How to read a monitor spec sheet

Every monitor tells you its diagonal and its resolution and leaves the useful figure implicit. Pixels per inch is the first step — √(width² + height²) ÷ diagonal — and across this catalog it runs from 91 on the BenQ ZOWIE XL2566X+ to 232 on the UPERFECT GR19BU, with the densest desktop panel — the Samsung ViewFinity S9 — at 218.

But PPI cannot be compared between two screens you sit different distances from, and that is exactly what people try to do with it. A travel panel on a tray table and a 32-inch display two feet away are not on the same scale. So the figure this site sorts and ranks by is pixels per degree of vision: pixels per inch multiplied by the viewing distance and by π/180. At the 24-inch desk distance used throughout, that is PPI × 0.419.

pixels per inch = √(w² + h²) ÷ diagonal
pixels per degree = PPI × 24 × π / 180

The threshold is 60, and it is not a marketing figure: one arcminute per pixel is the resolving limit of 20/20 vision and an arcminute is a sixtieth of a degree. Above 60 pixels per degree the grid is gone; between 40 and 60 text is clean but pixel edges show if you look; below 40 the grid is visible at a desk. This catalog splits 23 / 56 / 5 across those three bands, with a median of 57.

The sharpest panel here is the UPERFECT GR19BU at 97 pixels per degree, and it is a 19-inch screen — which is the entire argument for using this figure instead of the diagonal. The softest is the BenQ ZOWIE XL2566X+ at 38, and it is deliberately soft: it is built for frame rate, and nobody reads body text during a match.

Four families, and each has a characteristic weakness worth knowing before you buy rather than after. IPS — 43 here, plus 5 of the faster variant — gives accurate colour and wide viewing angles, and glows in the corners on dark content. That glow is a panel lottery, not a defect the maker screens for, which is why it is the single most common first-week complaint about an otherwise good monitor.

TN is the oldest family and survives only where speed is everything: 2 panels here, all of them esports monitors. Colour and viewing angles are visibly worse and the response times are genuinely the fastest available in an LCD.

Mini-LED is not a panel type at all — it is an LCD with its backlight divided into hundreds or thousands of small dimmable zones. 5 here, and the reason to want one is that it produces real HDR contrast without any wear mechanism. Its failure mode is blooming: a bright object on a dark field lights its whole zone, so a mouse cursor on a black background drags a faint halo. The Samsung Odyssey Neo G9 G95NC has 2,392 zones, which is enough that the halo is small; a backlight with a few dozen zones has zones the size of a window.

OLED is emissive: each pixel makes its own light, so contrast is effectively infinite and response is measured in hundredths of a millisecond. There are 28 here — 12 WOLED and 16 QD-OLED. The failure modes are real and specific: full-field brightness far below an LCD’s, coloured fringing on small text from the non-standard subpixel layout, and differential wear. Why the text fringes, and what the warranty covers.

One specification that sits with the panel rather than with the picture: the coating. 65 monitors here are matte and 8 are glossy. A glossy OLED has deeper-looking blacks and mirrors the room; a matte one spreads a lamp into a dull patch and lays a faint grain over large white areas. Neither is better, and the room decides.

Refresh rate is how often the panel can change, and its reciprocal is what your eye responds to: 60 Hz holds each frame for 16.7 ms, 144 Hz for 6.9, 240 Hz for 4.2, and the fastest panel here — the Acer Predator X27U F5 at 500 Hz — for about 2 ms. Motion blur on a sample-and-hold display comes from the frame standing still while your eye tracks across it, so halving persistence halves that blur. Each doubling gives a real improvement and a smaller absolute one than the last.

Response time is the other half, and it is where the marketing gets loose. A rated grey-to-grey figure is the best transition the panel manages with overdrive applied; its slow dark-to-dark transitions can be several times that. At 500 Hz a frame lasts 2 ms, so a pixel with a real 5 ms transition has not finished changing before the next frame arrives, and you are seeing a smeared average rather than distinct frames. This is why an OLED at 240 Hz reads as clearer in motion than an LCD at 360.

