skills/ai-film-pipeline-master/references/phase-07-image-prompt-engineering/vertical-short-form.md

# Vertical Short-Form

25 cards. These decide a frame: what is in it, how it is lit, and what it is made of.

Every card carries what it is, what it does to the audience, where it works best, what to avoid, and the prompt written out in full.

---

### The Hook Window / First Three Seconds

**Also called:** 0 to 3, the scroll gate  
**What it is:** The first three seconds are budgeted separately: a strong opening frame, motion, and a clear reason to stay. Everything else is pushed past second three.  
**Effect on the audience:** The platform decides distribution from how many viewers clear the opening seconds, and the viewer decides in the same window. This is where the video is won or lost.  
**Used for and where it works best:** Plan every reel on this split: hook (0 to 3 seconds), promise, delivery, loop. In vertical micro-drama open on the problem, not the introduction; in a vertical ad let the product or the result appear inside the first second. Every second of set-up spent here costs you half the audience.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Meme / Comedy Edit, Vertical Micro-Drama, Commercial / TVC, Product Video / Motion Design, Music Video, Kids Content | genres: Comedy, Drama, Thriller / Suspense, Slice of Life, Family, Science / Educational, Romance, Action  
**Avoid when:** Do not treat it as decoration added at the end. A logo or an intro card — a separate screen ahead of the picture — inside this window ends the video before it starts; a short title laid over a first shot that is already moving is `vertical.title_card_zero`, not an intro card.  
**Prompt:** `open on the single strongest image of the story, full frame, motion already happening, no establishing beat`  
**Source:** reference — TikTok for Business, Video Advertising Guide for Small Businesses, https://ads.tiktok.com/business/en/guides/video-advertising-guide, checked 2026-09-24; supports hooking the viewer in the first three seconds and skipping long intros or logo animations, not the claim about how distribution is ranked, which Shelf life already dates.  
**Shelf life:** dated — review 2027-03. That the platforms decide distribution from the opening seconds is their ranking, not a craft fact, and it changes when they change it; the viewer's own decision in that window is the durable part.  
`vertical.hook_window`

### Visual Hook

**Also called:** scroll stopper, image hook  
**What it is:** The opening frame is a strange image, a finished result or a sight the viewer does not meet every day, with no dialogue and no set-up. The picture alone stops the thumb.  
**Effect on the audience:** It works before the viewer reads a word or hears a sound, which makes it the only hook that still lands with the sound off.  
**Used for and where it works best:** Best for food, craft, travel and product: the cake cut open, the dye pouring, the city from above, the artefact lit in the dark. In micro-drama make it the frame with the wrong thing in it: water flooding, a broken door, a person standing where they do not belong.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Commercial / TVC, Product Video / Motion Design, Travel / Real Estate / Hospitality, Kids Content, Music Video, Meme / Comedy Edit | genres: Slice of Life, Family, Science / Educational, Comedy, Nature / Wildlife, Heritage (civilisation-focused), Romance, Thriller / Suspense  
**Avoid when:** If the image needs an explanation to read, it is not a hook. A striking frame with no link to the rest of the video buys a view and loses a follower.  
**Prompt:** `first frame is the finished result in close-up, filling the vertical frame, bright and high contrast, slight motion`  
**Yields to:** `vertical.text_hook` — The image needs explaining to read.  
`vertical.visual_hook`

### Face In The First Frame

**Also called:** human first, face rule  
**What it is:** A full, clearly lit human face is present in the opening frame, not a landscape, a logo or text on an empty background.  
**Effect on the audience:** The human eye locks onto faces faster than any other shape, and a face looking down the lens creates a feeling of being addressed directly.  
**Used for and where it works best:** A practical rule for talking content, advice, commentary and clipped interviews. Keep the face in the upper third and large enough to read on a small screen. In heritage content that hides faces, substitute a working human hand so the human element is still there.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Vlog / YouTube Long-Form, Interview / Video Podcast, Commercial / TVC, Explainer, News Report / Reportage, Vertical Micro-Drama | genres: Slice of Life, Comedy, Drama, Science / Educational, Family, True Crime, Romance  
**Avoid when:** Do not force it on content whose subject is a product or a place; a face jammed in reads as an intruder. A small face lost in the frame is worse than no face at all.  
**Prompt:** `vertical medium close-up, face in the upper third looking straight into the lens, bright soft key, background simple and dark`  
**Yields to:** `vertical.visual_hook` — Subject is a product or a place.  
`vertical.face_first_frame`

