Do megapixels matter? Sensor size, pixels and real detail

Twelve megapixels already fill an A4 print at 300 ppi. Past that, lens diffraction, pixel size and light decide how much real detail you get.

The exposed 26.2-megapixel full-frame CMOS sensor inside a Canon EOS RP mirrorless camera
Photo: Islander61, CC BY-SA 4.0, via Wikimedia Commons

Megapixels matter only up to the point where something else becomes the limit, and for most people that point is around 12 to 24 MP. A 12 MP image already prints at A4 size near 300 pixels per inch, about what a sharp eye can resolve at 30 cm. Beyond that, extra pixels add real detail only if the lens is sharp enough, the aperture is wide enough to beat diffraction, and there is enough light for each smaller pixel; that is why 200 MP phone sensors save 12 to 50 MP files by default.

Key takeaways
  • 8.7 MP fills an A4 print at 300 ppi; 12 MP covers screens and A4 with room to spare.
  • Diffraction caps detail: a perfect f/8 lens blurs a point to about 10.7 µm.
  • 0.6 µm pixels on a 200 MP phone sensor are smaller than the lens can feed.
  • Binning 16 pixels into one gains about 4× signal-to-noise.
  • Sensor area, not pixel count, sets how much light the whole image collects.

The question keeps coming up because spec sheets keep climbing. At MWC in March 2026, the vivo X300 Ultra and OPPO Find X9 Ultra each carried two 200 MP cameras, while full-frame cameras for professionals still ship at 24 to 61 MP. This post works through the numbers:

  • how many pixels a print or screen can use;
  • what a lens can actually deliver to a pixel;
  • what happens to noise as pixels shrink;
  • when more megapixels are genuinely worth paying for.
The story

On 27 February 2012, at Mobile World Congress in Barcelona, Nokia unveiled the 808 PureView with a 41-megapixel sensor. The headline number hid the real idea. The sensor was large for a phone, 1/1.2-inch, with 1.4 µm pixels behind a Carl Zeiss f/2.4 lens, and by default the camera did not save 41 MP files at all. It combined pixels and saved 5 MP images, turning the spare resolution into cleaner pixels.

The phone won best new handset at MWC 2012 and a Gold Award from Digital Photography Review. A decade later, every 200 MP phone sensor works the same way: count huge, save small.

Wikipedia: Nokia 808 PureView (launch, sensor and 5 MP oversampling)
0.6 µmpixel pitch of Samsung's 200 MP ISOCELL HP2
3.76 µmAiry disk diameter of a perfect f/2.8 lens at 550 nm
8.7 MPneeded for an A4 print at 300 ppi

How many megapixels do you need for a print?

An A4 print at 300 pixels per inch needs about 8.7 MP; an A3 needs about 17.4 MP. A print's pixel needs are just its width and height in inches multiplied by the target pixels per inch (ppi), and many commercial printers, such as PsPrint, ask for 300 ppi at the final printed size.

Why 300 ppi

The 300 ppi figure is not arbitrary. Normal 20/20 vision resolves detail of about 1 arcminute (1/60 of a degree). At a 30 cm reading distance, 1 arcminute spans 30 × tan(1/60°) ≈ 0.087 mm, which works out to about 290 dots per inch.

At arm's length the number falls fast:

  • a poster seen from 1 m needs about 87 ppi;
  • a hoarding seen from 20 m needs under 5 ppi.

That is why a 6 MP image can make a perfectly good billboard and why the same file looks soft in a photo book held in the hand.

Megapixels needed to print at 300 ppi
See the data as a table
Megapixels at 300 ppi (MP)
6 × 4 in postcard2.2
A54.4
A48.7
A317.4
A234.8
A169.6

Source: Calculated from ISO 216 paper sizes at 300 ppi (PsPrint artwork requirement) · A4 is 8.27 × 11.69 in, so 2481 × 3508 pixels. Each step up the A-series doubles the area and the pixel count.

