00:07.20]: skeletal fleshy arm bursts through cracked screen opening and grabs the doll.
SHOT_4 [00:07.20-00:09.50]: arm drags the clown backward into the television, water and fabric shifting under the pull.
SHOT_5 [00:09.50-00:11.07]: television remains alone, doll vanished, room returns to dreadful stillness.

SPEECH PACK:
Timecoded transcript: No speech. The reference relies on object movement, CRT hiss, and horror ambience only.

[00:00-00:02.00]
TAKE_A: [quiet room hum, faint wet ambience]
TAKE_B: [old electronics idle tone, still water]
TAKE_C: [soft ambient silence]
Prosody / emotion: calm before intrusion.

[00:02.00-00:04.20]
TAKE_A: [CRT static hiss rising]
TAKE_B: [electrical buzz and blue-screen crackle]
TAKE_C: [screen noise intensifies]
Prosody / emotion: dread, awakening.

[00:04.20-00:07.20]
TAKE_A: [glass stress, static rupture, fleshy reach]
TAKE_B: [screen crackle with scraping pull]
TAKE_C: [unnatural tearing and electronic distortion]
Prosody / emotion: violation, shock.

[00:07.20-00:09.50]
TAKE_A: [fabric drag, water disturbance, pulling tension]
TAKE_B: [wet scrape and electronic hiss]
TAKE_C: [sudden abduction sound bed]
Prosody / emotion: helpless extraction.

[00:09.50-00:11.07]
TAKE_A: [ambient room returns, residual static]
TAKE_B: [low hum, silence after the event]
TAKE_C: [empty room dread]
Prosody / emotion: aftermath, unease.

GLOBAL LOCK: keep one old boxy cathode-ray television on a low wooden stand in a dim vintage living room with faded wallpaper, framed wall art, patterned rug, and partial sofa visible at frame right; keep one pale female ghost with long wet dark hair, hollow exhausted face, and thin crawling body emerging from the TV. Preserve the same blue-white spectral glow, smoky screen distortion, and quiet domestic horror tone throughout.

Create a 15-second haunted television horror scene where a CRT left on after programming ends becomes a portal for a ghostly woman to crawl into the room. The sequence should escalate from eerie screen static to a face in the glass, then to full supernatural emergence and a final crawl across the rug. No subtitles and no on-screen text.

0:00-0:03 - Open on the old television alone in the living room. The screen glows with unstable blue-white noise and vague moving shapes inside. The room is still, dim, and ordinary, which makes the electronic disturbance feel more threatening.

0:03-0:06 - Let the ghostly face begin to resolve inside the screen. The woman’s pale features appear behind the glass, staring outward while the image warps with analog distortion. Keep the TV physically grounded on the stand while the impossible image intensifies.

0:06-0:09 - Escalate into supernatural breakthrough. Blue-white mist and energy spill out from the screen edges and the ghost starts forcing herself through the television frame. Her shoulders and upper torso cross the threshold as if the set has become liquid and solid at once.

0:09-0:12 - The ghost drops or crawls down from the TV onto the rug. Keep her low to the ground on hands and knees, hair hanging forward, body slick with spectral light and fog. The television continues to glow behind her like an open portal.

0:12-0:15 - Finish with the ghost fully on the floor, crawling toward camera or across the room while the TV still burns with cold blue light in the background. End before any attack, holding on the completed emergence.

CAMERA: locked or near-locked living-room composition, slight cinematic push-in only if subtle, wide enough to include TV, stand, rug, and ghost movement out of the screen. No handheld shake.

LIGHTING: dim practical room ambience with the TV as the dominant blue-white light source; use spectral spill, mist glow, and soft shadow on walls and floor.

GRADE: desaturated domestic horror palette, dusty browns and greys in the room, cold electronic blue in the screen and fog, restrained contrast with supernatural highlights.

MOTION: analog flicker, screen distortion, mist seepage, slow unnatural crawl from the set, low floor movement, no jump-scare sprinting, no violent camera effects.

SPEECH: no dialogue, no narration, no captions, no lip-sync. The scene is entirely visual and atmospheric.

SPEECH PACK: none. Sound design should imply CRT hum, static crackle, low electrical buzz, spectral hiss, and soft crawling movement only.

NEGATIVE PROMPT: no subtitles, no text overlays, no comedy, no modern flat-screen TV, no extra people, no blood spray, no gore dismemberment, no bright daylight, no monster army, no possession of other furniture, no identity drift, no cartoon ghost, no aggressive sprint toward camera, no futuristic room design.

[Visible Inventory]
Subject(s)
- No human or character subjects visible.
- Abstract signal texture only.

Clothing & materials
- Not applicable (no clothing items).

Props/objects
- Full-frame grayscale analog-TV-like static pattern.
- Dominant horizontal bands: bright noisy strip near top, thick dark band through upper-middle, mid-gray dense grain zone, bright band near bottom.
- Fine white speckle noise distributed across entire frame.
- No text, no logos, no icons, no interface overlays.

Environment
- No physical environment depicted; purely abstract signal/noise field.
- Feels like a close-up of interference on a CRT/analog monitor.

