More Zones, More Problems
Local dimming raises LCD contrast by controlling the backlight in sectors. More zones tighten the dark floor — but each zone boundary is a potential halo.
How the Geometry Works
An LCD panel cannot produce black on its own. The liquid-crystal layer attenuates light; it does not extinguish it. Even the best VA cell, closed as tightly as physics allows, transmits a fraction of a percent of incident light — which is why a full-on/full-off contrast ratio measured in a dark room still depends largely on how dark you can make the backlight, not how tightly the cell shuts.
Local dimming addresses this by dividing the backlight into independently driven zones. In edge-lit designs, a row of LEDs along one or two edges feeds a light-guide plate; zones are carved out by controlling LED clusters, but the light-guide itself smears the dimming spatially, typically yielding 8–32 effective zones before the blur becomes indefensible. Direct-lit (full-array) designs mount LED clusters directly behind the panel, giving genuine two-dimensional control. A 65-inch panel with a 5 × 9 grid of LED clusters has 45 zones; high-end panels now ship with hundreds, and some Mini-LED implementations — using LEDs roughly 100–200 µm across — push into the low thousands.
The gain is real and measurable. A single-zone backlight running at full brightness to support a small white object on a black field sets the black floor across the entire screen. Dim that region independently and the local contrast ratio climbs by an order of magnitude or more, which is why peak nits and the conditions that produce them are meaningless without also specifying what happens to the surround.
Why More Zones Introduce Their Own Artefacts
Each zone has a finite physical extent, so the dimming boundary does not coincide with image content boundaries. A bright star in a dark sky subtends a few pixels. The nearest zone boundary may be centimetres away in panel coordinates, meaning the algorithm must either illuminate the entire zone — creating a visible halo of elevated luminance around the star — or clip the zone to nearly black and accept that the star's own peak brightness is suppressed.
The algorithm must walk this trade-off frame by frame. Common strategies include computing the maximum pixel luminance within each zone and targeting the backlight to support that peak, then compensating in the LCD cell by scaling pixel values to match. This works when content is static. When content moves, the zone luminance must track it: a bright object crossing a zone boundary triggers a rapid drive change in two adjacent zones simultaneously. Slew-rate limits on LED current, combined with the milliseconds of thermal rise in the LED junction, mean the backlight cannot keep up with fast-moving bright objects without either producing trailing halos or applying predictive lookahead — which introduces latency, typically one to two frames, into the display pipeline.
More zones compress the halo geographically: a 1,000-zone panel's halo is smaller than a 45-zone panel's. But the halo does not disappear; it simply becomes subtler. It also becomes more irregular, because fine zone grids interact with image content in complex ways, and aggressive algorithms can produce a flickering halo that attracts the eye precisely because temporal contrast sensitivity is high in peripheral vision. Tuning the dimming algorithm — attack time, release time, minimum zone floor, pixel compensation curve — is where engineering hours actually go, and it is why two panels with identical zone counts can produce very different subjective results.
Zone count is the headline. Algorithm quality is the product.