Viewing distancePixel pitch the eye resolves (1 arcmin)ppi neededLargest print from 24 MP (6000 × 4000)
30 cm (hand-held print)0.087 mm29152 × 35 cm
60 cm (desk, wall at close range)0.175 mm145105 × 70 cm
1 m (wall print)0.29 mm87175 × 117 cm
5 m (hall banner)1.45 mm178.7 × 5.8 m
20 m (roadside hoarding)5.8 mm4.435 × 23 m

Screens need even fewer pixels. A 4K television has 3840 × 2160 pixels, or 8.3 MP, and a 1440p monitor 3.7 MP. A 50 MP photo shown full-screen on either is resized down to well under a fifth of its pixels.

Do more megapixels give more detail?

Only if the lens can put that much detail on the sensor. Light passing through a circular aperture spreads into a blur spot, the Airy disk, whose diameter is about 2.44 × wavelength × f-number. For green light at 550 nm, that is 3.76 µm at f/2.8 and about 10.7 µm at f/8, even for an optically perfect lens.

Roadmap for CMOS image sensors: Moore meets Planck and SommerfeldPeter B. Catrysse and Brian A. Wandell · Proceedings of SPIE 5678, Digital Photography · 2005

Using Stanford's ISET camera simulator, the authors asked what shrinking pixels does to image quality given two fixed properties of light: photon noise and diffraction. For a monochrome sensor, pixel pitch needs to be about the radius of the Airy disk; finer sampling adds no real resolution.

To avoid aliasing across all visible wavelengths, a diffraction-limited f/2.8 lens needs pixels of 1.36 µm or smaller, and an f/8 lens about 3.9 µm. On noise, they simulated an ideal 3 µm pixel and showed colour noise plainly visible at f/2.8 in a dim 10 cd/m² scene (about 0.01 lux-seconds) and an unusable image at f/8, because smaller pixels and smaller apertures both cut the photons each pixel counts. Limit: ideal lenses and sensors; real ones are worse.

Read the paper (free)

Apply that to a phone. A main camera at f/1.7 has an Airy disk about 2.3 µm across at 550 nm, so a radius of about 1.14 µm. Samsung's 200 MP ISOCELL HP2 has 0.6 µm pixels, roughly half the Airy radius, meaning the sensor oversamples even a perfect lens. Real phone lenses, made of moulded plastic elements a few millimetres across, blur further, especially at the corners. The 200 MP readout is useful mainly for cropping and digital zoom in bright light, not for full-frame detail.

0.6 µm pixels (200 MP mode)2.4 µm binned pixel (12.5 MP mode)One blur spot covers about 11 small pixelsThe spot fits inside one large pixelAiry disk, f/1.7≈ 2.3 µm
The same diffraction blur from a perfect f/1.7 lens, drawn to scale over the HP2's native 0.6 µm pixels and over its 16-to-1 binned 2.4 µm pixel. Small pixels sample the blur finely but cannot find detail the lens never delivered.

Why do 200 MP phone cameras save 12 MP photos?

Because combining pixels cuts noise and the lens cannot feed 200 MP anyway. This is pixel binning. The HP2, announced in January 2023 for the Galaxy S23 Ultra, groups pixels in 2 × 2 blocks to give a 50 MP image with effective 1.2 µm pixels, or in 4 × 4 blocks for 12.5 MP at 2.4 µm. Sony's LYT-901, launched in November 2025 with 0.7 µm pixels on a 1/1.12-inch sensor, uses the same 50 MP and 12.5 MP binning modes.

16 × 0.6 µm pixels200 MP, full detail4 × 1.2 µm (2 × 2 bin)50 MP1 × 2.4 µm (4 × 4 bin)12.5 MP, lowest noiseSame patch of sensor, three read-out modes (Samsung ISOCELL HP2)
Binning adds the signal of neighbouring pixels. Photon noise grows only as the square root of the signal, so a 16-pixel bin gains about four times the signal-to-noise ratio of one small pixel. The colour filters are arranged so each binned block still records one colour.

