Phone cameras are still marketed on megapixel counts, and the figure has been close to meaningless as a quality indicator for a long time.
What actually determines image quality is a combination of things, none of which fits on a specification sheet.
What megapixels actually control
Resolution, and nothing else. How many pixels are in the final image.
For any normal use — viewing on a screen, printing at ordinary sizes, posting online — the resolution available from even a modest sensor exceeds what is needed by a wide margin.
The useful thing extra resolution buys is cropping headroom, which matters if you frame loosely and crop later.
What it does not buy is better images at the same framing, and beyond a point it actively costs you, because more pixels on the same sensor area means smaller pixels.
Sensor size is the thing that matters
The single most consequential factor and the one least prominently marketed.
A larger sensor collects more light. More light means less noise, better performance in dim conditions, and greater dynamic range.
This is why a camera with a large sensor and modest resolution outperforms a phone with an enormous pixel count in almost any difficult lighting.
Phone manufacturers have been increasing sensor size, which is a genuine improvement, and it is constrained by physics — a bigger sensor needs a bigger lens and more depth, which is why the camera bumps keep growing.
Pixel binning, which explains the huge numbers
The technique that reconciles enormous megapixel figures with usable images.
A sensor with a very high pixel count combines groups of adjacent pixels into one larger effective pixel, producing a lower resolution image with better light gathering.
A camera advertised at a very high figure is frequently producing images at a quarter of that by default, and the full resolution mode is usable only in bright light.
Which means the headline number describes a mode most people never use, while the actual output is a much smaller image.
This is not deceptive exactly, since the sensor really does have those pixels. It is a number chosen because it is large.
The lens and the aperture
Underdiscussed and genuinely important.
A wider aperture admits more light, which helps in the same way a larger sensor does.
Lens quality determines sharpness across the frame, flare behaviour, and distortion. Phone lenses are tiny and manufacturing tolerances matter enormously at that scale.
Optical stabilisation allows longer exposures without blur, which is a substantial advantage in low light and is not present on every camera in a multi-camera array.
Processing is doing most of the work
The thing that has changed most and that most determines how photographs look.
Modern phone photographs are not single exposures. They are composites of several frames, aligned and merged, with substantial computational processing for noise, dynamic range, sharpening and colour.
This is why two phones with similar hardware produce visibly different images, and why software updates can change how a camera performs.
It is also why comparing specifications tells you very little. The processing pipeline is proprietary, undocumented, and responsible for most of the perceived difference.
The consequence is that the only meaningful way to compare cameras is to look at photographs taken with them, in conditions resembling those you shoot in.
The multi-camera question
What the extra lenses actually do.
An ultra-wide and a telephoto are genuinely different focal lengths, offering framing that cannot be reproduced by cropping or by digital zoom.
They are frequently of lower quality than the main camera — smaller sensors, narrower apertures, no stabilisation — which is why images from them look noticeably worse in anything other than good light.
Some additional cameras on cheaper phones exist mainly to increase the number on the box, contributing very little. Depth sensors and low-resolution macro cameras are the usual examples.
Counting cameras is about as informative as counting megapixels.
What I would actually assess
Look at full-resolution sample images, ideally taken by somebody independent, in low light and in high contrast, which are the conditions that separate cameras.
Check whether the telephoto is optical or a crop from the main sensor, which is frequently disguised.
Check whether stabilisation is optical or software only.
Consider whether you want the processing style, which varies substantially between manufacturers — some produce heavily sharpened and saturated images, others aim for neutrality, and this is a preference rather than a quality difference.
And ignore the megapixel figure entirely, which has been decorative for years.
Raw capture, briefly
Worth mentioning for anybody who wants control over the processing.
Most phones now offer a raw or minimally processed capture mode, which produces a file with far more latitude for editing and none of the automatic decisions about sharpening, colour and dynamic range.
The files are much larger and require editing to look good, since the processed image you normally see is doing substantial work.
For anybody who finds a manufacturer's processing style objectionable, this is the answer, and it is more accessible than it used to be.