Composition
- 640x508 frame (near 5:4), full-bleed texture.
- Horizontal layering is the key structure.
- No focal object; composition relies on tonal band contrast and grain density.
- Minimal perspective cues, almost flat frontal plane.

Lighting
- Monochrome luma-driven contrast rather than scene lighting.
- Brightest areas at top and bottom bands; darkest band in upper-middle.
- Soft bloom in bright noise areas; no directional shadowing.

Color palette
- Strict monochrome: black, charcoal gray, mid-gray, off-white.
- Low chroma, medium contrast.

Image style
- Analog interference / VHS static / CRT scanline aesthetic.
- Heavy film grain and noise texture.
- Intentionally degraded, lo-fi, archival-tech mood.

[MASTER PROMPT]
[Subject] Abstract monochrome analog signal interference, no people, no objects, no typography.
[Environment] Full-frame CRT-like static noise texture with layered horizontal scan bands and dense granular speckle.
[Composition/Camera] Flat frontal capture, full-bleed texture, near 5:4 canvas, dominant horizontal banding: bright noisy top strip, dark central stripe, textured gray mid-field, bright lower strip.
[Lighting] Luma-only tonal contrast; brightest noise clusters at top and bottom, deep dark stripe in upper-middle, no directional key light.
[Style/Rendering] Retro broadcast glitch, VHS/CRT interference realism, high micro-noise detail, analog artifact simulation, intentionally degraded and atmospheric.
[Detail constraints] Keep monochrome palette, preserve horizontal band layout, avoid any figurative subjects, avoid text/logos/UI marks, keep dense speckled grain across the frame.

[Negative Prompt]
people, face, eyes, body, landscape, architecture, text, letters, watermark, logo, color tints, rainbow glitch blocks, geometric clean patterns, vector art, smooth gradients, sharp modern UI, low-noise clean image, anime elements.

[Suggested Parameters]
- Aspect ratio: 5:4 (or 640:508 close match)
- Lens/focal length feel: N/A (texture plate / no perspective)
- Depth of field: deep/uniform (flat texture clarity)
- Steps: 28-40
- CFG / style strength: 5-7
- Sampler: DPM++ 2M Karras (or Euler a for rougher grain)
- Seed suggestion: 90577124

[Delta Prompt Strategy: 10 Drift Risks + Corrective Micro-Prompts]
1) Drift: Colored artifacts appear.
   Fix: "strict grayscale only, no color channels, black-gray-white palette"
2) Drift: Human silhouettes emerge.
   Fix: "non-figurative abstract signal texture, zero human forms"
3) Drift: Banding becomes vertical.
   Fix: "dominant horizontal interference bands across full width"
4) Drift: Image becomes too clean.
   Fix: "dense analog speckle noise and micro-grain over entire frame"
5) Drift: Central dark stripe missing.
   Fix: "retain one thick dark band through upper-middle"
6) Drift: Bright top strip too weak.
   Fix: "high-luma noisy strip near top edge"
7) Drift: Bottom highlight disappears.
   Fix: "secondary bright noisy band near bottom edge"
8) Drift: Modern digital glitch blocks appear.
   Fix: "avoid RGB datamosh blocks; use analog CRT-style interference only"
9) Drift: Texture over-sharp and synthetic.
   Fix: "soft-edged noisy bloom, organic analog degradation"
10) Drift: Added text/UI overlays.
   Fix: "no text, no logo, no HUD, pure interference frame"

GLOBAL LOCK:
Subject: Young woman, pale complexion, neutral/melancholic expression, short dark wavy bob hair.
Wardrobe: Light yellow/cream camisole with thin straps and ruffled hem.
Environment: Dimly lit, yellowish bedroom with a bed and posters.
Style: Surreal horror, cinematic but with a low-fi/found-footage edge.
Lighting: Sickly yellow indoor lighting, high-contrast shadows, occasional harsh flash.
Color Grade: Desaturated, warm tints, film grain, slight motion blur.
Speech: Single speaker, calm educational tone, mechanical/animatronic cadence.

[00:00–00:02]
Subject: The woman sits centrally on the edge of a bed, looking directly at the camera.
Action: A heavy medieval sword falls vertically from the top of the frame.
Camera: Static medium shot.
Lighting: Dim, overhead yellow light.
Motion: The sword accelerates downward.

[00:02–00:04]
Action: The sword strikes the woman's head. Upon impact, her body splits perfectly down the middle into two giant, vertical loaves of French baguette.
Environment: Crumbs and bread fragments explode outward into the air.
Motion: The two baguette halves tilt away from each other.
Speech: A distorted, breathy gasp.

[00:04–00:06]
Subject: A terrifying animatronic baby doll with a bloody mouth, exposed neck wires, and large blue eyes.
Environment: A colorful classroom background with yellow chairs.
Action: The baby's mouth moves in a mechanical, jerky fashion.
Camera: Medium close-up, centered.
Speech: "And now children..." (Calm, teacher-like tone).