### Silent First Second Rule

**Also called:** sound off design, muted opening  
**What it is:** You design the first second to work fully with the sound off: a clear image and large text that carries the idea without being heard.  
**Effect on the audience:** A large share of views begin muted, so a video whose idea lives only in the audio loses that entire audience without you noticing.  
**Used for and where it works best:** Apply it to every reel: watch the first second with your eyes only and ask whether the idea lands. Add text or an image that carries the first spoken line. In a vertical ad, put the product name or the problem on screen rather than only in the voice.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Commercial / TVC, Explainer, News Report / Reportage, Product Video / Motion Design, Video Essay, Corporate / Training | genres: Science / Educational, Slice of Life, Comedy, Drama, Family, True Crime  
**Avoid when:** Do not turn the whole video into text; that kills the audio for the viewers who unmute. Designed text that repeats the spoken line word for word becomes visual filler after the opening; the first line is the exception, written to work as both (`sf.first_line_both_ways`). The caption track is a different layer, and it carries every spoken word by design (`vertical.captions`).  
**Prompt:** `opening frame readable with sound off, large clear caption plus a self explaining image, no reliance on voice over`  
**Yields to:** `vertical.visual_hook` — A self-explaining image beats more text.  
`vertical.silent_first_second`

### Cover Frame / Thumbnail Frame

**Also called:** cover image, feed thumbnail  
**What it is:** The frame shown in the grid and on the profile before playback. It is either chosen from inside the video or designed separately in the same ratio.  
**Effect on the audience:** It is the video's static advertisement and it decides a large share of the views that arrive from the profile and from search rather than the feed.  
**Used for and where it works best:** Pick a frame with a clear face or a striking result and two to four words of large text. Remember the profile grid crops the sides, so keep the important part centred. For a series, standardise the design so the profile grid reads as one body of work.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Vertical Micro-Drama, Explainer, Vlog / YouTube Long-Form, Video Essay, Kids Content, Music Video | genres: Slice of Life, Science / Educational, Drama, Comedy, Family, Heritage (civilisation-focused), True Crime  
**Avoid when:** A frame caught mid-movement or with a half closed eye usually kills the click, because the face reads as a mistake at thumbnail size; the exception is an action cover where the blur is the promise. A cover that promises something the video does not contain damages trust in the account.  
**Prompt:** `single clean vertical frame with the subject centred, sharp face or hero result, space at top and bottom for overlay text`  
**Shelf life:** dated — review 2027-03. How the profile grid crops a cover is the platform's interface, not a craft fact, and it changes when the platform redesigns the grid; keeping the important part centred is the durable part.  
`vertical.cover_frame`

### Hook Stack

**Also called:** layered hook, double hook  
**What it is:** More than one hook in the same instant: a strong image plus a short caption plus a sound hit, all three on the first frame.  
**Effect on the audience:** It covers every kind of viewer at once: the reader, the looker and the listener. The chance of a stop rises because every channel is working.  
**Used for and where it works best:** The working rule for a serious reel: the result image, a written question and the recognisable opening of a sound. In a vertical ad: the product in a human hand, a caption naming the problem and the real sound of using it. Never go past three layers or the frame becomes crowded.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Commercial / TVC, Product Video / Motion Design, Meme / Comedy Edit, Explainer, Vertical Micro-Drama | genres: Comedy, Science / Educational, Slice of Life, Drama, Family, Action  
**Avoid when:** Four layers or more cancel each other out and the frame turns to chaos. Contradicting hooks, where the caption promises one thing and the image another, usually split the attention and lose both; the exception is a deliberate contradiction the video resolves in the first seconds.  
**Prompt:** `opening frame combines a striking hero image, a two line caption in the upper third and a single sharp audio hit`  
`vertical.hook_stack`

### Call-Out Hook

**Also called:** targeting hook, if you are hook  
**What it is:** A line that calls out a specific group of people, spoken or written over the image: for example anyone who works in this trade.  
**Effect on the audience:** The targeted viewer feels personally addressed and stops; everyone else scrolls on, which is useful because it teaches the algorithm who your audience is.  
**Used for and where it works best:** Best for professional, educational and niche advertising content: designers, cooks, parents of small children, camera operators. Write the call-out in the largest type in the video and keep it above the face and below the top interface band `vertical.safe_zones` reserves. The image under it must prove you really belong to that field.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Commercial / TVC, Explainer, Corporate / Training, Vlog / YouTube Long-Form, Video Essay | genres: Science / Educational, Slice of Life, Comedy, Family, Drama  
**Avoid when:** A call-out as broad as everyone does nothing. A false call-out that does not serve the group it names loses that group's trust.  
**Prompt:** `large caption naming the target group across the upper part of frame one, below the top interface band, presenter already working in that environment behind it`  
**Yields to:** `vertical.question_hook` — No narrow group worth naming.  
`vertical.callout_hook`

