DOOH specs: sizes, formats and file limits

DOOH has 50+ screen types, so one creative gets transcoded, resized and letterboxed across them — but you still build to spec, and the specs differ by supply platform. This reference gathers the verified build requirements for the major DOOH platforms, the sizes that actually carry spend, and the two universal rules that catch most newcomers: build silent, and build vertical.

The short answer

Build portrait 1080×1920 as the master, and supply a landscape 1920×1080 variant. Those two sizes together took 77% of programmatic DOOH video spend in the most recent transaction data (Broadsign / Place Exchange 2026 trends — primary, two-SSP).
Video: MP4, H.264, no audio track. Ströer’s public spec is 5–30 seconds, ≤12 Mbps, no audio (Ströer — primary). Ten seconds is the common build.
Static: JPG or PNG. Vistar caps stills at ≤10 MB and video at ≤50 MB; keep text at ≥15 px (Vistar — primary).
HTML5 acceptance varies, and dynamic creative often renders to a static image at serve time — design DCO variants as static layouts (Vistar — primary).
One master is transcoded across many screen types — keep critical content inside a safe area so it isn’t cropped (Vistar — primary).
File-size caps are mostly per-network, not platform-wide — confirm them with your supply path before you build.

The per-platform table

Build to the supply platform you’re transacting through. Verified specs (platform docs — primary/directional):

Platform	Static	Video	HTML5	Notes
The Trade Desk	D6 vertical 1080×1920, JPG/PNG	MP4	—	D6 is the cross-SSP reach format
Ströer	—	MP4 / H.264, no audio, 5–30 s, ≤12 Mbps, 16:9	—	The canonical no-audio spec
Vistar	JPG, ≤10 MB, text ≥15 px	MOV/MP4, ≤50 MB	Not accepted	Transcodes/letterboxes; DCO → static image
Hivestack	JPG/PNG	MP4/MOV, under 10 MB	—	Common dims 1920×1080, 1080×1920, 1280×720
Broadsign	PNG/JPEG (no animated GIF)	MP4/MOV/WebM, H.264 ~8 Mbps	Supported (Chromium v87+)	No published platform-wide file-size cap

A figure to ignore: a “Broadsign 750 KB max” limit circulates but could not be verified in Broadsign’s own docs — caps in DOOH are usually per-network, so confirm rather than design to an unverified number.

The sizes that actually carry spend

The long tail of DOOH resolutions is real, but the money is concentrated in a handful of canvases. In the 2026 transaction data across 1.7 million programmatically-enabled screens (Broadsign / Place Exchange — primary, two-SSP):

1080×1920 (9:16 portrait) and 1920×1080 (16:9 landscape) together took 77% of programmatic video spend.
The top three display sizes — 1080×1920, 1920×1080 and 1400×400 — took 79% of display spend.

So a two-file delivery (one portrait, one landscape) addresses the overwhelming majority of buyable inventory, and a third asset at 1400×400 covers the wide banner shapes used in transit and retail. That finding, and what it means for selling salon inventory, is covered in programmatic DOOH runs on PMPs and portrait.

For a beauty network, the practical consequence is that the mirror and styling-station formats sit on the dominant canvas, not an exotic one — a buyer with an existing portrait asset can run on them with no new production. The physical panel requirements behind those canvases — brightness, duty cycle, panel type — are a separate question, covered in the screen hardware spec benchmark.

Spot length: build ten seconds unless told otherwise

Ströer’s published range is 5–30 seconds, and most DOOH networks sell in fixed slots rather than variable durations. The reason is arithmetic, not preference: a screen’s capacity is loop length ÷ spot length, so spot length sets how many advertisers a screen can carry and what each share of voice is worth — the model worked through in loop length and ad slots.

A 10-second master is the safe default: it divides cleanly into the 60- and 120-second loops most venue networks run, and it can be cut down rather than padded out. Deliver longer only when the operator has confirmed the slot length, because a 30-second file in a 10-second slot is not a creative decision — it simply won’t schedule.

The two universal rules

Whatever the platform, two rules hold across DOOH and trip up most first builds:

Silent by default. DOOH is experienced without sound — platform specs are explicit (e.g. Ströer’s public-video spec is no audio track). Carry the message in type and captions, never voice-over. In a salon this is doubly true: the venue owns its own ambience.
Vertical first. Portrait 1080×1920 (9:16) is the dominant panel format and the natural build for a mirror or styling-station screen. Make it the master; supply a landscape 1920×1080 variant for lobby and standalone screens.