[00:06–00:08]
Environment: A dark kitchen.
Action: Glowing, molten orange lava flows rapidly across the floor toward the camera.
Camera: POV, low angle, shaky handheld movement.
Lighting: Intense red/orange glow from the lava, deep shadows elsewhere.

[00:08–00:09]
Subject: Return to the animatronic baby doll.
Action: Slight head tilt, eyes blinking out of sync.
Speech: "...we learn..."

[00:09–00:11]
Subject: Multiple pale, hairless humanoid creatures with gaunt faces.
Environment: A dark forest at night.
Action: The creatures crawl rapidly on all fours toward the camera through mud and leaves.
Camera: Night vision, green tint, grainy, handheld "found footage" style.
Lighting: Single bright spotlight (camera light) hitting the lead creature.

[00:11–00:17]
Subject: The animatronic baby doll at the front of the class.
Environment: Classroom, back of students' heads visible in the foreground.
Action: The baby speaks with wide, exaggerated mouth movements.
Speech: "...about the father of lies." (Slow, deliberate cadence).
Sync: High lip-sync strictness on "father" and "lies."

[00:17–00:18]
Subject: A man in a dark suit and blue tie.
Action: His mouth opens unnaturally wide, and thick, black, oily tentacles erupt outward toward the lens.
Camera: Close-up, sudden jump-cut.
Lighting: Harsh, direct flash photography style, black background.

NEGATIVE PROMPT:
Smooth plastic skin, bright happy colors, stable video, clean edges, 3D animation look, cartoonish, high-key lighting, friendly faces, slow movement, robotic voice without emotion, perfect symmetry, watermark, text.

SPEECH PACK:
Transcript: "And now children... we learn about the father of lies."

TAKE_A (Educational): [00:04] And now children... [00:08] we learn... [00:12] about the father of lies. (Paced like a storybook reading).
TAKE_B (Sinister): [00:04] And now children... [00:08] we *learn*... [00:12] about the father... of lies. (Heavy emphasis on "learn" and "lies").
TAKE_C (Mechanical): [00:04] And-now-children... [00:08] we-learn... [00:12] about-the-father-of-lies. (Flat, monotone, robotic).

Prosody:
"And now children..." (Rising intonation, pause)
"we learn..." (Slightly lower pitch)
"about the father of lies." (Falling intonation, cold delivery)

HIGH-GRANULARITY INVENTORY
- Subject(s): No people; pile of VHS tapes on wooden surface.
- Clothing & materials: N/A
- Props/objects: Multiple black VHS cassettes stacked chaotically, each with red word "BURN" on labels, moody smoke/haze.
- Environment: Dark indoor tabletop scene with vintage cinematic mood.
- Composition: Vertical close-up, center tape angled toward viewer, surrounding tapes partially occluded.
- Lighting: Low-key directional light from upper left, soft haze bloom.
- Color palette: Dark charcoal blacks with red label accents and muted brown wood.
- Image style: Moody still-life product shot / retro analog aesthetic.

MASTER PROMPT (English)
[Subject] Photograph a dramatic pile of black VHS cassettes stacked on a wooden table, each labeled with the red word BURN.
[Environment] Use a dark smoky studio backdrop with minimal distractions.
[Composition/Camera] Vertical close composition with one central tape sharply in front and others layered behind.
[Lighting] Low-key cinematic side light with subtle haze and soft vignetting.
[Style/Rendering] Retro analog still-life, gritty and atmospheric.
[Detail constraints] do not add/remove objects; match counts, placement, and materials; keep background consistent. Keep multiple VHS tapes and repeated BURN labels; no people or modern digital devices.

Negative prompt: extra people, wrong object count, incorrect perspective, lowres, blur smear, watermark, random text, logo drift, overprocessed skin, cluttered background, bright high-key studio, DVDs, clean white background, missing red label text

Suggested parameters for reproducibility:
- aspect ratio: 9:16
- lens/focal length: 50mm macro-ish
- depth of field: shallow f/2.8
- steps: 24-32
- CFG/style strength: 6-7
- sampler: DPM++ 2M
- seed suggestion: 378015942

Delta prompt strategy (top 10 drift-prone details):
1. Drift detail: Tape type | Corrective micro-prompt: classic black VHS cassettes only
2. Drift detail: Label word | Corrective micro-prompt: red uppercase BURN text on multiple tape labels
3. Drift detail: Stack arrangement | Corrective micro-prompt: irregular layered pile with one tape angled on top
4. Drift detail: Background darkness | Corrective micro-prompt: dark low-key backdrop with subtle smoke haze
5. Drift detail: Surface | Corrective micro-prompt: worn wooden tabletop visible at bottom
6. Drift detail: Lighting direction | Corrective micro-prompt: soft side light from upper-left
7. Drift detail: Color control | Corrective micro-prompt: mostly black/gray with red text accents only
8. Drift detail: No people | Corrective micro-prompt: remove hands and faces entirely
9. Drift detail: No modern media | Corrective micro-prompt: no USB drives, no discs, no screens
10. Drift detail: Sharp focus anchor | Corrective micro-prompt: front-center tape sharp, rear tapes softer

GLOBAL LOCK: Cinematic horror aesthetic, 90s VHS/CCTV grain, low-key lighting, high contrast, muted colors with vibrant red blood. Subject consistency: Pale skin, distorted features, dark hair. Environment: Cluttered labs, server rooms, liminal bedrooms.