### Number Hook

**Also called:** list promise, countable hook  
**What it is:** The opening promises an explicit number: three ways, five mistakes, seven pieces. The number stays written on screen from the start.  
**Effect on the audience:** A number sets a known length and a clear end, so the viewer knows what they are investing and completion rates rise.  
**Used for and where it works best:** Strong for knowledge and marketing content: five facts about an ancient city, three mistakes in filming. Pair it with numbered corner cards so the viewer can follow the progress. Make the last item the strongest one.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Explainer, Video Essay, Corporate / Training, Commercial / TVC, Vlog / YouTube Long-Form, Heritage / Historical Documentary | genres: Science / Educational, Heritage (civilisation-focused), Historical / Biopic, Slice of Life, Family, Comedy  
**Avoid when:** A big number like fifteen frightens the viewer in a short video. Promising a number and delivering fewer is what costs you the next video, because the count is the contract; delivering more is the only safe direction to miss in.  
**Prompt:** `bold numeral and short promise line on the opening frame, small numbered tag stays in the corner through the video`  
`vertical.number_hook`

### POV Caption

**Also called:** POV text, scenario caption  
**What it is:** A short line at the top of the frame places the viewer in a situation so the shot reads as if seen through their own eyes.  
**Effect on the audience:** It turns an ordinary shot into a story, because the text hands the viewer a role and the image completes it. It costs nothing and delivers strong identification.  
**Used for and where it works best:** Best for comedy, everyday situations and imagined historical content: you are standing in an ancient market, you are seeing this place for the first time. Shoot the shot at eye level or straight on, and hold the caption for the whole shot, set just below the top interface band `vertical.safe_zones` reserves, never inside it.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Meme / Comedy Edit, Vertical Micro-Drama, Heritage / Historical Documentary, Kids Content, Music Video | genres: Comedy, Slice of Life, Heritage (civilisation-focused), Historical / Biopic, Romance, Fantasy / Epic, Horror  
**Avoid when:** A caption describing something other than what is on screen breaks the whole device. Overusing it on one account makes the content feel repetitive.  
**Prompt:** `eye level first person framing of the scene with a short scenario caption pinned near the top of the vertical frame, just below the top interface band`  
**Yields to:** `vertical.pov_camera` — Put the viewer inside with the lens.  
`vertical.pov_caption`

### On-Screen Captions

**Also called:** word by word subtitles, animated captions  
**What it is:** Every spoken word appears written at the same instant, in a heavy typeface at large size, usually one word at a time or in short groups. This is the caption track, distinct from hook copy, labels and titles.  
**Effect on the audience:** They make the video fully understandable with the sound off and add a visual rhythm because the text changes with the speech.  
**Used for and where it works best:** Mandatory on any vertical talking video. Use a heavy face with an outline or shadow so it reads over any background, and keep its position fixed in the middle third, or just under it only while it stays clear of the platform interface band that `sound.sound_off` finds over the lower third. For Arabic, respect the right to left flow and keep it to a single line.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Vlog / YouTube Long-Form, Explainer, News Report / Reportage, Video Essay, Corporate / Training, Commercial / TVC, Interview / Video Podcast | genres: Science / Educational, Slice of Life, Comedy, Drama, True Crime, Family, Political Thriller  
**Avoid when:** Captions covering the face or falling into the platform interface band become a burden rather than help. A different colour on every word tires the eye.  
**Edit Overlay Rule:** this text is an editing and graphics overlay added in post-production. It is never burned into the generated image or video frame itself, which is what `heritage-rules.md` rule 5 forbids and what makes the text uncorrectable afterwards.  
**Prompt:** `[Edit Overlay Instruction]: word by word captions in a heavy sans face with outline, fixed position in the middle third, synced to the voice`  
**Shelf life:** dated — review 2027-03. Where the platform interface band sits is the platforms' to move, so the caption position follows it; the band's current depth is written only in `vertical.safe_zones`, and Facebook's in `vertical.facebook_safe_zones`.  
**Yields to:** `vertical.text_beyond_picture` — Text should add, not transcribe.  
`vertical.captions`

### The Caption Bar

**Also called:** fixed title bar, static header  
**What it is:** A fixed text bar in one position for the whole video carrying the title or the topic, which never moves and changes only at a major turn.  
**Effect on the audience:** It provides a constant identity and reminds a viewer who arrives mid-video what they are watching, which reduces mid-video exits.  
**Used for and where it works best:** Strong for series and knowledge content: the series name at the top and the episode number in a corner. In vertical documentary carry the site name or the period. Keep the position and typeface identical across episodes so it becomes a recognisable mark.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Documentary Series, Heritage / Historical Documentary, Explainer, News Report / Reportage, Corporate / Training, Video Essay | genres: Heritage (civilisation-focused), Historical / Biopic, Science / Educational, True Crime, Political Thriller, Nature / Wildlife  
**Avoid when:** A thick bar eats a quarter of the vertical frame and strangles the image. Movement or flashing on the bar usually steals attention from the content; the exception is a single word highlight landing on the beat you want heard.  
**Prompt:** `slim fixed title bar pinned in the upper safe area, same position and typeface for the whole video, no animation`  
**Yields to:** `vertical.title_card_zero` — The frame cannot spare a permanent band.  
`vertical.caption_bar`

