Part VI — Shot Design, Blocking, and the Vertical Camera
Chapter 30. Vertical Composition and Phone Readability#
In this chapter
30.1 9:16 is not a widescreen frame with the sides removed#
Vertical framing emphasizes vertical hierarchy, faces, hands and props. Wide lateral blocking gets compressed, while platform buttons and subtitles occupy the edges. Composition has to start from the phone interface, not end with a crop.
Display the safe areas over a screenshot template for the target platform: the title or status bar at the top, the interaction buttons on the right, and the caption and subtitle band at the bottom. Test each platform and each distribution format separately.
30.2 Three safe areas#
The core narrative area holds faces, hands and proof objects. The subtitle area must not let text cover key action. The platform occlusion area must not carry important information. Safe areas are not a fixed percentage truth; measure them against the current platform UI and real devices.
Keep separate overlay templates for the ad version, the episode version and the UI-free master.
Figure 30-1 Four vertical composition tasks. A face close-up protects the eyes and headroom. A two-shot preserves the power relationship inside a narrow frame. An evidence shot keeps the fields that must be read outside the subtitle safe area. The failure case lets platform buttons, the comment layer and the top UI obscure the character simultaneously.
A safe area is not a permanently universal transparent frame. Platform, account type, ad buttons and subtitle styling all change the usable area, so templates carry versions. Composition approval uses real platform screenshots as an overlay rather than judging "subject centered" on a clean canvas.
30.3 Framing the face#
Close-ups need room in the direction of the gaze, without the top of the head being randomly cut or the chin sitting on the subtitles. When a character looks toward frame right, leave appropriate looking room; compressing it deliberately can express pressure, and it should be a choice.
Two characters in frame can use foreground/background layering, upper/lower offset, or one near and one far — rather than being squeezed into two side-by-side portraits. Height and power must still agree with the space.
30.4 Two-person vertical layouts#
Common solutions: one shoulder in the foreground with the other person clear; vertical offset; one standing and one seated; a door frame or table edge dividing them; a mirror or screen establishing the relationship. Symmetrical side-by-side suits public confrontation and invites face bleed and flat space.
Keep motion small in vertical two-shots, and cut to singles for important lines. Do not have both people striding laterally out of frame together.
30.5 Hands and proof objects#
A document held low in frame is easily covered by subtitles and platform captions. Introduce important props in the core area first, then cut to a separate insert. When face and document must both be sharp, test depth of field and size in practice.
The audience only needs the title, a name, a date, an amount or a result. Do not push a full page of small type into a vertical frame.
30.6 Minimum legibility testing#
Test at real phone size and normal viewing distance, not on an enlarged desktop preview. Show it once to a cold viewer and ask what they read. If they had to pause and zoom, the shot failed.
Testing should also simulate compression, lower brightness, screen glare and muted playback. Hold time and existing context determine legibility together.
30.7 Negotiating between subtitles and picture#
Subtitle position is not permanently fixed. When a key prop occupies the lower frame, subtitles can move up briefly. When the face sits high, avoid covering the mouth and chin. The subtitle system records avoidance zones, adjusted automatically or by the editor.
Dialogue writing should control the sense units per line rather than relying on shrinking the type. Accent colors for emphasis should agree with the proof object's accent color, avoiding multicolored lines.
30.8 Vertical depth#
Vertical framing can build power through foreground, mid-ground and background: the document in front, Lin Yun in the middle, Lin Wei behind; or security's shoulder in front creating obstruction. Depth of field must not fully defocus critical background evidence.
Doors, columns, screens and windows can frame the vertical composition, and the decoration must not occupy the core readable area.
30.9 Establishing shots in vertical#
You do not need an extreme wide. Shoot down the length of the table from the door side so three principals and the screen are visible at once, or establish in two shots: door to table, then screen to head seat.
A vertical wide still needs enough hold time for the audience to orient. Complex space flashing past only produces noise.
30.10 Platform cropping and multiple versions#
A 9:16 master may be cropped by a platform preview to other ratios. Test the center crop for the first frame and the ad thumbnail. Where 1:1 or 16:9 is required, do not simply scale and crop — recompose from high-resolution assets.
Keep key graphics as re-layoutable layers. Leave a little reframing room when shooting or generating character shots, without letting the subject become too small as a result.
30.11 Dynamic composition#
Camera push-ins, characters approaching and props entering all change the relationship to subtitles and safe areas. Check the whole duration, not only the first frame. When a document enters from the bottom, subtitles can be delayed or displaced.
Video models can let characters drift out of the core area over time; the last frame must be re-checked for head, eyes and prop position.
30.12 Composition templates and exceptions#
Templates increase stability: single close-up, over-shoulder, two-person confrontation, document insert, phone close-up, entry through a door, crowd reaction. A template stores safe areas and typical positions without locking every shot.
