Screen Resolution, PPI and Pixel Density Explained

Open our device readout on a MacBook and it reports something like 1512×982 @2.0x. The panel is 3024×1964. Neither number is wrong — they’re answering different questions, and the gap between them explains most of the confusion around resolution, PPI, and “Retina.”

Four numbers, four questions

Resolution — how many pixels the panel physically has. 1920×1080 (Full HD), 2560×1440 (QHD), 3840×2160 (4K). This is a count, not a quality; a 55-inch TV and a 27-inch monitor can share a resolution and look nothing alike up close.

PPI (pixels per inch) — how densely those pixels are packed: divide the diagonal pixel count by the diagonal inches. The same 3840×2160 grid is 163 PPI on a 27-inch monitor and 80 PPI on that 55-inch TV. PPI is the number your eyes actually experience.

Viewing distance — the missing variable in every PPI argument. What matters is pixels per degree of vision: 80 PPI is razor sharp from a couch and chunky at a desk. This is why phone panels race past 450 PPI while desktop monitors do fine at 110 — the phone lives at a third of the distance.

Device pixel ratio (DPR) — how many physical pixels the operating system spends per logical “point” of UI. This is the @2.0x in our readout, and it’s the key to the MacBook mystery above.

Why high-PPI screens lie about their resolution

At 220 PPI, rendering UI at 1:1 would make every button microscopic. So the OS renders logical layout at a comfortable size and maps each logical pixel onto a 2×2 (or 3×3) block of physical ones. Software sees a 1512-wide screen; the panel draws 3024. Everything is half the size in logical units and twice the density in physical ones — that’s the whole trick behind “Retina,” Windows display scaling, and Android’s dp units.

So when our readout says 1512×982 @2.0x, read it as: logical canvas 1512×982, drawn with 2×2 physical pixels each — a 3024×1964 panel. Screenshots coming out “too big” and web images looking soft on new laptops are both this mechanism seen from different sides. (For web developers, DPR is also why you ship 2× images: a 300-pixel-wide image element on a 2× screen has 600 physical pixels to fill, and a 1× asset gets stretched.)

Fractional scaling (1.25×, 1.5×) is the messy middle: logical pixels no longer align with the physical grid, so the OS renders at a higher resolution and downsamples. Usually fine, occasionally shimmery on fine text — if a Windows monitor looks subtly soft, check whether it’s sitting at 125% or 150%.

How sharp is sharp enough?

The old “Retina” rule of thumb — the density where individual pixels stop being resolvable — works out to roughly 60 pixels per degree for typical vision. In practice:

Sharpness also saturates: past the resolvable threshold, extra pixels buy nothing your eye can collect. The upgrade from a 110 PPI to a 160+ PPI desktop monitor is transformative for text; from 160 to 220 is nice; beyond that you’re paying GPU tax (every frame is rendered at those physical dimensions — resolution is a permanent performance bill, which our benchmark deliberately caps out of the score so screens don’t skew device comparisons).

Reading your own screen

The device readout on our homepage shows the logical resolution, the DPR, and the derived numbers live, plus your measured refresh rate. Two checks worth doing on any new machine:

1. Is the DPR what you expect? An external 4K monitor at DPR 1.0 means the OS is running it unscaled — everything tiny, and text not benefiting from the density.
2. Is the resolution native? A panel fed a non-native resolution interpolates, and no amount of PPI rescues interpolated text. The test patterns make this obvious: PM5544’s fine gratings should resolve into distinct lines, not moiré soup.

And if you’re comparing two used phones at a market stall, remember the hierarchy: a clean panel at 300 PPI beats a scratched, burnt-in one at 500. Density is a luxury; uniformity is a necessity.