### Progress Bait

**Also called:** step counter, progress bar  
**What it is:** A visual indicator showing where the video has reached: a filling bar, or a step counter such as three of five in the corner.  
**Effect on the audience:** It makes the viewer feel close to the end so they finish, and this works even though the actual length has not changed.  
**Used for and where it works best:** Suits lists, tutorials, recipes and challenges. Keep the indicator small in an upper corner, below the top interface band and away from the right-edge buttons `vertical.safe_zones` reserves, and clear of the captions. In recipes tie each step to a clean image so the viewer can pause and read.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Explainer, Vlog / YouTube Long-Form, Corporate / Training, Kids Content, Video Essay | genres: Science / Educational, Slice of Life, Family, Comedy, Sports Drama  
**Avoid when:** An indicator promising five steps and delivering three breaks trust. In a dramatic story it reveals the length of the scene and kills the suspense.  
**Prompt:** `small step counter in an upper corner below the top interface band and away from the right-edge buttons, updating on each new step, thin progress bar filling just above the bottom interface band`  
**Yields to:** `vertical.watch_till_end` — Drama; showing the length kills suspense.  
`vertical.progress_bait`

### Sound Sync

**Also called:** SFX sync, hit sync  
**What it is:** Every visual event is locked to a precise sound effect: text appearing with a click, a product entering with a whoosh, a cut with an impact.  
**Effect on the audience:** It makes movement felt in the body rather than only seen, and gives a polish that raises the perceived value even when the footage is simple.  
**Used for and where it works best:** Essential in motion design, product and explainer work: never let an element enter without a sound. Place the effect two frames before the event so it feels locked. In heritage content use real textures such as stone and water instead of synthetic hits.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Product Video / Motion Design, Commercial / TVC, Explainer, Kids Content, Meme / Comedy Edit, Corporate / Training, Title Sequence / Opening Credits | genres: Science / Educational, Family, Comedy, Action, Slice of Life, Heritage (civilisation-focused)  
**Avoid when:** A sound effect on absolutely everything becomes noise and fatigue. An effect that lags behind its event ruins the whole impression.  
**Prompt:** `every text pop and every object entrance carries its own short sound effect placed two frames early`  
`vertical.sound_sync`

### Green-Screen React / Overlay React

**Also called:** green screen duet, image overlay commentary  
**What it is:** The presenter is keyed out of their background and placed in front of an image, a video or a post, commenting on it and pointing at it.  
**Effect on the audience:** It puts the source and the opinion in one frame, so the viewer sees the evidence and the explanation at the same time.  
**Used for and where it works best:** The first tool of news analysis, historical images and product commentary: put the relief, the map or the photo behind you and point at the detail. Keep yourself in the lower third with the image above, your face clear of the bottom interface band `vertical.safe_zones` reserves. Change the background image every couple of sentences so the frame refreshes.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Video Essay, News Report / Reportage, Explainer, Heritage / Historical Documentary, Vlog / YouTube Long-Form, Meme / Comedy Edit | genres: Science / Educational, True Crime, Heritage (civilisation-focused), Historical / Biopic, Political Thriller, Comedy, Mystery / Whodunit  
**Avoid when:** A dirty key around hair and shoulders looks cheap. Do not put a busy, detailed image behind you; it pulls attention off what you are saying.  
**Prompt:** `presenter keyed into the lower third of the vertical frame, face kept above the bottom interface band, with the reference image filling the area above, pointing at a detail`  
**Yields to:** `vertical.split_react` — The source image is busy or moving.  
`vertical.green_screen_react`

### Split-Screen React

**Also called:** reaction split, picture in picture react  
**What it is:** The screen splits horizontally: the source above, your reaction or explanation below, both playing at once.  
**Effect on the audience:** The viewer sees the event and how they are meant to feel about it at the same instant, which raises the emotional response.  
**Used for and where it works best:** Suits sport, surprising clips and analysis: give the source two thirds of the frame and yourself the bottom third, your face kept above the bottom interface band `vertical.safe_zones` reserves. In teaching content make the lower area a board or a diagram rather than a face. In 9:16 the horizontal split is the correct one, not the vertical split.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Sports Broadcast / Sports Film, Video Essay, News Report / Reportage, Vlog / YouTube Long-Form, Explainer, Meme / Comedy Edit | genres: Sports Drama, Comedy, True Crime, Science / Educational, Slice of Life, Political Thriller  
**Avoid when:** A vertical split inside a vertical frame makes both halves too narrow to read. Do not let your face be larger than the source when the source is the subject.  
**Prompt:** `horizontal split of the vertical frame, source clip filling the top two thirds, presenter reaction in the bottom third, face kept above the bottom interface band`  
**Yields to:** `vertical.green_screen_react` — You need to point into the image.  
`vertical.split_react`

