Each face is a separate canvas
A wall is one rectangle. A cube is four, and the control system does not join them into a single surface by itself. Before any creative work begins, the pixel dimensions of one face have to be written down: divide the edge length by the pixel pitch and that number is both the width and the height.
A 960 mm edge at 2.5 mm pitch gives 384 by 384 per face. That is unusual content geometry. It is neither landscape nor portrait in the conventional sense, so any asset produced for a normal screen needs reframing.
The practical consequence is that cube content is produced to the face size. Reframing a 16:9 master into a square face discards either the sides or the top and bottom, and the discarded part is usually where the logo was.
Choose how the four faces are fed
Three arrangements cover most cubes. Independent canvases give each face its own input, which suits wayfinding, menus or four languages. A single wide canvas sends one signal split across all four faces, which suits continuous motion. A mirrored canvas feeds the same content to every face, which is the cheapest to produce and reads as a repeating pattern.
The single wide canvas needs the most planning. Four faces of 384 pixels laid side by side make a 1,536 by 384 strip, and the content has to be drawn in that aspect with the corners sitting where the folding happens.
The arrangement is a control-system setting, not a content decision made late. It determines how many outputs the processor needs and whether one face can be updated without re-issuing the rest.
The corner is where continuity breaks
Two faces meeting at a vertical edge do not share pixels. The image stops at the last column on one face and starts again at the first column on the next. Anything crossing that line, a moving object, a band of colour, a line of text, is cut there.
Three approaches work. Draw content that reads per face and let the corner be a deliberate break. Draw content that treats the corner as a fold and shifts the image by the width of the corner profile. Or keep motion away from the corners so each face carries its own composition.
What does not work is assuming the corner behaves like a seam between two cabinets on a flat wall. On a wall both sides sit in the same plane and the eye joins them. On a cube they are at right angles, and the eye reads two surfaces.
Motion behaves differently in the round
On a wall, motion travels in one direction across a surface everyone sees from roughly the same angle. On a cube, a viewer standing at a corner sees two faces at once, and motion crossing between them appears to change direction at the edge.
Slower motion hides this. Fast horizontal panning draws attention to it. Where the creative includes fast movement, test it on the actual cube geometry before sign-off, because the effect is hard to judge on a flat preview monitor.
Match brightness across faces, then calibrate
Faces sit in different light, so mapping the content is only half the job. The four faces also have to look like one object.
Set the level for the face with the most ambient light and calibrate the others to it. Uniformity across faces matters as much as uniformity within a face, since a viewer at a corner compares two of them side by side with nothing in between.
What to hand to the content team
Four items: the pixel dimensions of one face, the feeding arrangement, the frame rate the control system will run, and the position of the corners inside the canvas.
Add safe zones at the vertical edges so text and logos never land on a fold, and agree the loop length and file container before production starts. These are ordinary pre-production questions. They cost nothing to settle early and a full re-render to settle late.
Square face geometry
Reframing a 16:9 master into a 384 by 384 face discards either the sides or the top and bottom. Cube content is produced to the face size, which is why it cannot be lifted from wall artwork.
The single wide canvas
Four faces of 384 pixels laid side by side make a 1,536 by 384 strip. The content has to be drawn in that aspect with the corners sitting at the fold, and tested for fast motion before sign-off.
Steps for a project: mapping content on an led cube
- Write the face pixel size down first A 960 mm edge at 2.5 mm pitch gives 384 by 384 pixels per face, which is neither landscape nor portrait. Any asset produced for a normal rectangle will not map cleanly, so the face size drives everything else.
- Choose the feeding arrangement second Independent canvases, one wide canvas split across faces, or a mirrored feed are a control-system setting, not a late content decision. The choice sets how many outputs the processor needs.
- Design for the corner third Two faces meeting at a vertical edge do not share pixels; the image stops at the last column on one face and starts at the first on the next. Add safe zones at the edges so text and logos never land on a fold.
Clearing up mapping content on an led cube
“The corner behaves like a seam between two cabinets on a flat wall.” — On a wall both sides sit in the same plane and the eye joins them. On a cube they are at right angles, so the eye does not join them, and motion crossing the fold needs testing on the actual geometry.
A note for the supplier: mapping content on an led cube
Write the face size down first, choose the feeding arrangement second, and design for the corner third. A cube is four canvases at right angles to each other, and content that ignores the corners reads as four screens stacked together instead of one object.