[00:00–00:02] Medium shot of Harley Quinn (Margot Robbie style), platinum blonde hair with teal ponytail, wearing a white t-shirt with red trim and a red skirt. She stands in a dark doorway, looking into a dimly lit room with a look of anticipation. Cinematic lighting, high contrast.

[00:02–00:04] Medium shot of a giant, pale human head attached to a tiny body in a white jumpsuit. The figure stands in a cluttered, clinical lab. A thick, continuous stream of dark red blood pours from its mouth. Static camera, fluorescent lighting.

[00:04–00:09] A pale, multi-armed humanoid creature crouched in a dark server room filled with wires. A scientist in a white coat is visible in the foreground typing on a laptop. Sudden bolts of white electricity strike the creature's head, causing it to writhe and glow. High-speed motion.

[00:09–00:12] Wide shot, low angle, of a massive pyramid made of hundreds of people in dark clothing, piled high in a modern city square. The sky is overcast and gloomy. Subtle movement of the people in the pile.

[00:12–00:16] Close-up of a creature with a cracked, skull-like mask and long dark hair. A thick, twisting fleshy umbilical cord or tentacle erupts from a hole in the top of its skull, reaching toward the ceiling. Organic, wet textures.

[00:16–00:17] Quick cut to a multi-armed creature sitting at a desk with multiple glowing CRT monitors, typing frantically with all hands. Dark, shadowy room.

[00:17–00:22] Low angle shot looking up at a metal ceiling vent. Dark red blood begins to drip, then pour out in a heavy stream, drenching a pale creature sitting directly beneath it.

[00:22–00:28] Over-the-shoulder shot of a girl with black pigtails sitting on a bed, watching a small TV. On the screen, a pale, bloody man with long black hair crawls toward the camera. The man then physically crawls out of the TV screen into the bedroom, blood pooling on the floor. High-tension movement.

NEGATIVE PROMPT: bright colors, sunny, happy, clean, high-definition, smooth skin, 3D animation look, watermark, text, logos, friendly faces, stable anatomy, daylight.

SPEECH PACK:
[00:00-00:28]
Transcript: (No speech, only ambient horror soundscape)
TAKE_A: Low-frequency industrial drone with wet squelching sounds and electrical buzzing.
TAKE_B: High-pitched cinematic riser ending in a distorted TV static sound.
TAKE_C: Whispering voices layered over a heavy, rhythmic heartbeat.
Prosody: N/A (Non-verbal)
Sync: Sound effects must align with blood splashes and lightning strikes.

Create a vertical 9:16 poetic dark-fantasy hospital tableau in desaturated near-monochrome tones. In a quiet clinical room under flat ceiling lights, an ancient gaunt figure stands beside a hospital bed holding a large weathered round clock with cracked face and Roman numerals. The figure should look like a personification of time: bald, deeply lined skin, severe expression, frail but unyielding, wearing worn layered garments that blur the line between patient, monk, and deathless witness. In the foreground, a woman lies in the hospital bed looking weak and distant, slightly out of focus compared to the looming clock-bearer. The image should feel like the instant between judgment and reprieve, as if time itself has paused to negotiate one final promise. Keep the mood solemn, intimate, and metaphysical, with no subtitles, no extra logos, no cartoon styling.

[Inventory]
- Subject(s): no people; a pile/stack of VHS videocassettes in the foreground.
- Clothing & materials: none.
- Props/objects:
  - Multiple black VHS tapes (at least 6) scattered/stacked at different angles.
  - Each tape has a label with bold red text reading "BURN".
  - One small flame (a lit match or small burning wick) near the top center, creating a warm highlight.
  - Wooden surface beneath (rough, dark wood planks/board) with visible grain.
- Environment: dim, moody interior; smoky/hazy atmosphere; no clear room context.
- Composition: vertical portrait (9:16); close-up still life; central tape dominates mid-frame; other tapes create layered depth; vignette/dark falloff around edges.
- Lighting: low-key cinematic lighting; cool/neutral ambient with a warm flame accent; soft directional light from upper-left; shallow highlights on plastic; deep shadows.
- Color palette: charcoal/black tapes; desaturated brown wood; bright red typography; small warm orange/yellow flame.
- Image style: cinematic photo/AI-film-still look; slight film grain; high contrast; shallow depth cues.
- Text/typography: "BURN" in bold uppercase red on multiple labels (repeated).

[MASTER PROMPT]
[Subject] A cinematic close-up still life of a messy stack of black VHS tapes on a rough wooden surface, no people, with one central VHS tape in focus and several others layered behind and below at different angles.
[Objects/Labels] Each VHS tape has a label with bold uppercase red text reading exactly: "BURN" (repeated across multiple tapes). Include visible clear reels/windows on the cassettes.
[Environment] Dark moody interior with subtle haze/smoke, no identifiable background.
[Composition/Camera] Vertical 9:16 frame, close-up top-down three-quarter angle, central tape dominating the middle, layered depth, dark vignette around edges.
[Lighting] Low-key film lighting with a small warm flame near the top creating a warm accent, otherwise cool/neutral shadows, soft directional key from upper-left, deep contrast.
[Style/Rendering] Photoreal cinematic film still, subtle grain, high contrast, tactile plastic texture, moody atmosphere.
[Detail constraints] Do not add people; keep multiple tapes, repeated "BURN" labels, the wooden surface, and the small flame; keep the overall scene dark and cinematic.