### Clap Transition

**Also called:** clap cut, hand clap wipe  
**What it is:** The presenter claps or brings both palms together in front of the lens, and the cut lands on the frame of contact, changing the wardrobe or the location.  
**Effect on the audience:** It gives a visual and audible reason for a sudden change, so the viewer reads it as a satisfying trick rather than an edit mistake.  
**Used for and where it works best:** Suits wardrobe changes, two-state comparisons and the start of a new segment. Shoot exactly the same framing before and after on a locked camera. Cut on the real clap sound rather than a library effect.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Commercial / TVC, Fashion Film, Kids Content, Meme / Comedy Edit, Music Video | genres: Comedy, Slice of Life, Family, Musical, Romance  
**Avoid when:** If the framing shifts between the two takes, the clap reads as an ordinary visible cut. Avoid it in serious content, where it looks childish.  
**Prompt:** `locked vertical framing, the hands clap in front of the lens and the cut lands on the contact frame with the outfit changed`  
**Yields to:** `vertical.walk_past_cut` — Serious content; the clap reads childish.  
`vertical.clap_transition`

### Walk-Past Cut

**Also called:** body wipe, cross frame transition  
**What it is:** A person or an object passes in front of the lens and fills the frame completely for an instant, and the cut happens on that dark frame.  
**Effect on the audience:** It moves between two places with no effect at all, keeping the feel real and cinematic rather than digital.  
**Used for and where it works best:** Ideal for travel, weddings and vertical documentary: have someone cross close to the lens at the end of the shot and another cross at the start of the next at the same speed and direction. A wall, a column or a hanging cloth works the same way.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Travel / Real Estate / Hospitality, Wedding / Event Film, Vlog / YouTube Long-Form, Heritage / Historical Documentary, Music Video, Vertical Micro-Drama | genres: Slice of Life, Romance, Heritage (civilisation-focused), Adventure, Drama, Musical  
**Avoid when:** If the crossing speeds differ between the two shots the cut looks broken. Do not use it when the passing figure pulls more attention than the subject.  
**Prompt:** `a figure crosses close past the lens filling the frame, cut on the full cover frame into a matching cross in the new location`  
**Yields to:** `vertical.whip_transition` — Nothing available to cross the lens.  
`vertical.walk_past_cut`

### Vertical Stacking

**Also called:** layered vertical composition, top and bottom layout  
**What it is:** You use the height of the frame in layers: text at the top, the subject in the middle, a detail or a hand or a data line at the bottom.  
**Effect on the audience:** It delivers information density without clutter and walks the eye down the frame in a controlled order instead of letting it wander.  
**Used for and where it works best:** The core solution for explainers, comparisons and recipes: title on top, the action in the middle, the current step below. In vertical documentary put the map above, the object in the middle and the site name below. Keep the layer order identical through the video, and keep every layer inside the area `vertical.safe_zones` leaves clear: the top layer sits below the platform's reserved top band and the bottom layer above its reserved bottom band, never in them.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Explainer, Heritage / Historical Documentary, News Report / Reportage, Corporate / Training, Product Video / Motion Design, Video Essay | genres: Science / Educational, Heritage (civilisation-focused), Historical / Biopic, Slice of Life, Political Thriller, Family  
**Avoid when:** More than three layers makes a frame that cannot be read on a small screen. A moving layer and a static layer sharing one band usually fight each other, because the eye goes to the motion and abandons the rest; the exception is when the static layer is only a label for the moving one.  
**Prompt:** `vertical layout in three bands, caption in the top band below the platform's reserved top area, the action centred in the middle, a labelled detail just above the platform's reserved bottom area`  
**Yields to:** `vertical.fill_frame` — Nothing to layer; give the subject everything.  
`vertical.vertical_stack`

