How to tell a sharp screen from a sharp camera

As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.

Two pixel counts that listings print in the same breath

Almost every imaging instrument in this category has two resolutions and most listings mention both without ever distinguishing them. One belongs to the thing that gathers light or heat. The other belongs to the thing that shows you the result. They are measured in the same unit, written in the same format and have almost nothing to do with each other.

The detector figure is a hard physical limit. An array of a given size can separate two features only if they fall on different elements, and no processing recovers a separation that never existed in the measurement. The display figure is a presentation choice. It determines how finely the captured frame can be drawn, how much interface furniture fits around it and how legible the numbers printed over the image are. A display can always be made finer than the detector, and on thermal instruments it very often is.

The HSFTOOLS HP96 is a clear illustration because it publishes three figures at once: a 96 by 96 infrared sensor, a 640 by 480 visible light camera and a 3.5 inch touchscreen, alongside a 50 degree field of view, automatic tracking of maximum, minimum and centre temperatures and a set of imaging modes that blend the two cameras. Two of those three figures are counts of pixels and they differ by a factor of more than thirty.

Counting how many screen pixels one detector pixel must fill

The arithmetic is worth doing once because the result is so lopsided. A 96 by 96 array contains 9,216 elements. A 320 by 240 array contains 76,800. A display of 480 by 640 contains 307,200 pixels. Divide the display count by the detector count and you get the number of screen pixels each measured value has to be spread across before any menu or cursor is drawn over the top.

The second row comes from the MiLESEEY TR20, which states a 320 by 240 infrared array, a 25 Hz refresh rate, a sensitivity below 40 mK, a 50 degree wide angle lens and a 2.8 inch display at 480 by 640. That is an unusually complete set of figures, and it shows a ratio of four screen pixels per measured value, which is a sane design. The last row is the common case: a camera figure with no display figure beside it, where the comparison cannot be made at all.

Why a thermal detector count is the harder ceiling

Visible sensors and thermal detectors are not built the same way and do not scale the same way. A photosite on a visible sensor is a small well that collects photons and can be made very small indeed. A thermal detector element is a microbolometer: a suspended plate that has to absorb infrared radiation, change temperature measurably and stay thermally isolated from its neighbours. Isolation requires physical space, so the elements stay comparatively large and the arrays stay comparatively small.

That is why a single instrument can carry a 640 by 480 visible camera next to a 96 by 96 thermal array without anything being wrong. The two numbers come from different manufacturing problems. It also explains why blended and picture in picture modes exist: they borrow the visible camera to supply the outlines the thermal array cannot resolve, so the operator can tell which warm blob is which component.

Element count is only half the thermal story. A detector that cannot distinguish a small temperature difference from its own noise will produce a grainy image however many elements it has, and the figure that governs that is sensitivity rather than resolution. The MiLESEEY figure of below 40 mK is a claim about that second property, and what it means in practice is set out in what NETD means on a thermal camera. The broader question of which instrument suits which kind of hidden fault is covered in the comparison between a thermal camera and a borescope.

The moving test that separates the two

There is a bench test that settles the question in under a minute and needs nothing but a target with repeating fine detail. A printed page, a brick wall, a coarse file, a radiator with closely spaced fins for a thermal instrument. Point the camera at it from a distance where the detail is just about distinguishable, then close the gap steadily and watch what happens to the features at the limit of visibility.

If new separations appear as you approach, features that were one smudge resolving into two, then the detector had spare capability at the original distance and something else was the limit. If the existing features simply grow larger and their edges grow softer without any new detail emerging, the detector was already at its limit and magnifying its output changes nothing. The same test works on a probe: advance the tip within the focus window rather than walking forward, and watch for the same two outcomes.

A second check applies to instruments with two cameras. Switch from the thermal channel to the visible one on the same target at the same distance. The visible channel shows what the optics and the geometry allow, so the gap between the two views is a direct demonstration of the thermal detector being the ceiling rather than the screen or the lens.

Detector and display figures across these four listings

Three rows out of four give a diagonal in inches and no pixel count, which is the normal pattern and the reason the comparison usually cannot be completed from a listing alone. The DEPSTECH DS620 is interesting for a different reason: it states three separate lenses selected by a button, each capturing 2 MP stills and 1080P video at a 76 degree field of view, on a 16.5 foot semi rigid cable with ten adjustable LEDs. Three lenses is a genuine optical feature rather than a display one, which puts it in a different category from the single lens cameras reviewed here. The Soliluxelink borescope sits at the other end, publishing 1080P and a 4.3 inch screen with a 16.4 foot metal cable, eight LEDs and a battery rated for five hours, and leaving the reader with one pixel count and no way to pair it with the display.

Reading the specification line before the screen size

Work through a listing in a fixed order and the confusion disappears. Find the sensor or detector figure first and write it down. Find the display figure second, and if it is given only as a diagonal, note that the comparison is unavailable rather than assuming the two match. Check whether any resolution figure is marked as enhanced, upscaled or super resolution, because those describe output after processing rather than what was measured. Only then look at the screen size, and treat it as a comfort specification.

The judgement that follows is usually straightforward. If the detector is coarse, a large display will not help you see smaller faults but will help you read values and interpret patterns on site, which is a real benefit for long days of inspection. If the detector is fine and the display is small, you may be throwing away capability that a later export to a computer would recover. Matching the two is the point, and the same reasoning applies when deciding between instrument types altogether, as in a thermal camera against an infrared thermometer and across the individual camera reviews on this site.

Frequently Asked Questions