[Negative prompt]
people, hands, modern devices, DVD cases, clean bright studio lighting, neon colors, extra text, misspelled words, watermark, cartoon, anime, 3D render look, overly sharp HDR, glossy showroom reflections

[Suggested parameters]
- aspect ratio: 9:16
- lens/focal length feel: 35mm–50mm equivalent
- depth of field: medium (central tape sharp, others slightly softer)
- steps: 28–40
- CFG/style strength: 5–7
- sampler: DPM++ 2M Karras (or equivalent)
- seed: 90355127

[Delta prompt strategy: top drift risks + corrective micro-prompts]
1) Label text becomes wrong: "exact red text on labels: BURN, readable, uppercase"
2) Scene becomes too bright: "low-key lighting, dark shadows, cinematic contrast"
3) VHS tapes become DVDs: "VHS videocassettes with visible reels/windows"
4) Loses wooden texture: "rough wooden planks surface, visible grain"
5) Flame disappears: "small warm flame near top of frame, warm glow"
6) Adds hands/people: "no people, no hands"
7) Colors drift away from red/black: "black tapes, red labels, desaturated brown wood"
8) Background becomes a room: "no identifiable background, dark haze"
9) Composition changes to flat-lay: "three-quarter top-down angle, layered stack depth"
10) Style turns too clean/CGI: "film grain, moody atmosphere, tactile plastic texture"

A vertical prompt-education social post featuring a surreal cinematic still set inside a large traditional Japanese martial arts hall. At the top, place a wide image of a polished wooden dojo lit by afternoon sun through high narrow windows, casting long parallel shadow bars across the floor. On the left, five older Japanese masters sit in silence behind a long low wooden table. At center-right, a small white cat in a crisp white gi and fresh yellow belt stands alone on the open floor as if undergoing a formal martial arts examination. The atmosphere is still, symmetrical, and absurdly serious. Beneath the image, add a yellow heading “Prompt” followed by a dense block of small instructional text describing the hall, the judges, the cat’s kata performance, the 1970s film look, and the deadpan pacing. Finish with a glowing yellow CTA line: “Save this post!” Black background, static poster layout, cinematic AI prompt education content.

GLOBAL LOCK: 
The video must maintain a consistent cinematic horror aesthetic with high-contrast lighting, clinical white highlights, and deep, ink-black shadows. The subject identity in the first shot is a pale-skinned young woman with dark hair, vibrant purple eyeshadow, and a silver chain necklace. The second subject is a male with short brown hair in a black t-shirt. The environment transitions from a dark, cluttered room to a clinical hallway, then a night highway, and finally a war-torn urban setting at night. All textures must emphasize "uncanny valley" realism—fleshy, wet, and slightly translucent. No speech is present, but the audio vibe is a low-frequency, sedated drone.

[00:00–00:03]
Subject: A pale woman with dark hair and heavy purple eyeshadow, silver chain necklace.
Action: Four hands in white latex medical gloves cup her face. The skin of her face begins to stretch outward like elastic taffy, eventually splitting the face into two distorted, mirrored versions of herself with visible teeth and gums.
Camera: Extreme close-up, static.
Lighting: Harsh, top-down clinical white light.
Motion: High-intensity skin stretching and tearing physics.

[00:04–00:06]
Subject: A man resembling Mark Zuckerberg, wearing a plain black t-shirt.
Action: His face is a smooth, featureless mask of skin—no eyes, no nose, no mouth. He walks forward through a hallway.
Environment: A dimly lit, industrial hallway with overhead fluorescent lights and a camera operator visible on the left.
Camera: Medium shot, tracking backward as he walks forward.
Lighting: Flickering, cold industrial light.

[00:07–00:09]
Subject: A white semi-truck.
Action: The front cabin of the truck is transformed into a giant, fleshy human skull. It has massive teeth, exposed gums, and a long, wet pink tongue lolling out of the mouth area.
Environment: A dark highway at night.
Camera: Low-angle side-tracking shot, moving at the same speed as the truck.
Lighting: Dim ambient night light with highlights on the fleshy textures.

[00:10–00:12]
Subject: A massive, pale human face with piercing blue eyes and a bloodied mouth.
Action: The giant face looms over the top of a multi-story concrete building. In the foreground, three soldiers in desert camouflage and helmets fire assault rifles at the building.
Environment: A dark, urban combat zone at night.
Camera: Wide shot, static.
Lighting: Dark night scene punctuated by bright, rhythmic muzzle flashes from the rifles.
Motion: Muzzle flashes, smoke, and the subtle, menacing movement of the giant face.