### UI Safe Zones

**Also called:** platform safe area, interface margins  
**What it is:** The top band, the bottom band and one side edge are occupied by the platform's buttons, caption and account name, and anything placed there gets covered.  
**Effect on the audience:** Ignoring them produces a video that looks cropped and cluttered on a phone even though it was perfect on the editing monitor.  
**Used for and where it works best:** Keep the reserved bands at the top and bottom free of anything important, and stay off the right edge where the interaction buttons sit. Check every design on a real phone preview, not in the editing software. The same rule applies to the product, the text and the logo. **This card is the only place in the package where these Instagram Reels and TikTok figures are written; Facebook's own, which Meta publishes for Reels and Stories ads, are written only on `vertical.facebook_safe_zones`**, and all of them move. As measured in September 2026: **Instagram Reels reserves roughly 400-430 px of a 1920-tall frame at the bottom — about 22 percent — having grown its audio-attribution bar by some 50 px during 2026**, and about 220 px at the top. **TikTok reserves less at the bottom, roughly 320 px — about 17 percent** — with about 108 px at the top and a deeper right margin for the interaction rail. So **Reels, not TikTok, is the tighter of the two at the bottom**; this card said the reverse until September 2026, and a piece laid out to the old reading put the subtitle under the caption stack. Compose to the *tightest* platform the piece will ship on, take the figure from the platform's own current page before a delivery, and treat the percentages here as the shape of the problem rather than as the answer.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Vertical Micro-Drama, Commercial / TVC, Product Video / Motion Design, Kids Content, Music Video, Explainer, News Report / Reportage | genres: all  
**Shelf life:** dated — review 2027-03. These are platform interface measurements, not craft facts: Reels moved its bottom band during 2026 and the numbers here have been wrong once already. Every other file in the package defers to this card, so this is the single place a correction to these figures has to land; Facebook's are corrected on `vertical.facebook_safe_zones`.  
**Avoid when:** The face, the product or the key point placed inside the reserved bands is usually covered by the interface, and that is the one loss you cannot fix afterwards. Even a busy background there fights with the platform buttons.  
**Prompt:** `compose for a 9:16 frame keeping the top twelve percent and the bottom twenty-two percent clear of the subject and of any text, and the right edge clear of both`  
**Source:** measured — the platforms' own current specification pages, September 2026. This one genuinely was measured, and it is the model for the field  
**Yields to:** AUTHORITY (in words): the platform's own current specification page, checked before each delivery — The card defers its own figures outward.  
`vertical.safe_zones`


### Facebook Reels and Stories Safe Zones

**Also called:** Meta safe zone, Facebook interface bands, Reels and Stories ad safe area  
**What it is:** The interface bands Meta publishes for 9:16 video on Facebook, written as Meta states them, because Reels and Stories are not the same, and neither matches the organic Instagram band that `vertical.safe_zones` measured. **Facebook Reels:** leave roughly 14% of the top, 35% of the bottom and 6% on each side free of text, logos and key creative elements (Meta Ads Guide, Facebook Reels). Meta gives percentages only; at 1080×1920 they come to 268.8 px at the top, 672 px at the bottom and 64.8 px each side, and at 1440×2560, Meta's recommended resolution, to 358.4, 896 and 86.4 px — that arithmetic is this card's, not Meta's. **Facebook Stories:** roughly 14% (250 pixels) of the top and 20% (340 pixels) of the bottom, and no side figure at all (Meta Ads Guide, Facebook Stories); the pixel values Meta prints there are not 14% and 20% of either a 1920 or a 2560 frame, so plan from the percentages. **A Reels ad that carries a disclaimer** keeps the bottom 40% clear (Meta Business Help Center, which covers Facebook and Instagram). **Facebook Feed video** can be vertical, square, 4:5 or horizontal 16:9, and the brief decides, as it does on YouTube; Meta's page for it gives no safe-zone figure. For a horizontal feed video this card's working default keeps text and key elements off the bottom 15% (the player controls and caption overlay) and 5% at each side, recorded under `decisions`. **Every one of these figures is published for ads, and Meta publishes none for organic posts**; for an organic Facebook Reel the ad figure is the closest official figure there is, because the organic Reel carries the same profile, caption and action-icon overlay. **This card is the only place in the package where Facebook's figures are written**; any other file that needs one cites this id.  
**Effect on the audience:** A face, a caption or a logo placed inside Meta's bands is covered by the profile icon, the caption or the call-to-action on the viewer's phone, however clean it looked in the edit.  
**Used for and where it works best:** Any 9:16 piece that ships to Facebook, as a Reel or as a Story, and any piece boosted or run as an ad there. Take the band for the placement the piece actually ships to: the bottom band is 35% on Reels and 20% on Stories, and Stories gives no side margin, so a layout built for Stories puts its text in the Reels bottom band. Meta's own Instagram Reels ad page gives the same three Reels figures, worded "at least" rather than "roughly". On screens taller than 9:16 Meta may zoom the video to fill the screen, which crops whatever sits outside the safe zone, or show it at 9:16 on black. Where a piece ships to several platforms, compose to the tightest band across all of them, as `vertical.safe_zones` instructs — on the bottom edge Meta's Reels figure is deeper than any organic band that card measured. Check the placement against the safe-zone overlay Meta's Help Center points to in Ads Manager, or on a real phone, before delivery.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Vertical Micro-Drama, Commercial / TVC, Product Video / Motion Design, Explainer, Heritage / Historical Documentary, Music Video | genres: all  
**Shelf life:** dated — review 2027-03. These are Meta's figures for Meta's interface, not craft facts. Meta's pages carry no date, and third-party pages already describe a 2026 change that Meta's own Stories page does not show; every figure here is re-read from Meta's page before a delivery.  
**Avoid when:** A figure is taken from a third-party summary instead of Meta's page. Several blogs report that Meta unified the Stories and Reels safe zones in March 2026 and quote pixel values for it; no official page states that date, and Meta's Stories page still read 14% top and 20% bottom on 2026-09-23. Do not carry the 9:16 bands onto a 4:5 Feed video, for which Meta gives no figure. And do not read the ad figures as organic measurements: they are Meta's advice for ads, carried over to organic posts only because nothing closer is published.  
**Prompt:** `compose for a 9:16 Facebook Reel keeping the top fourteen percent, the bottom thirty-five percent and six percent at each side clear of the subject, the text and the logo; for a Facebook Story, the top fourteen percent and the bottom twenty percent` Text treatment: the text and the logo are set in the edit inside the safe area (`textimg.text_safe_zone`).  
**Source:** reference — Meta Ads Guide, the Facebook Reels video page (facebook.com/business/ads-guide/update/video/facebook-facebook-reels) and the Facebook Stories video page (facebook.com/business/ads-guide/update/video/facebook-story), and the Meta Business Help Center article "About text overlays and the safe zone for ads on Facebook and Instagram" (facebook.com/business/help/980593475366490); read on the live pages 2026-09-23, none of them dated. The Reels pixel conversions are this card's arithmetic, not Meta's  
**Yields to:** `vertical.safe_zones` — the piece also ships on Instagram Reels or TikTok: those organic bands are measured on that card, and the layout composes to the tightest band across every platform the piece ships on, as that card instructs. AUTHORITY (in words): Meta's own current Ads Guide page for the placement, checked before each delivery — it wins wherever it and this card differ.  
`vertical.facebook_safe_zones`
### Text-Safe Zone

