Four menu boards can play one takeover video two ways: run all four from one player through a video wall processor, or give each screen its own player with a sync rule that re-anchors every loop. Left free-running, cheap player clocks drift 100 ms apart in about 17 minutes, which is when the mismatch becomes visible to a viewer.
Four menu boards above the counter spend the day on their own playlists, then a promotion needs all four screens to behave like one video. Two architectures do that job. One player feeds a video wall processor that crops the frame into quadrants, so every screen shares a single clock and cannot drift. Or every screen carries its own player, and a sync rule starts the four quadrant files together and re-anchors them each time the video loops. The deciding number is how long four clocks stay close on their own: at worst, two players started together pass 100 ms of offset in about 17 minutes, and 100 ms is where one video starts reading as two. This page works the arithmetic and ends with the decision for each layout.
Two architectures for one video on four screens
One player, one clock. The player outputs a single large frame and a video wall processor splits it: the processor takes the input, crops one quadrant per output and scales each crop to its screen. One device generates the picture, so the screens cannot disagree. The everyday menus live on that same big canvas: the layout tool draws four zones on one 3840 x 2160 slide for a 2x2 wall, the processor shows each screen its zone, and the switch from menus to promo is just the next slide. Drive an LED video wall from an Android signage player walks through the single-player route, and video wall sync architecture: how players stay in step explains what software sync is racing against.
One player per screen. Each screen gets a player that plays its own 1920 x 1080 quadrant file, so every panel keeps full pixels, which is what a 1x4 row over a counter needs. Synchronization turns into a software job: one player acts as host, triggers the start, the other three follow, and the group re-anchors at every repeat. Why your sync group screens play video out of sync covers the drift that appears when the re-anchor is missing. The same players then serve their solo playlists for the rest of the day and for the other screens around the restaurant.
The shortcut worth knowing. If the takeover can be the same video on all four screens, there is nothing to synchronize: one player into an HDMI splitter feeds all four the same full frame. That is a different look, four copies of one video instead of one picture spread across four panels. HDMI splitter or video wall processor for two screens sorts out which box does which job.
The number that decides it
Count in frames first. At 30 frames per second, one frame lasts 1 s / 30 = 33.3 ms; at 60 frames per second, 16.7 ms. Motion crossing a seam offset by more than one frame reads as two clips, so set the planning figure at about 100 ms of offset between adjacent screens, three frames at 30 fps. Past that, viewers see two videos where one was intended. Design the takeover to stay under 100 ms.
Next, the drift rate. The quartz oscillator in an inexpensive media player is typically specified to between ±20 and ±50 ppm. The worst pair on a wall is one player 50 ppm fast and its neighbor 50 ppm slow: a relative rate of 100 ppm, which is 100 μs (microseconds) per second of playback, or 6 ms per minute.
Sound tightens the budget. The film guideline for lip sync is 22 ms in either direction, so a talking promo with room audio has a 22 ms budget, and 100 μs per second spends it in 220 s. Either re-anchor a talking promo every 220 s or keep the voice off it.
One relief: screens that run separate playlists most of the day only need to agree during the takeover, so the drift budget applies to the takeover window, not to the whole day.
Example: four menu players started together at 9:00
Inputs. Four identical players, one per screen in the row above the counter, all triggered by one scheduled start at 9:00 and left to free-run until the next scheduled trigger at close. Each crystal is specified to ±50 ppm, the standard grade for an inexpensive oscillator. The promo runs at 30 fps with no speech, so the limit is the 100 ms picture figure.
Rule. Relative drift between two players is the sum of their tolerances when their crystals drift in opposite directions, and offset equals drift rate multiplied by elapsed time.
Arithmetic. Worst-case relative rate: 50 ppm + 50 ppm = 100 ppm = 100 μs per second. After 1 minute: 100 μs/s × 60 s = 6 ms. Crossing point: 100 ms ÷ 100 μs per second = 1000 s, about 17 minutes. After 1 hour: 100 μs/s × 3600 s = 360 ms, a third of a second.
Result. The wall reads as one video for roughly the first quarter hour and visibly falls apart over the following hour.
Decision. A start time alone is not a sync system; something must pull the players back together before the 1000 s crossing point. Re-anchor at every loop and the error resets each pass: a 45-second promo re-triggered every repeat accumulates at most 100 μs/s × 45 s = 4.5 ms per pass, invisible. If the players only take a timed start, keep the takeover under 15 minutes: 900 s × 100 μs per second = 90 ms, still inside the figure.