One master, many screens

You don’t build a file per screen — platforms run a transcoding service that resizes and letterboxes one master across similar aspect ratios (Vistar — primary). The practical consequences:

Keep critical content in a safe area. A logo or QR jammed to the edge can be cropped when the asset is re-fit to a different screen shape.
Design for legibility at distance. Keep text large — Vistar specifies a ≥15 px minimum height — and high-contrast, because the same asset may render on a small mirror screen or a large lobby display.

Dynamic creative: design the static frame

Contextual and dynamic creative (daypart, weather, nearest-store, promos) lifts outcomes, but the production reality constrains the build: many platforms render dynamic creative to a static image at serve time, animated dynamic creative is region-limited, and raw HTML5 is often not accepted (Vistar — primary). So design each DCO variant as a static layout that stands on its own — don’t assume live animation will play on the panel. The creative craft around these specs is in creative for salon and mirror screens, and the scheduling side in dynamic creative and dayparting for salon screens.

Before you deliver: six things to confirm

Specs published on a platform’s website are the floor; the network you are actually running on sets the ceiling. Confirm these in writing before production, not after:

Exact pixel dimensions and orientation per screen group — not just “portrait.”
Maximum file size and bitrate, which are usually per-network rather than platform-wide.
Spot length and tolerance — whether a 10.2-second file is rejected or trimmed.
Audio — assume none, and confirm the panel has no speakers before designing around sound.
HTML5, DCO and animation support, including whether dynamic variants are flattened to static at serve time.
Delivery deadline and route — how far ahead of the flight the file is due, and whether it goes to the DSP, the SSP or the operator directly. Lead times are set by the operator, not by a universal industry rule, so ask rather than assume.

Then ask the reporting question in the same conversation: what proof of delivery comes back, and whether it evidences display or only scheduling — the distinction that decides what an invoice actually substantiates, laid out in proof of play: scheduling vs display.

Common questions

What creative specs are required for digital OOH screens?

A portrait 1080×1920 master and a landscape 1920×1080 variant, delivered as MP4/H.264 video with no audio or JPG/PNG stills, built to around 10 seconds, with critical content inside a safe area. Everything else — file-size caps, bitrate, HTML5 support — varies by supply platform and by network.

What file formats does DOOH accept?

MP4 (H.264) is universal for video, with MOV and WebM accepted on some platforms. JPG and PNG are the standard stills; animated GIF is commonly rejected. HTML5 is accepted by some platforms and not others.

Do DOOH ads have sound?

No — assume silent. Most DOOH panels have no speakers, and platform specs frequently require no audio track at all. Carry the message in typography and captions.

How long should a DOOH ad be?

Ten seconds unless the operator specifies otherwise. Published ranges run 5 to 30 seconds, but slot length is fixed per network and derives from the loop, so a file that doesn’t match the slot cannot be scheduled.

What file size limits apply?

Where published, they are platform-specific: Vistar caps stills at ≤10 MB and video at ≤50 MB; Hivestack restates an under-10 MB video cap. Broadsign publishes no platform-wide cap. Treat caps as a per-network question.

Does one file work across every screen?

Effectively yes, with care. Platforms transcode and letterbox a single master across similar aspect ratios, which is why the safe area and large type matter more than producing a file per panel.

Methodology & caveats

Verified (primary): portrait 1080×1920 as the dominant build and the transcoding/letterboxing reality, the ≥15 px text minimum, JPG ≤10 MB / video ≤50 MB, HTML5-not-accepted and DCO-renders-to-static facts from Vistar; the D6 1080×1920 format from The Trade Desk; the MP4 / H.264 / no-audio / 5–30 s / ≤12 Mbps spec from Ströer; Broadsign’s supported formats (PNG/JPEG, MP4/MOV/WebM, HTML5 via Chromium 87+) from Broadsign docs; the 77% of video spend / 79% of display spend format concentration and the 1.7 million-screen dataset from the Broadsign / Place Exchange 2026 Programmatic DOOH Trends Report (two-SSP transaction data, full-year 2025 and Q1 2026). Directional: Hivestack’s under-10 MB cap and common dimensions (via reseller restatement). Not asserted: a “Broadsign 750 KB” file cap (not in Broadsign’s own docs — per-network caps apply); any universal delivery lead time (operator-set, no industry standard published); any beauty-specific creative benchmark. The 10-second default is our reading of the loop arithmetic, not a published standard. Specs change; confirm against your supply platform’s current creative guidelines before delivery. Updated 22 Aug 2026 with the 2026 format-concentration data and the delivery checklist.