**Also called:** caption band, typographic safe area  
**What it is:** The middle band between the platform's reserved top and bottom bands is the area where burned text has the best chance of staying readable on every platform the piece ships on; how deep those bands reach is written only in `vertical.safe_zones`, with Facebook's on `vertical.facebook_safe_zones`, and the layout composes to the tightest of them.  
**Effect on the audience:** It gives the title and the captions the best chance of being read on every screen and keeps them clear of the platform interface and the face; it is a design target, confirmed on a preview of the actual export, not a guarantee.  
**Used for and where it works best:** Split the band: the title at its top above the face, the captions at its bottom below the chin. In Arabic keep it to one centred line so it stays inside the band. In heritage video put the subtitles here, since burned text cannot be removed later.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Vertical Micro-Drama, Explainer, News Report / Reportage, Kids Content, Heritage / Historical Documentary, Commercial / TVC, Lyric Video | genres: all  
**Avoid when:** Text running to the frame edge gets clipped on some devices in full-screen playback. Do not stack two text blocks in one band; one of them will be covered.  
**Edit Overlay Rule:** this text is an editing and graphics overlay added in post-production. It is never burned into the generated image or video frame itself, which is what `heritage-rules.md` rule 5 forbids and what makes the text uncorrectable afterwards.  
**Prompt:** `[Edit Overlay Instruction]: place all overlay text inside the central safe band, single centred line, clear of the face and of the platform bars`  
**Shelf life:** dated — review 2027-03. Which devices clip text at the frame edge in full-screen playback is device and platform behaviour, not a craft fact, and it moves; the band depths are dated once, in `vertical.safe_zones`, and Facebook's in `vertical.facebook_safe_zones`.  
**Yields to:** AUTHORITY (in words): heritage-rules.md rule 5 - overlay text is added in the edit, never burned into the generated frame — Generation stage, not the edit.  
`vertical.text_safe_zone`

