How PixelBid calculates

Cabinet layout

Columns = target width ÷ cabinet width; rows = target height ÷ cabinet height, each rounded to whole cabinets. Resolution, physical size and face area follow from the whole-cabinet grid, never from the target.

A whole-cabinet size within 12.7 mm (half an inch) of the target counts as exact. Otherwise the grid rounds up to the next whole cabinet and shows the size you actually get, so you can take a cabinet off if the opening cannot take it. The aspect ratio is reported from the real grid, and flagged when rounding moves it noticeably.

Power and circuits

Circuits are sized from each cabinet's maximum draw, so the wall cannot trip at full white. Each circuit is loaded to at most 80% of its breaker — the continuous-load rule of the US National Electrical Code (NEC 210.20(A)). Circuit capacity = volts × amps × 0.8 on single phase, and √3 × volts × amps × 0.8 on three phase.

Cabinets per circuit = circuit capacity ÷ cabinet maximum watts, rounded down, because a cabinet draws from one circuit only. The circuit count is the larger of the arithmetic minimum and the number of runs actually drawn: runs are laid out so that sibling runs read alike on site, which can take one more circuit than the division, and the electrician works from the drawing.

Heat is sized from typical draw, not maximum: BTU/h = typical watts × 3.412. Typical draw is the vendor's published average. Some vendors publish it as a fixed fraction of the maximum rather than a measurement, so treat heat as indicative.

Real draw depends on brightness and content. You can enter the brightness limit set in the processor: maximum draw at the limit = maximum × (10% + 90% × limit). The 10% is an estimate of what does not dim — receiving cards, driver ICs, supply losses, fans — since no vendor in the library publishes black-screen power. Circuits stay sized at full brightness unless you choose to size them at the limit; then the results, the report, the power drawing and the quote all say so, because at full brightness those circuits trip and the limit has to be locked in the processor.

Outside the US: the same 80% rule is applied to your country's voltages and breaker sizes, but local practice differs — RCDs and derating rules in Europe, for example. Check circuits against your local code before you build.

Data and processors

Pixels per port = port speed × 0.936 ÷ (bits per pixel × frame rate), rounded down to the 10,000 that capacities are published in. Bits per pixel are those NovaStar specify: 24 at 8-bit, 32 at 10-bit and 48 at 12-bit — not three times the bit depth. The 0.936 efficiency is set so that a 1G port at 8-bit and 60 Hz carries 650,000 pixels, NovaStar's published figure; the same port carries 480,000 at 10-bit and 320,000 at 12-bit.

The processor is chosen from each NovaStar model's own datasheet: its port count, its published pixels per port and its total load at each bit depth it states. Where a sheet does not state a bit depth, its 8-bit figure is rescaled with the bits-per-pixel above. The MX series is rated as a whole: a port carries the unit's total load divided by its ports — an MX40 Pro, 9,000,000 pixels over 20 ports, is 450,000 a port. 23 NovaStar models are in the library.

Cabinets per data run = pixels per port ÷ pixels per cabinet, rounded down, where pixels per port is the smaller of the port's own figure above and what one port of the quoted processor carries. Runs are drawn port by port, with the same sibling rule as power.

Viewing distance

Three distances, each a multiple of the pixel pitch in millimetres, given in metres: 1× is the closest anyone should sit before the pixel grid is obvious; 2× is where the trade puts the best seat; 3.44× is where one pixel subtends one arcminute and the structure disappears. All three are floors: nothing is too far away.

Distance is measured from the eye to the centre of the image, so the nearest seat in any audience is the centre of the front row — the only seat that can be too close. PixelBid checks that seat and says by how much it falls short, and maps every part of the seating block by the tier it reaches.

Weight and structure

Weight = cabinets × the cabinet's published weight. Face load = weight ÷ face area, and a wall above 50 kg/m² is flagged for structural review.

These are cabinet weights only. Frames, rigging, cables and processors are not included, and PixelBid does not do rigging or structural engineering. A flag is a prompt to get that done, not a substitute for it.

Where the data comes from

The library holds 59 cabinets from Absen, Gloshine, INFiLED, Leyard and Unilumin. Each row is transcribed from the vendor's own datasheet and names that document; the calculator shows it beside the cabinet you pick. Vendors print power and weight in different ways — per panel, per square metre, maximum and average or ceilings — and each is converted to per-cabinet figures before it is stored.

A row that has not been checked against its datasheet is marked as such in the cabinet picker and on the report. A custom cabinet uses your own figures; where you leave power or weight blank, they are estimated from typical values for the pitch, and the calculator says so.

Receiving cards are not counted. No vendor in the library publishes how many cards a cabinet carries, and a number nobody can source does not belong in a bill of materials.

Who stands behind it

PixelBid is built and maintained by AI Home, Inc., a California-licensed contractor (CSLB #1063528) that designs and installs LED walls. The rules on this page are the ones we quote and build our own walls by. Questions, corrections and disagreements go to support@pixelbid.ai, and we reply within one business day (US Pacific time).