Walk any show floor and every fine-pitch LED wall carries the same three letters: COB. Chip-on-board. It reads like a specification. It is not — it is a packaging method, and packaging is a commodity. Every factory sells it.
What separates a reference display from a commodity video wall happens above the emitters, in a layer most spec sheets never mention.
Why so much COB is hazed and blackened
Look at a bare COB module in daylight and the problem is obvious. The PCB solder mask shifts colour from batch to batch. Solder pads, copper traces and bond wires glint back through the translucent encapsulant. Switch the wall off and instead of one black slab you get a faint patchwork of grey, green and metallic pinpoints.
The industry's fix is simple: load the encapsulant with haze and black pigment. It works. Light scatters, the metal disappears, and the panel reads as uniform black when it is off. But look closely at what that solution actually is. It is not an optical design. It is a cosmetic process — one that solves an appearance problem at the factory and hands the bill to the image.
Haze is volume scattering. Black is volume absorption. Neither knows which light is which.
This is the crux. A scattering particle does not check whether a photon arrived from the room or left the LED. It scatters both. Carbon black does not care whether it is absorbing ambient light or your own emitted light. It absorbs both. So the industry trades black for white, every single time:
- Haze softens the white field. Scattered emitted light reads as a hazy, slightly dirty white. Edges lose their edge. Neighbouring pixels bleed into one another. The dirty screen effect.
- Black eats luminance. To reach the brightness number on the spec sheet you raise drive current, and pay for it in power, in heat, and in low-grey uniformity.
- Neither reduces reflection. Masking hides what is underneath; it does not cancel Fresnel reflection at the surface. The moment a lamp or a window enters the scene, black lifts back to grey.
- Neither touches pixel structure. Discrete emitters plus a black matrix still read as grain up close, and still strike moiré the instant a camera appears.
There is a quieter problem too. Black-loaded encapsulant varies from batch to batch. Using a non-uniform material to hide a non-uniform substrate adds a fresh source of the very colour variation you were trying to remove — and then you compensate for it module by module. Binning is where that starts.
We cancel the reflection instead of absorbing it
At NovaVista we do not mask the surface. We engineer it. Two layers sit above the emitters:
- One. An interference-based AR optical surface. External light is arranged to cancel at the interface rather than be soaked up: 2.0% reflectance (SCI) or lower, at a calibrated 40 GU gloss. Emitted light is neither scattered nor absorbed, so clarity and efficiency stay intact — that is exactly where the matte anti-glare approach and ours part company.
- Two. An orthogonal grating (Grating Series) that reconstructs discrete point emitters into a square, continuous emissive plane at 85% fill or better. The grid is not hidden. It is gone. Without giving up grey scale.
Same emitters. Two surfaces. Measured.1
We ran the A/B on an identical P1.5625 panel — bare versus treated.
- Black-state luminance: 0.3–0.5 cd/m² → 0.07 cd/m² (4–7× darker)
- Ambient contrast: roughly 800:1 → 4,088:1 (about 5×)
- Colour shift from the optics: none measurable, Δu'v' ≤ 0.002
That last line matters most. A black-loaded encapsulant always moves your colour point, and you correct for it afterwards. Ours does not have to.
And because we are not buying black with drive current, the rest of the picture holds: 98% brightness uniformity or better, Δx/Δy ≤ ±0.003 chromaticity uniformity, ΔE ≤ 3.0, 10-bit 1,024-level greyscale, and 3,840 Hz refresh or higher, up to 7,680 Hz. On the Grating Series, 119.63% of the DCI-P3 volume at 95.5% coverage.2 Visual comfort holds too: VICO ≤ 1.8, with 415–480 nm blue-light energy attenuated by 40% or more.
The point
Most COB specifications quote contrast in a black room — a room your client will never watch in. We publish measured black-state luminance and ambient contrast alongside gamut, uniformity and refresh, because integrators deserve numbers, not adjectives.
The package is a commodity. The surface is the product.
1 Internal laboratory measurement, identical P1.5625 panel, bare versus with grating plus AR surface; results vary by pitch and configuration.
2 DCI-P3 refers to the SMPTE ST 2113 colour space — a performance benchmark, not a product certification. DCI has not tested, approved, or endorsed this product.
Grating Series figures apply to the Grating Series only. The Deep Black Series is a standard chip display with black-level enhancement surface treatment.
See the difference instead of reading about it. NovaVista runs remote live demos and can supply evaluation panels for side-by-side testing against your current reference wall.