Climaxes may break the template with intent. Stable templates are what give an unusual composition its force.
30.13 SOP for phone composition#
First, load the target platform UI template. Second, mark the core, subtitle and occlusion areas. Third, arrange composition around characters, eyelines and props. Fourth, choose a vertical hierarchy for two-shots. Fifth, insert key text separately. Sixth, test with subtitles overlaid. Seventh, watch at normal speed on a real phone. Eighth, test compression and center cropping. Ninth, check first, middle and last frames of moving shots.
30.14 Fault tree#
Symptom: legible on a computer, illegible on a phone. No real-size testing. Enlarge the prop fields, extend the hold, or move to an insert.
Symptom: subtitles always cover hands and documents. Composition and subtitles were designed separately. Establish avoidance zones and dynamic positions.
Symptom: two-shots look like ID photographs. Only side-by-side was used. Use depth, vertical offset, seated versus standing, and over-shoulder layering.
Symptom: the wide flashes past and nobody knows who is where. The establishing shot is too complex or too short. Split it into two simple geography shots.
Symptom: platform buttons cover a key reaction. The UI template was never loaded. Move faces and proof into the core area.
30.15 Checklist, exercises and deliverables#
Check that it is natively 9:16; that platform UI is overlaid; that faces and eyelines have room; that two-shots have hierarchy; that props avoid the subtitle band; that key fields can be read in one pass; that subjects stay inside the safe area through motion; and that other ratios are recomposed rather than cropped.
Exercise one: draw the three safe areas over a platform screenshot. Exercise two: design four vertical layouts for a two-person confrontation. Exercise three: test minimum legible solutions for the authorization's title, name and amount. Exercise four: re-storyboard a widescreen scene rather than cropping it.
Deliverables for this chapter: platform safe-area templates, the vertical composition kit, the two-person layout library, the proof readability test, the subtitle avoidance map, and the multi-ratio recomposition plan.
30.16 Responsive framing rather than a single crop#
One master may output 9:16, 4:5, 1:1 and 16:9. Responsive framing first defines the semantic objects that cannot be lost, the objects that can be rearranged, and the background that can be cropped. The lead's face, the evidence document and the opponent's reaction cannot be lost. Bare wall and decorative lighting can be cropped. Subtitles and CTAs can be repositioned. Recalculate the layout for each output ratio rather than mechanically trimming from the center.
responsive_frame:
semantic_objects:
CH_LINYUN_FACE: {priority: 1, crop: forbidden}
PROP_AUTHORIZATION: {priority: 1, min_screen_height: 18_percent}
CH_LINWEI_REACTION: {priority: 2, crop: partial_allowed}
background_window: {priority: 4, crop: allowed}
layouts:
9x16: stacked_face_and_document
4x5: diagonal_two_subject
16x9: relational_two_shot
When a composition cannot retain its core objects at a target ratio, return to shot design and generate a separate version. Outpainting the sides solves the canvas; it does not automatically solve character relationships and eyelines.
30.17 Dynamic safe areas and attention collisions#
Platform UI, safe areas and subtitles are not a static masking problem. Character movement, phone popups, comment layers and animated subtitles collide with proof objects over time. Composition previsualization should place every overlay on the timeline and check who owns attention in each frame.
When Lin Yun raises the authorization, subtitles can move up briefly or end early. If a subtitle still covers the signature, the audience is reading two sets of text at once. Resolve in order: change subtitle timing, change subtitle position, restage the shot — and only then consider shrinking the evidence. Never reduce core proof to illegible decoration to make room for UI.
Dynamic safe areas record the platform, its version and the date, because platform interfaces change. Verify before release with an actual private upload rather than trusting a template from six months ago.
30.18 Device acceptance for vertical composition#
Review must cover at least a small screen at low brightness, normal handheld distance, one-handed viewing, and a real page carrying platform UI. Legible at 200 percent in design software does not mean a viewer sees it at a glance. Accept evidence shots on whether a first-time viewer can read the key fact within the allotted time — not on pixel dimensions.
For faces, check that eyes, mouth and identity fall in the visual center. For two-shots, check that distance and power read instantly. For wides, check that vertical layering still explains the location. For action, check that hands and props are not cut off by a body or by subtitles. Keep passing and failing examples for each shot type, forming the vertical composition kit.
Finally, run a thumbnail test. Acquisition entrances, episode lists and paused frames may show only a small image, where key silhouettes and color masses must still be identifiable. The episode's composition need not look like a poster throughout, and the first frame and the cut-point frame deserve extra verification.
A note on sources#
Vertical microdrama reads around faces and proof objects. Genuine mobile composition brings platform UI, subtitles, compression and viewing distance into the design together — rather than merely setting a 9:16 parameter.