The distribution here is a reasonable map of the market: 32 monitors at 240 Hz or faster, 15 at 60 Hz or below, and a median of 165 Hz. Where the diminishing returns actually start.

A fixed-refresh display and a variable frame rate cannot agree, and the two ways to reconcile them are both bad: tear the frame, or hold it and hitch. Variable refresh fixes the argument by letting the display wait for the GPU. It is the single largest improvement to perceived smoothness available, and it matters more than another 120 Hz of headroom.

The badges are certification programmes rather than different technologies. Plain Adaptive-Sync is the VESA baseline. FreeSync adds AMD’s own testing; FreeSync Premium requires at least 120 Hz at 1080p plus low-framerate compensation, which matters because it is what keeps the display working when the frame rate falls below the panel’s minimum; FreeSync Premium Pro adds HDR requirements. G-Sync Compatible means Nvidia tested the monitor’s Adaptive-Sync implementation and it passed. Modern GPUs from both vendors drive all of them.

The specification nobody prints is the one that bites: the variable range’s lower bound. A panel with a 48-to-240 Hz window and no low-framerate compensation stops varying below 48 frames a second, which is exactly when you need it. And on OLED specifically, variable refresh can produce visible brightness flicker in dark scenes when frame times are uneven — a known characteristic of the panel type rather than a fault in any particular monitor.

VESA’s DisplayHDR 400 tier asks for 400 nits peak, 95% of sRGB and 8-bit input processing. It does not require wide colour gamut, it does not require 10-bit, and it explicitly does not require local dimming of any kind. A monitor can therefore carry the badge, accept an HDR signal, switch to an HDR tone curve, and show you a flatter and greyer picture than its own SDR mode — which is the most common HDR experience there is.

So this site does not rank by the badge. It asks what the hardware can physically do, and there are only two ways to make an HDR image: every pixel produces its own light, or the backlight is bright enough and divided enough to put a highlight beside a shadow. The threshold used here is 600 claimed nits with at least one dimming zone. By that test 38 of the 84 monitors here can display HDR. The rest split in two: 21 carry an HDR certificate their backlight cannot honour — they accept the signal and change the tone curve, nothing more, the LG UltraGear 27GR93U among them — and 25 make no HDR claim at all, which their pages say outright rather than dressing up as a tier.

Brightness alone is not the story either, because HDR is a contrast claim. The brightest claim in this catalog is 1,600 nits on the ASUS ProArt PA32UCXR, and an OLED rated at a third of that can look more convincing, because it can put a bright highlight next to an absolutely black pixel. Full-field brightness is the figure OLED loses on: the dimmest SDR rating here is 200 nits on the LG UltraGear 27GR95QE, which is a real constraint in a bright room.

One number is treated differently from all the others on this site. A peak-brightness figure that a named third party measured is marked in its own colour wherever it appears, and there are 12 of them here, because a figure somebody put a meter behind is a different kind of fact from a figure on a box. Why HDR mode often looks worse.

A monitor’s resolution and refresh rate are a claim about the panel, not about the port, and the two disagree more often than the box admits. The arithmetic is not controversial:

required Gbps = width × height × refresh × bits-per-pixel × 1.20

1.20 is blanking overhead — 4K60’s CTA-861 total timing is 4,400 × 2,250 against an active area of 3,840 × 2,160, which is 1.19× — and bits per pixel is 30 for a panel that takes a 10-bit signal, 24 for one that does not. The hungriest panel in this catalog is the Samsung Odyssey Neo G9 G95NC, which needs 143 Gbps.

