Resolution is the pixel grid and nothing else. A file labelled 4K has 3840 columns and 2160 rows of pixels available to it — that is the entire claim the number makes. It says nothing about whether those pixels contain real detail.
That gap is why the label disappoints people so often. Resolution is a ceiling, not a floor. Everything that happens between the camera sensor and your screen — capture resolution, compression, re-encoding, upscaling — decides how much of that ceiling gets used, and any one of those steps can leave a 4K file looking like an ordinary 1080p one.
What does the number in 1080p or 4K actually count?
Two different things, which is the source of most of the confusion.
The older labels count vertical lines: 480p is 480 rows of pixels, 720p is 720 rows, 1080p is 1080 rows. The 4K and 8K labels count horizontal pixels instead — around four thousand and eight thousand across — a convention inherited from cinema. Both naming systems now sit side by side in the same dropdown menu, which is why the sequence looks inconsistent.
| Label | Pixel grid | Pixels vs 1080p | Where the extra pixels tend to show |
|---|---|---|---|
| 480p (SD) | 854×480 | about 0.2× | Visibly soft on anything larger than a phone |
| 720p (HD) | 1280×720 | about 0.44× | Acceptable on small screens, soft on a TV |
| 1080p (Full HD) | 1920×1080 | 1× (reference) | The practical baseline for most viewing |
| 4K (UHD) | 3840×2160 | 4× | Large screens, close viewing, cropping and zooming |
| 8K | 7680×4320 | 16× | Very large screens; rarely a real source resolution |
The multipliers are worth internalising. Going from 1080p to 4K is not "a bit sharper" — it is four times as many pixels to store, transmit and decode. That cost is real whether or not the extra detail exists.
8K sits in the table for completeness rather than because you will meet it. It is far more often a container size than a capture resolution.
What does the p in 1080p mean, and does i still matter?
The p stands for progressive: every frame contains all of its lines, drawn in one pass. The i stands for interlaced, where each frame carries only the odd or even lines and the two halves are woven together.
Interlacing was a bandwidth trick from the broadcast era, and it is mostly historical now. It still matters in one situation: older material mastered for television can carry interlacing artefacts into a digital transfer, showing up as comb-shaped horizontal tearing on fast movement. Most players can deinterlace this, and most modern releases are progressive throughout, so the letter after the number is rarely a decision point today.
Why do two files with the same label look different?
Because the label describes the output grid, not the detail that filled it. Three separate stages can hollow out a resolution while leaving its name intact.
Capture resolution. If a scene was shot at 1080p, no later step can invent detail that the sensor never recorded. Exporting it at 3840×2160 produces a 4K file containing 1080p worth of information.
Upscaling. Software enlarges the frame by interpolating new pixels from existing ones. Modern upscalers are good at edges and poor at fine texture — they can make a picture look cleaner without making it more detailed.
Re-encoding. Every time a file is compressed again, information is discarded. A 4K file that has been through several rounds of re-encoding can carry less usable detail than a 1080p original that was encoded once and left alone.
The practical consequence: a resolution label tells you what a file cannot exceed, and nothing about what it contains.
How can you tell an upscaled file from a native one?
Look at fine texture on a paused frame, not at the overall impression.
- Edges look sharp but surfaces look empty. Upscalers reconstruct boundaries well. Skin pores, individual hairs, fabric weave and printed text are where invented pixels give themselves away.
- Halos and ringing around high-contrast edges. A bright outline hugging a dark boundary is a sharpening artefact, not detail.
- Pausing does not help. On a genuine high-resolution source, freezing the frame reveals more than motion allowed. On an upscale, the still frame looks the same as the moving one — soft in the same places.
- The file is unexpectedly small for its label. Real 4K detail costs data to store. A 4K file no larger than a typical 1080p one has had that detail compressed away, whether or not it was ever captured.
Which resolution should you actually care about?
Match it to your screen and your distance from it — beyond a certain point the pixels fall below what your eye can resolve, and additional ones change nothing you can perceive.
The broadcasting standards bodies put a number on that point. ITU-R defines the optimal viewing distance as the one at which a single pixel spans one arcminute of visual angle, which it treats as corresponding to a visual acuity of 1.0. Expressed as a multiple of picture height, that works out at about 4.8× picture height for 720p, 3.1× for 1080p, 1.5× for 4K and 0.75× for 8K. Further back than the figure for your resolution and the extra pixels stop being resolvable — and the 4K figure in particular puts you considerably closer to the screen than most people actually sit.
Source: Report ITU-R BT.2246-9 (2025), §2.5.2 and Table 1; Recommendation ITU-R BT.1845-1; checked 2026-08-03.
| Your viewing setup | Resolution worth paying for | What is more likely limiting you |
|---|---|---|
| Phone, held at normal distance | 1080p | Encode quality and screen brightness |
| Laptop or tablet | 1080p | Bitrate; ambient light and reflections |
| Desktop monitor, close up | 1080p to 4K | Whether the source was natively captured at 4K |
| Large TV, sitting back | 1080p to 4K | Compression during streaming delivery |
| Large TV or projector, close up | 4K | Storage and bandwidth cost |
| VR headset | As high as available | Headset panel resolution and lens quality |
The right-hand column is the point of the table. In most setups, the thing degrading your picture is not a shortage of pixels — it is how heavily those pixels were compressed. That is a separate variable, and usually the one worth optimising first.
VR is the genuine exception. A headset magnifies the image across a wide field of view, so the same pixel count is spread much thinner than on a flat screen. Resolution numbers that are academic on a phone become clearly visible there.
To give the scale: current standalone headsets publish per-eye panel resolutions in the region of 2064×2208 to 2448×2448, which sounds enormous next to a flat screen until you remember it is being magnified across roughly a hundred degrees of vision rather than the thirty or so a correctly placed TV occupies.
Sources: vendor specification pages (developers.meta.com, picoxr.com, vive.com); Report ITU-R BT.2246-9 (2025), Table 1; checked 2026-08-03. Panel figures change with each hardware generation.
Does resolution matter when you are trying to identify someone in a frame?
Much less than the whole-frame resolution suggests, because face matching works on the face region alone.
A face detector crops the face out and normalises it before encoding it as a vector. What matters is how many pixels landed on the face and whether the landmarks — eye centres, nose tip, mouth corners — are visible. A 720p frame with a large, unobstructed, roughly front-facing face carries more usable information than a 4K frame where the subject is small, turned away, or in shadow.
So if a screenshot is not returning results, raising the capture resolution is rarely the fix. Choosing a different moment in the same scene almost always helps more.
Related questions
- Why does bitrate matter more than resolution?
- Can you actually see the difference between 4K and 1080p?
- What do the numbers in VR video formats mean?
- What actually changes between mobile and desktop viewing?
Resolution labels vary in how they are stated from one catalogue to the next — our site list shows which sources we index.