NEGATIVE PROMPT: 
Cartoonish style, low resolution, bright colors, happy atmosphere, stable human anatomy, smooth transitions, friendly faces, daylight, clean environments, robotic movement, text, watermarks, logos, blurry textures, low contrast.

SPEECH PACK:
(No speech present in reference video. Audio is non-verbal ambient horror.)
TAKE_A: [Low frequency sub-bass drone]
TAKE_B: [Distorted, slowed-down industrial hum]
TAKE_C: [Wet, squelching foley sounds synchronized to the skin-stretching moments]

GLOBAL LOCK: vertical social prompt-demo post with a split layout, top half showing a photoreal cinematic 1970s Japanese schoolyard scene in heavy rain, bottom half a persistent black prompt card with yellow-white copy. Main visual concept: a small orange tabby cat wearing a soaked white Japanese school shirt and dark trousers, standing barefoot in gravel while three older Japanese teenage boys in matching short-sleeve white school shirts and dark trousers surround it in a loose semicircle. Setting is a small wooden school building with corrugated iron roof, rusted tetherball pole, late-afternoon downpour, muted grey-green palette, degraded Fujifilm 16mm texture, real rain on lens, deadpan tense mood. Bottom text remains readable and unchanged throughout as a shareable AI prompt resource.
[00:00-00:04] Top half establishes the rain-soaked schoolyard from a sheltered or framed doorway viewpoint, the orange tabby-cat child figure bent low in the gravel while three teenage boys stand around it, all drenched, rusty roof and school wall behind, prompt card fixed on the lower half.
[00:04-00:08] The cat-like figure shifts and wipes water from its face or braces itself, boys subtly adjust posture and step within the semicircle, rain continues striking the gravel hard, 16mm-style softness and grime preserved, prompt text still locked below.
[00:08-00:12] The cat figure backs toward the school wall or side post as the boys hold position and watch, puddles and rain craters visible in the yard, scene maintains tense observational framing above the persistent prompt card.
[00:12-00:15] Final hold on the same confrontation dynamic: the cat in soaked school uniform low and vulnerable, the three boys still surrounding it in the storm, bottom prompt card remains visible as the post ends like a shareable AI cinema prompt example.

GLOBAL LOCK: vertical prompt-demo social post with split layout, top half showing a moonlit cedar forest training scene outside Nara at genuine midnight, bottom half a persistent black prompt card with yellow-white text and bright yellow CTA reading 'Comment AI for prompts'. Top sequence uses only real full-moon illumination through tall cedar trunks, cool blue-black shadows, strong 1970s Japanese cinema mood. Main subject is a brown tabby cat in black training clothes performing strikes against tree trunks and moving through patches of moonlight. Secondary subject is a 50-year-old Japanese trainer wrapped in a dark blanket sitting cross-legged on a broad flat boulder, holding a small oil lamp with the wick turned very low. No artificial light, only moonlight and the tiny lamp glow.
[00:00-00:04] Establish the genuine midnight cedar forest with full moon visible through the canopy, then reveal the tabby cat in dark training clothes darting among the trunks and striking bark in a fast, disciplined martial-arts rhythm above the static prompt card.
[00:04-00:08] Cut to the older Japanese trainer seated on a flat stone wrapped in a blanket, small lamp glowing beside him, posture calm and observant, deep forest blackness behind, prompt text fixed below.
[00:08-00:12] The cat returns into frame near the trainer, tail raised, movement slowed after practice, moonlit fur flickering as clouds pass the moon, the trainer remains still and silent, bottom prompt panel unchanged.
[00:12-00:15] Final hold on the quiet aftermath: trainer on the boulder, tabby settling beside him in the cedar darkness, moonlight and tiny lamp providing the only illumination while the prompt card and comment CTA remain visible until the end.

GLOBAL LOCK: A series of grotesque, horror-themed entities in various domestic and industrial settings. The visual style is "Analog Horror" or "Found Footage," characterized by heavy 35mm film grain, low-resolution textures, high contrast, and moody, low-key lighting. The color palette is desaturated with earthy browns, sickly grays, and deep blacks. All creatures share a pale, sickly skin tone and distorted, wide-mouthed facial expressions. Camera movement is handheld and shaky, mimicking a person filming in fear.

[00:00–00:01] 
A wide shot of a cluttered storage room packed with cardboard boxes and white plastic bags. On the ceiling, a stylized, dark mouth with red dripping paint is visible. Suddenly, a pale, limbless humanoid torso appears hanging from a thick rope noose in the center of the frame. The lighting is flat and fluorescent.

[00:01–00:02] 
A jump-cut to a closer view of the hanging torso. The creature has a distorted, screaming face with dark, hollow eyes. The skin is pale and textured. The background remains the cluttered storage room.

[00:02–00:04] 
A medium close-up of a pale, fleshy creature. Its chest cavity is surgically split open, revealing raw, red muscle and white bone. Inside the cavity, several delicate, blooming white flowers (like lilies) are growing. The lighting is warm and dramatic, coming from the side. The creature's face is pale with a wide, toothy grin.