The options compared
| Approach | How the screens stay together | Hardware per wall | Best for | Watch out |
|---|---|---|---|---|
| One player + video wall processor | One clock; the processor crops and scales each quadrant | 1 player, 1 processor | 2x2 walls; menus and promo living on one canvas | A 1x4 row divides a 3840 x 2160 input into 960 x 1080 per screen, so pixels soften |
| Per-screen players with a sync group | Host triggers the loop; the group re-anchors every repeat | 4 players | 1x4 rows that need full 1920 x 1080 on every panel | Drop the re-anchor and drift crosses 100 ms in about 17 minutes |
| Per-screen players with timed starts | Clocks set from one source; all four start on the same minute | 4 players | Screens that mostly run separate playlists; takeovers under 15 minutes | Offset grows 6 ms per minute after the start |
| Cloud CMS with sync playlists | The service coordinates the four start times | 4 players | Fleets already committed to a CMS across 20+ locations | Per-screen subscriptions multiply: 4 screens at 20 locations is 80 renewals |
| HDMI splitter | Same full frame on all four screens; nothing to drift | 1 player, 1 splitter | Takeovers where every screen should show the whole video | Four copies of one video, so no single-picture effect |
When to pick something else
Geometry rules out the processor route more often than anything else. A 3840 x 2160 input over a 2x2 block hands each screen 1920 x 1080, every pixel used. The same input over a 1x4 row hands each screen 3840 ÷ 4 = 960 x 1080, stretched 2x across a 1920-wide panel, and text near the seams goes soft. So a 1x4 menu row points to per-screen players; a 2x2 block points to the processor. Feed the processor a smaller master and each screen gets a fraction of its native pixels: a 1080p master on the same 2x2 block leaves every panel at 960 x 540, a quarter of its pixel count.
Pick the splitter when the same promo on every screen is acceptable and you want zero sync engineering.
Stay with per-screen players when the room has extra screens that need their own playlists anyway: the players that form the wall during the promo serve those screens the rest of the day.
And whatever file route you cut, test the encode on the player before it ships to every location. Video bitrate for signage player: test before you deploy gives the test method.
What to ask before you buy
- Wall geometry first: 1x4 row or 2x2 block, screen size and panel resolution. Every other choice follows from this.
- Processor limits: its maximum input resolution and output count. Full-HD quadrants on a 2x2 wall need an input of 3840 x 2160.
- Sync behavior: does the group re-anchor at every loop, and can you set the trigger? A feature that only fires once a day is the example above with a worse ending.
- Content: who cuts the four 1920 x 1080 quadrant files, or who builds the 3840 x 2160 canvas.
- Cabling: per-screen players need the store LAN to reach each screen; the processor route needs HDMI runs from one player only.
- Fleet math: 4 screens at 20+ locations is 80 players to manage. A per-screen subscription turns into 80 renewals; offline card-based players carry none. Our card-based menu boards include the software with no subscription, listed in the menu boards with no monthly fee category.
- Screens themselves: size them from the displays category, and if the player side is new, start with the digital signage player guide.
The decision
Recommendation: For four screens in a row above the counter, put one player on each screen, cut the promo into four 1920 x 1080 quadrant files, and run a sync group that re-anchors at every loop; a 45-second promo then holds its worst-case seam error to 4.5 ms, and the everyday menus stay on each player's own playlist. For a 2x2 block, run one player into a video wall processor fed 3840 x 2160 so every panel gets a full-HD quarter, and draw the menus as four zones on one canvas. Whichever route you take, keep any free-running stretch under 15 minutes, and under 220 s when the promo carries speech.
Questions
Can I mix players from different makers in one sync group?
Only if the software states it supports mixed hardware. Different players take different times to decode and present the first frame, so a mixed wall starts with an offset the sync rule cannot trim, and the seam error is there from the first frame. Keep one player model per wall wherever motion crosses the seam. Mixed models are harmless on screens that run their own playlists most of the day and never join the takeover.
Do I need one 4K video file for a four-screen wall?
Only on the processor route. Four per-screen players each play their own 1920 x 1080 quadrant file, so no 4K file exists anywhere in the system. One player plus a processor needs the master encoded at the full canvas, 3840 x 2160 for a 2x2 wall, so each panel receives full-HD pixels; a 1080p master leaves each screen at 960 x 540. Test the encode on the actual player before copying it to every location.
Can menu boards run offline with no monthly fee?
Yes on our panel-mount range: the player software is included with no subscription, and playlists load from a microSD card or USB. The larger wall models in the current catalog are quoted per project, so confirm the software model, ports and warranty on the quote before you standardize across all your locations.
Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.