The square-root law of photon noise

The physics of noise is simple. Light arrives as photons, and the count in any pixel varies randomly with a standard deviation equal to the square root of the mean count.

Photons collectedNoise (√N)Signal-to-noise
100 (one small pixel)1010
400 (4× the area, or a 2 × 2 bin)2020
1,600 (16× the area, or a 4 × 4 bin)4040

Shrinking pixels to fit more on the same sensor therefore lowers per-pixel quality, and binning reverses it at the cost of resolution.

Catrysse and Wandell's simulations show this is a law of light, not a technology problem to be engineered away.
Microscope view of the red, green and blue Bayer colour filter mosaic on an old Sony camcorder CCD sensor
An RGGB Bayer filter seen at 80× magnification. Each pixel records only one colour, so a sensor's "megapixels" are not full-colour samples; the camera interpolates the missing two-thirds of the colour data.Photo: Binarysequence, CC BY-SA 3.0, via Wikimedia Commons

Is sensor size more important than megapixels?

For image quality in anything but bright light, yes, because sensor area sets how much light the whole image collects. A full-frame sensor (36 × 24 mm, 864 mm²) has about 12 times the area of a 1/1.3-inch phone sensor (about 9.8 × 7.3 mm, 72 mm², for the HP2), so at the same exposure it gathers about 12 times the photons for the whole picture, whatever the pixel count.

Within one sensor size, more megapixels mostly trade per-pixel noise for finer sampling, and when you resize to the same output, the total noise is similar. The older Chen et al. study set out the trade-off formally, and the conclusion still holds for modern sensors: there is an optimum pixel size for a given technology, and it is not "as small as possible".

How small should pixel size be?Ting Chen, Peter Catrysse, Abbas El Gamal and Brian Wandell · Proceedings of SPIE 3965 · 2000

The Stanford group modelled CMOS active-pixel sensors and scored simulated images with the S-CIELAB perceptual colour metric instead of a single engineering number.

Smaller pixels gave higher spatial resolution and better modulation transfer, but larger pixels gave more dynamic range and signal-to-noise ratio, so an optimum exists. For the 0.35 µm CMOS process of the time, the best pixel was about 6.5 µm at a 30% fill factor; the optimum shrinks as process technology improves, but more slowly than the transistors do. Limit: the numbers belong to 2000-era processes, and backside illumination and microlenses have since moved the optimum smaller.

Read the paper (free)
A large full-frame CMOS image sensor next to a tiny camera sensor for size comparison
A full-frame sensor from a Canon EOS 5D Mark III beside a tiny hobby-camera sensor. At the same exposure, the area difference, not the pixel count, decides how much light the image collects.Photo: 根川孝太郎 (Kotaro Negawa), CC BY-SA 3.0, via Wikimedia Commons

When do more megapixels actually help?

More megapixels help when youThey help much less when you
crop hard (wildlife, sport)shoot in low light
print big and view closeuse cheap zooms
use sharp lenses at moderate aperturesstop down to f/11 or smaller
shoot in good lightonly post to screens
f-numberAiry disk diameter at 550 nmPixel pitch that stops adding detail (Airy radius)Rough full-frame MP at that pitch
f/1.82.4 µm1.2 µm600
f/2.83.8 µm1.9 µm240
f/45.4 µm2.7 µm119
f/5.67.5 µm3.8 µm60
f/810.7 µm5.4 µm30
f/1114.8 µm7.4 µm16
f/1621.5 µm10.7 µm7.5

Read the table as a ceiling, not a target. A 61 MP full-frame sensor has 3.76 µm pixels, matched to a perfect lens at about f/5.6; stop down to f/11 for depth of field and the sensor has nearly four times the pixels the optics can feed.

My own Sony ZV-E10 has a 24.2 MP APS-C sensor (23.5 × 15.6 mm, so about 3.9 µm pixels), which is diffraction-matched at roughly f/5.6 too. See depth of field explained for why landscape shooters stop down anyway, and the exposure triangle for what that costs in shutter speed and ISO.