[00:04–00:07] 
A low-angle, handheld shot in a dark bedroom. A messy bed with gray sheets is in the background. From the dark gap under the bed, a pale face with wide, staring eyes and a massive, blackened grin slowly emerges. Long, stringy black hair frames the face. The lighting mimics a shaky flashlight beam hitting the face directly.

[00:07–00:09] 
A medium shot of a charred, skeletal humanoid figure standing in a dark basement. The figure's skin is black and textured like burnt wood, with a wide-open, screaming mouth. Behind it are blue plastic tarps and wooden rafters. The lighting is harsh and overhead, creating deep shadows in the eye sockets.

[00:09–00:11] 
A medium shot of a pale, bloated creature in a dark room with framed paintings on the wall. The creature has a massive, distended mouth. It is spewing a dense, chaotic swarm of dark insects and golden-brown particles directly toward the camera. Shaky cam effect is intense. The lighting is dim, highlighting the particles in the air.

NEGATIVE PROMPT: clean digital look, high definition, smooth textures, bright lighting, cheerful colors, 3D render look, cartoonish, stable camera, symmetrical faces, healthy skin tones, flowers looking plastic, slow motion.

SPEECH PACK:
(No speech present in the video. Audio consists of low-frequency drones and jump-scare stingers.)
TRANSCRIPT: [Silence/Ambient Drone]
TAKE_A: [Low guttural growl]
TAKE_B: [Wet squelching sound]
TAKE_C: [High-pitched digital screech]

[GLOBAL LOCK]
Create a modern Japanese horror short set inside a small apartment with a narrow hallway, compact kitchen, sofa area, dining table, and television. The recurring protagonist is a young Japanese woman with a short black bob haircut, navy long-sleeve top, and anxious exhausted expression. The haunting force is a long-haired female ghost in a loose white dress with pale skin, plus shadowlike manifestations and impossible positions that suggest she is everywhere and nowhere at once. Keep the apartment layout and the woman consistent throughout. Vertical 9:16, 43 seconds, cinematic horror realism, cool blue nighttime grade, minimal warm practical light in selected early beats, oppressive domestic silence.

[SHOT 1 | 00:00:00-00:00:04.00]
Open with eerie domestic unease in the hallway and bathroom threshold. Show a ghostlike figure crouched or clinging in an impossible position near the ceiling or wall, then cut to a dark corridor where a distant silhouette stands at the far end. The geography should feel narrow and claustrophobic.

[SHOT 2 | 00:00:04.00-00:00:08.00]
Use partial-obstruction shots through blinds or slats to suggest the entity without giving a full clean reveal. Cut to strange movement near the ceiling and corners. Keep the image dim and uncertain, with the apartment feeling watched from multiple angles.

[SHOT 3 | 00:00:08.00-00:00:12.00]
Transition to the kitchen where the woman sits or stands with a bowl of food under cooler light. She looks over her shoulder as if sensing someone behind her. The room is ordinary, which makes the next disturbance more violent.

[SHOT 4 | 00:00:12.00-00:00:16.00]
The ghost suddenly appears deeper in the kitchen or living area as a dark-haired shape. A black impact-like burst or supernatural force erupts near the cabinets, and the woman is thrown backward through the kitchen space. Plates and objects scatter as the haunting becomes physical.

[SHOT 5 | 00:00:16.00-00:00:21.00]
Continue the apartment chaos. The woman falls or scrambles near the sofa while a shadow form moves through the blue-lit living room. The entity seems to occupy multiple positions across the room, matching the line "They are everywhere, they are nowhere."

[SHOT 6 | 00:00:21.00-00:00:27.00]
Escalate with the television. Show static or bright screen noise, then the long-haired woman in white emerging or climbing out of the TV into the room. Keep this sequence iconic and unmistakably cursed-media horror, with dark fluid shadow growth spreading across walls and furniture.

[SHOT 7 | 00:00:27.00-00:00:34.00]
The living room becomes overrun with spectral presence. Show the ghost in full figure, shadow swarms or duplicate dark masses in the background, and the victim cornered on the floor. The woman in white advances with heavy hair obscuring the face while the room darkens further.

[SHOT 8 | 00:00:34.00-00:00:39.00]
Move into intimate terror close-ups: the woman's face crying and screaming near a wall, then the ghost looming directly over her with wet hair and corpse-pale features. Use drips or streaks on the wall behind her to intensify the sense of contamination.

[SHOT 9 | 00:00:39.00-00:00:41.00]
Hit a final monstrous eye close-up in near darkness. The iris should feel inhuman and singular, like the last impossible stare before blackout.

[SHOT 10 | 00:00:41.00-00:00:43.18]
Return briefly to the woman holding her bowl in shocked silence as if trapped in a nightmare loop, then end on black with the title text "Don't Sleep."

[CAMERA]
Use a mix of slow dread framing, reactive handheld panic, hallway long-lens compression, low room-level angles, and TV-facing compositions. Keep the apartment readable but never safe.

[LIGHTING]
Cool blue nighttime ambience dominates the apartment, with occasional warm practical kitchen or lamp light in the early domestic scenes. Let the TV provide a harsh screen glow during the cursed-media sequence. Shadows should deepen as the haunting escalates.