Link rates are the usable data rates after line coding, not the marketing figures: HDMI 2.0 signals at 18 Gbps and delivers 14.4 after 8b/10b; HDMI 2.1 signals at 48 and delivers 42.67 after 16b/18b; DisplayPort HBR3 signals at 32.4 and delivers 25.92; the UHBR tiers use 128b/132b, which is why UHBR20 keeps 77.37 of 80. And “HDMI 2.1” is a specification version rather than a bandwidth guarantee — a port can be labelled 2.1 while running at 24 Gbps, because every 2.0-era capability was folded into the 2.1 document.

Display Stream Compression closes the gap, at up to 3:1 and visually losslessly, which is how a 4K 240 Hz monitor exists on a DisplayPort 1.4 input at all. 38 monitors here state that they support it, and 7 have a maker that publishes no DSC row at all — where that decides whether an input carries the panel, those pages say so instead of guessing. Without it a panel cannot exceed its port, and 3 monitors here, among them the LG UltraWide 40WP95C, cannot reach their advertised refresh rate at 10 bits per channel over any input, and run at 8 bits instead, which is what the specification sheet expresses by not mentioning bit depth. A further 8 cannot reach their rated refresh over any input at any bit depth, because they are 8-bit panels already: their headline refresh rate needs either a lower resolution or the reduced-blanking timing a DisplayPort link can use in place of the television timing this figure assumes.

Where this decides a purchase outright is a console, which has HDMI and nothing else. 41 monitors here have an HDMI 2.1 input. 43 have an HDMI input of any version that carries the panel’s own maximum. The two sets are not nested — a 1080p portable clears its own panel over HDMI 2.0, while a 1440p 240 Hz OLED does not clear its own over HDMI 2.1 uncompressed. Every monitor page runs this arithmetic per input and prints the result. The full method.

47 monitors here take video over USB-C, which means they can replace a dock: video in, power out, peripherals through the monitor’s own hub, one cable to the laptop. The wattage is what decides whether it actually works. Power delivery negotiates — the laptop asks, the monitor grants up to its ceiling — so a 65-watt monitor and a 140-watt laptop will settle on 65, and the laptop will discharge under sustained load while the operating system shows a charging icon.

Sixty-five watts holds a 13-inch ultrabook, 100 covers a 14-inch machine, and 140 is what a 16-inch workstation wants at full draw. 25 monitors here deliver 90 W or more, and the ceiling is 140 W on the Dell UltraSharp U2725QE.

Thunderbolt — 12 monitors — tunnels PCI Express alongside DisplayPort, which is what allows a daisy-chained second display or a fast external drive through the monitor. For a laptop plus a keyboard, a mouse and a webcam, plain USB-C DisplayPort Alt Mode does the same job for less.

The specification retail sites leave off entirely is the KVM switch: one keyboard and mouse plugged into the monitor, following whichever input is on screen, so a work laptop and a desktop share the peripherals. 28 monitors here have one. What the wattage has to be.

Gamut is which colours the panel can reach; bit depth is how finely it can subdivide the space between them. They are independent, and a monitor can be good at one and poor at the other.

Coverage is quoted against DCI-P3 for anything sold on colour, and the figures in this catalog run to 100% on the ASUS ZenScreen MB16QHG with a median of 98.5%. 6 monitors here publish no gamut figure in any colour space, and their pages leave that row out rather than convert one from another space — an NTSC or sRGB percentage relabelled as DCI-P3 is a different measurement, not a rounded one. Two further cautions. A percentage can be coverage of the target or volume relative to it, and the second number is always larger and means less. And a wide-gamut panel with no sRGB clamp oversaturates everything that is not colour-managed, which on Windows is most things — so a good clamp mode is worth more day to day than the last five percent of coverage.

Bit depth has three honest answers and this catalog uses all three. True 10-bit panels — 41 here — address 1,024 levels per channel. 8-bit+FRC panels dither between 8-bit levels fast enough to simulate 10, which is effective and is not the same thing. Plain 8-bit gives 256 levels per channel and shows visible banding in gradients, most obviously in a dark sky.