How big can you print your photo?

Use this to check what you just read. Enter your image's pixel dimensions, the ppi you plan to print at, and the distance people will view it from.

Print size and the ppi your viewers need

The defaults show the usual surprise: a 24 MP file viewed from 60 cm can be printed about a metre wide before a sharp-sighted viewer can see its pixels.

For vector logos the question disappears entirely; our vector vs raster guide explains why.

So should you buy for megapixels?

For a phone, compare sensor size, lens aperture and processing before pixel counts, because every flagship bins its 50 to 200 MP sensor anyway. The DXOMARK phone camera ranking and our explainer on computational photography show how much of today's phone image comes from multi-frame stacking rather than a single high-resolution exposure.

For a camera, 24 MP covers events, travel, video and A3 prints. Move to 40 to 61 MP if you:

  • crop wildlife or sport heavily;
  • shoot landscapes for large prints;
  • copy artwork.

Budget for lenses sharp enough to use the pixels and for files twice the size: an uncompressed 14-bit RAW from a 61 MP sensor is over 100 MB.

Shooting in RAW and exposing well, covered in the RAW and ETTR guide, often gains more real quality than another 20 MP. If you want to see the idea taken to its absurd end, read what if a photo had a trillion megapixels. My own kit is listed on the gear page.

A small CMOS image sensor in a glass chip-scale package, with the dark rectangular active area visible
A small CMOS sensor in a glass chip-scale package, the kind used in phones and webcams. Packing more pixels into an active area a few millimetres wide is what drives pixel pitch below one micrometre.Photo: Phiarc, CC BY-SA 4.0, via Wikimedia Commons

Questions people ask

Is 12 MP enough for photography?

For screens, social media and prints up to about A4 at 300 ppi, yes. 12 MP (4000 × 3000) gives 34 × 25 cm at 300 ppi and is sharp at that size from reading distance. It runs short only for heavy cropping or large prints viewed close.

Is a 50 MP phone camera better than a 12 MP one?

Not by itself. Most 50 MP phone sensors bin to 12.5 MP by default, and in that mode the result depends on sensor area, lens and processing. The 50 MP mode helps in bright light when you plan to crop.

Does crop factor reduce image quality?

A smaller sensor collects less total light at the same exposure, so it is noisier when the output size is the same, and it needs a shorter lens for the same view. With good light, an APS-C camera with a 1.5× crop factor produces excellent results; the gap shows mainly in dim scenes.

Do more megapixels mean more noise?

Per pixel, yes, because each smaller pixel collects fewer photons. Viewed at the same final size, the difference largely disappears, since resizing averages neighbouring pixels the way binning does.

Sources

  1. Wikipedia: Nokia 808 PureView (2012 launch, 41 MP 1/1.2-inch sensor, 5 MP oversampled default)
  2. Catrysse, P. B. and Wandell, B. A.: Roadmap for CMOS image sensors, Moore meets Planck and Sommerfeld, Proc. SPIE 5678 (2005)
  3. Chen, T. et al.: How small should pixel size be? Proc. SPIE 3965 (2000)
  4. CineD: Samsung 200MP ISOCELL HP2 announced (pixel size, binning modes) (2023)
  5. Notebookcheck: Sony LYT-901 200MP sensor launched (2025)
  6. RedShark News: Smartphone camera news from MWC Barcelona 2026 (2026)
  7. PsPrint Help Center: Artwork resolution (300 ppi at final size; no upsampling)
  8. Wikipedia: Visual acuity (20/20 equals 1 arcminute)
  9. PetaPixel: Sony ZV-E10 specifications
  10. Photography Stack Exchange: Do megapixels matter with modern sensor technology?
megapixelspixel binningsensor sizediffractionAiry diskprint resolution200MP sensorphoton noise
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of Artblog.

Keep reading in Artblog

From the other blogs