[GRADE]
Cold blue-gray horror palette with desaturated interior colors, pale skin highlights, black hair mass, and occasional white-screen TV contrast. Keep the image naturalistic but nightmarishly drained.

[MOTION]
The protagonist moves in hesitant turns, startle reactions, scrambling falls, and collapsed floor-level fear. The ghost moves through stillness, sudden proximity changes, ceiling or wall placement, crawling emergence from screens, and slow looming approaches.

[SPEECH / AUDIO]
No spoken dialogue. Use apartment ambience, distant hum, floor creaks, bowl and plate clatter, TV static, sudden impact bursts, wet hair movement, sub-bass dread, and one piercing sting for the TV emergence. The final text beat should land on near-silence.

[NEGATIVE PROMPT]
No bright daylight, no gothic mansion, no comedy horror, no excessive gore, no zombie makeup clichés, no modern jump-cut meme editing, no extra human cast, no colorful lighting, no possession exorcism tropes, no anime styling, no text except the final "Don't Sleep", no unreadable dark mush.

[GLOBAL LOCK]
Create a Japanese horror short set inside a traditional tatami room with shoji doors, wooden beams, sliding panels, and a small glowing TV in the corner. The recurring victim is a young man in a white short-sleeve shirt and dark pants, terrified and stumbling backward through the room. The recurring ghost is a female J-horror entity with long black hair, pale skin, a loose patterned robe, bare feet, and a grotesque head gimmick that shifts between upside-down, detached-looking, or unnaturally rotated orientation. Her eyes glow red in key reveal shots. Keep the same room layout and the same two characters throughout. Vertical 9:16, cinematic horror realism, handheld panic energy, dim warm practical light, deep shadow corners, oppressive indoor atmosphere.

[SHOT 1 | 00:00:00-00:00:02.50]
Open with the man moving nervously across the tatami room near the shoji doors. He glances toward the back of the room as if sensing something behind him. Camera follows with unstable handheld movement, keeping the room geometry clear.

[SHOT 2 | 00:00:02.50-00:00:04.50]
The ghost suddenly appears deeper in the room, hair hanging forward, body low and unnatural. Her head orientation is wrong and immediately unsettling. The man recoils. Keep the reveal fast and invasive, with the ghost advancing toward camera.

[SHOT 3 | 00:00:04.50-00:00:06.50]
Cut to a tighter horror image of the ghost's face or head with glowing red eyes. Her expression is inhuman and corpse-like. Use dim room light and shadow to make the red eyes the brightest point in frame.

[SHOT 4 | 00:00:06.50-00:00:09.00]
The camera whips across the room and catches the empty tatami space, then the ghost reappears lower to the ground, crawling or crouching in a new position. The man backs away toward the corner where the small TV throws a weak glow.

[SHOT 5 | 00:00:09.00-00:00:11.50]
The confrontation escalates as the ghost rushes closer. Her hair lashes outward, her robe drags across the floor, and the victim is forced toward the sliding doors. Show frantic close-quarters pursuit inside the tight room.

[SHOT 6 | 00:00:11.50-00:00:14.00]
The ghost launches into impossible movement, climbing or sticking to the wall and then appearing on the ceiling line. Her long hair drops downward as gravity seems wrong. This is the signature J-horror set piece and should feel abrupt and deeply unnatural.

[SHOT 7 | 00:00:14.00-00:00:16.99]
End in a near-black face close-up of the ghost with glowing red eyes and a stretched, distorted mouth. Overlay the in-world text "Don't lose your head" as the final stinger. The last image should feel like a cursed-screen jump scare rather than a clean title card.

[CAMERA]
Handheld panic camera with sharp reframes, short pushes, and sudden angle changes inside a confined room. Keep the geography readable enough to understand where the man and ghost are in relation to the tatami floor, doors, and ceiling.

[LIGHTING]
Dim warm practical interior light with some cold daylight leaking through shoji paper. Let the TV add a weak secondary glow in one corner. Eyes should glow saturated red only in the strongest scare beats.

[GRADE]
Muted brown, beige, and shadowy charcoal palette with low-saturation traditional interior tones. Preserve the warm wood ceiling, pale tatami mats, and patterned robe while allowing the red eyes to break the palette violently.

[MOTION]
The human moves in frightened stumbles, backing turns, and evasive pivots. The ghost moves in jerks, crouched glides, sudden lunges, wall-crawls, and inverted hanging poses. Hair motion should enhance every change of direction.

[SPEECH / AUDIO]
No spoken dialogue. Use creaking floorboards, breathy room tone, TV hum, sudden thuds, hair movement, scratchy impact sounds, and a sharp horror sting when the ghost appears on the ceiling. The final stinger should be supported by low sub-bass and a piercing high-frequency hit.

[NEGATIVE PROMPT]
No bright daylight, no comedy horror tone, no clean modern apartment, no extra ghosts, no gore splash, no zombie makeup cliché, no possession VFX, no superhero motion, no anime styling, no text except the final "Don't lose your head", no blurry unreadable ghost design, no duplicated limbs.