Bit depth is also the hidden cost in the bandwidth arithmetic above: 30 bits per pixel against 24 is a 25% larger signal, which is precisely why several monitors here reach their headline refresh rate only at 8 bits.

A curve radius is written in millimetres with an R: 1000R means the panel is a section of a circle with a one-metre radius, so the ideal viewing distance is one metre. A smaller number is a tighter curve. 15 monitors here are curved and the tightest is 800R on the LG UltraGear 34GS95QE.

The purpose is to equalise distance. On a flat 34-inch panel the far edges are measurably further from your eye than the centre and are seen at a sharper angle, which costs both focus and — on a VA or IPS panel — contrast. That is why almost every ultrawide past 34 inches is curved, and why a curve on a 24-inch 16:9 monitor is mostly decoration.

Aspect ratio is the related decision. 21:9 at 3440 × 1440 is the common ultrawide and it is exactly as tall as a 27-inch 1440p monitor; the 5120 × 2160 tier is the genuine upgrade, because it adds 720 lines of height as well as width. 32:9 is two 16:9 panels joined, and the largest here — the Samsung Odyssey Neo G9 G95NC at 57 inches — is the extreme of that idea. One ultrawide or two monitors.

Two mounting specifications belong here too, because they are the ones people discover after delivery. 45 monitors here pivot into portrait, which is the best arrangement there is for reading long documents or code. And 3 have no VESA mount at all — all of them portables — so a monitor arm is not an option however much desk you want back.

macOS renders the interface at exactly twice the logical size and downsamples the frame to fit the panel. It is uniform and pixel-perfect when the panel’s resolution is exactly twice a comfortable logical one, and slightly soft at every other factor, because the downsample becomes a fractional resample.

Windows takes the other approach: each application is told a scale factor and lays itself out accordingly. That is sharp at any factor for applications that participate, and produces bitmap-stretched blur or postage-stamp windows for the ones that do not.

The practical consequence is a different shortlist per machine. Apple designs around 218 PPI, which is why 27-inch 5K exists: 5120 × 2880 is exactly twice 2560 × 1440, and that is what puts the Samsung ViewFinity S9 at 218 PPI. A 27-inch 4K panel at 163 PPI is excellent on Windows at 150% and divisive on a Mac, where 2× gives a cramped 1920 × 1080 logical desktop. A 32-inch 4K at 140 PPI is the other sweet spot, and for the opposite reason: close enough to unscaled that many people run it at 100%.

This is also the real answer to the eye-strain complaint about large high-resolution ultrawides, which usually gets blamed on pixel density. Those panels are 140 PPI — the same as a 32-inch 4K — and at the distance they are used from they resolve better than a 27-inch 1440p monitor does. The full comparison.

Organic emitters dim in proportion to how hard they have been driven, so an OLED that has shown a bright white taskbar for two thousand hours has slightly dimmer pixels there than beside it. That is burn-in: differential wear, visible as a ghost of the interface on a uniform field, cumulative and not reversible. Blue emitters age fastest, which is why a static white element is the worst case.

Brightness is the strongest lever you control, and static high-contrast interface is the strongest risk. Mixed video and gaming is close to a non-issue on a modern panel; eight hours a day of the same three windows at full brightness is not.

Which makes the warranty the specification that matters, and specifically the cover that names image retention — a three-year panel warranty that excludes it covers nothing that actually happens to an OLED. Across the 28 OLED monitors here the stated cover runs 0 to 3 years — Samsung Odyssey OLED G8 G80SD is one of those at the top — and 2 makers publish no term at all, which this site prints as not published rather than assuming the usual three. On a monitor you expect to keep for five years, the difference between two years and three is the difference between a replacement and a purchase.

If the monitor will live on a desk showing static interface all day, a Mini-LED LCD gets most of the contrast with no wear mechanism at all — there are 5 of them here. How worried to actually be.