### Fill The Frame

**Also called:** tight vertical framing, no dead space  
**What it is:** The subject fills most of the vertical frame: medium close-up or closer, with wide shots used rarely and only for a reason. This is a sizing discipline for a small screen rather than a total fill, unlike the compositional Fill the Frame, which removes the surroundings entirely and makes texture the subject.  
**Effect on the audience:** It guarantees the detail reads on a small screen and prevents the empty feeling that sends the viewer back to scrolling.  
**Used for and where it works best:** The golden rule of vertical: a wide shot dies on a phone. Shoot closer than you think, and when you need a wide put strong movement or a large foreground element in it. On music videos with verse-tight/chorus-wide structure (`mv.verse_tight_chorus_wide`), open the chorus to a Medium-Wide (Cowboy/Knee) rather than an extreme wide to maintain mobile punch while providing musical expansion. On product, let the product fill half the frame height. In kids content comprehension outranks the tight frame: inside a teaching, counting or goodbye beat the clean wide wins (`kids.wide_clean_count`, `kids.wave_goodbye`), and in a dance or teaching shot the whole body stays in frame at child eye height (`kids.everyone_dances`), even though it is a wide. Tighten to the objects or to the bodies rather than to the room, but never past what the child has to see.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Vertical Micro-Drama, Commercial / TVC, Product Video / Motion Design, Kids Content, Music Video, Meme / Comedy Edit, Vlog / YouTube Long-Form | genres: all  
**Avoid when:** Framing too tight in an action beat destroys the geography and confuses the viewer. In a landscape it kills the sense of space you came for.  
**Prompt:** `tight vertical medium close-up, subject filling the frame top to bottom, minimal empty background`  
**Yields to:** `kids.wide_clean_count`, `kids.wave_goodbye`, `kids.everyone_dances` — in kids content, inside a teaching, counting, dance or goodbye beat, where the child must see the objects or the whole body.  
`vertical.fill_frame`

### First-Person POV

**Also called:** POV camera, head mounted view  
**What it is:** The camera stands in for the person's eyes: held at head height or mounted on the chest, with the hands entering frame from below.  
**Effect on the audience:** It puts the viewer inside the experience rather than watching it, and the effect is stronger in vertical because the frame shape echoes a standing person's field of view.  
**Used for and where it works best:** Best for cooking, craft, sport and walking through a historic site: keep the hands working in frame. In micro-drama use it for the moment of danger or discovery. Keep the horizon as level as you can so it does not induce nausea.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Vlog / YouTube Long-Form, Travel / Real Estate / Hospitality, Kids Content, Vertical Micro-Drama, Game Cinematic / Gameplay, Heritage / Historical Documentary | genres: Slice of Life, Adventure, Horror, Thriller / Suspense, Survival, Family, Action  
**Avoid when:** Violent shake in first person becomes nauseating fast on a phone. Do not use it in a beat where the viewer needs to see the character's face.  
**Prompt:** `first person point of view at head height, the subject's own hands entering the bottom of the vertical frame, level horizon`  
**Yields to:** `vertical.pov_caption` — Shake, or the beat needs a face.  
`vertical.pov_camera`

### Talk-To-Camera

**Also called:** direct address, piece to camera  
**What it is:** The presenter looks straight down the lens and addresses the viewer directly, with no interviewer and no second person in frame.  
**Effect on the audience:** It removes the distance and creates the feeling of a personal conversation, which is the strongest trust signal in short-form.  
**Used for and where it works best:** The basis of advice, explainers, opinion and creator-style advertising. Look at the lens itself, not at the screen, and frame to chest height. Pair it with jump cuts and captions to keep the pace up.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Vlog / YouTube Long-Form, Explainer, Commercial / TVC, News Report / Reportage, Corporate / Training, Video Essay | genres: Science / Educational, Slice of Life, Comedy, Family, Political Thriller, True Crime, Drama  
**Avoid when:** Looking at the screen instead of the lens makes the viewer feel you are watching something else. A long stretch of talk with no cuts or cutaways loses attention fast.  
**Prompt:** `presenter addressing the lens directly in a chest up vertical framing, soft key light, plain background`  
`vertical.talk_to_camera`

### Comment-Bait Freeze

**Also called:** comment prompt, question freeze  
**What it is:** The video freezes on a frame carrying a question, a choice or a deliberate mistake, and the viewer is pulled into typing an answer.  
**Effect on the audience:** A comment is the heaviest engagement signal on the platforms, and every reply pushes the video back into distribution.  
**Used for and where it works best:** Use a question that is easy to answer: which one is better, what is this place called, who recognises this piece. In heritage content ask about the symbol or the inscription. Reply to the first comments quickly so the thread takes off.  
**Best in:** formats: Vertical Short-Form (Reels, Shorts, TikTok), Video Essay, Kids Content, Explainer, Heritage / Historical Documentary, Meme / Comedy Edit, Vlog / YouTube Long-Form | genres: Heritage (civilisation-focused), Historical / Biopic, Science / Educational, Comedy, Slice of Life, Family, Mystery / Whodunit  
**Avoid when:** A deliberate mistake planted to farm comments costs you the specialist audience and damages your standing. Do not ask a personal or sensitive question just to collect replies.  
**Edit Overlay Rule:** this text is an editing and graphics overlay added in post-production. It is never burned into the generated image or video frame itself, which is what `heritage-rules.md` rule 5 forbids and what makes the text uncorrectable afterwards.  
**Prompt:** `[Edit Overlay Instruction]: freeze the last frame; the question and the arrow are added over it in the edit, centred in the safe band`  
**Yields to:** `vertical.end_cta` — No safe question to plant.  
`vertical.comment_bait`