Official translation of last year's Photo Mobile guide. Cliquez ici pour le français.
Let's face it, smartphones are becoming increasingly boring. The big bi-annual revolutions are a thing of the past. Why would anyone want to upgrade to a new device? Some people like faster processors, others like brighter screens. There are also the eternal promises of all-day battery life, but what stands above it all are the cameras. Being able to take better pictures is always a pleasing argument, but you still have to be able to find your way around a sea of cameras that all or almost all claim to offer superior photos.
| Understanding the pillar of mobile photography : the sensor
Let’s start with a hard wakeup call. Observe the two following pictures:
They were taken at the same place, at the same time by two different Samsung cameras, one at 12MP and the other at 32MP. Can you identify which photo belongs to the higher resolution camera?
If you have chosen the first photo, you have made a mistake. It's the second photo that comes from the 32MP sensor of the Galaxy A70. However, it is much worse, with limited contrast and dynamic range, and unrealistic shadows. So what about the resolution? After all, 32 is better than 12, right?
You've just learned your first lesson: megapixels are not a measure of quality. Now consider this image:
You see a camera with the lens removed, exposing its sensor, which is clearly larger than the cameras on the high-end phone next to it. The latter has a 108MP main sensor (the second one, under the red dot). That's huge! The camera on the left should be up to 1 Gigapixel! Oh wait, it only has 24MP. So, its sensor has much bigger pixels than the phone. And that matters a lot. You see, a camera works by capturing light. Every pixel must capture light. And the more light each pixel captures, the better. Demonstration:
Click to enlarge the image. Clicking again will allow you to download these in full resolution.
The first picture is from the camera, the second from the smartphone. The camera takes great advantage of its larger pixels despite its lower resolution. The field and its surroundings are much better defined, and they would be even more so hadn't I moved during the capture. The phone tries to compensate for the lack of light for each pixel by applying a line sharpening effect, which takes away a lot of detail and makes the vegetation look quite artificial. It's either that or a lot of electronic noise. Another example? But certainly.
It is now undeniable: the high-resolution sensor captures less detail. All because it is too small to give each pixel enough light! Note the washed out colours as well.
| More is less
This explains the importance of the pixel size illustrated by the previous comparison between the two phones. Now there's a new problem. Numbers sell, we all know that. And the concept of sensor size is not well known by consumers, unlike the concept of megapixels. Samsung has launched with its S20 Ultra a gigantic photo sensor. But to catch your attention, they've cut its pixels to be smaller than any Galaxy S released to date, reaching the preferred size for its mid to low-end devices, at a very meagre 0.8µm. And, picking up on what we've just learned about pixels, we know this is a problem. The reason why this choice was made will be discussed later.
These tiny pixels would be useless as soon as the light conditions deteriorate. And even in daylight, they can prove difficult to operate. Have a look and see for yourself:
This time the comparison is between the 108MP mode, the 12MP mode of the 108MP sensor and the telephoto lens of the camera. All photos are cropped to cover the same area as the telephoto lens. We are in the middle of winter and in daylight, in an environment full of detail. The 108MP sensor should give us a very sharp picture, but detail is lacking and the lines are very aggressively sharpened, again, to mask electronic noise. There is still more detail, especially on the parking rules plate, but there is not 9x more detail than with the 12MP mode. Textures pay the price. So, the 108MP mode does not add much value.
Could the solution be to simply revert to 12MP mode to get the benefits of the larger sensor without paying the consequences? Unfortunately, it's not as simple as that, let's look at why!
Imagine a crisp biscuit. What happens when you cut it in half? Crumbs fall out. On the floor, no less. So, they are irretrievable. You could try to put the two parts of the cookie together to make it look like a single biscuit, but the missing crumbs will still be visible. This is what happens with today's high-resolution sensors. Unfortunately, the process of virtually merging the pixels is not lossless and the image quality takes a hit. Does this mean that photos are no longer recoverable? Far from it:
This photo, produced by a 108MP sensor in 12MP mode, still produces a very good level of detail. The morning haze that diffuses the light is faithfully reproduced and the colour range is also very good. There could be more detail but nothing to spit on the sensor.
All this is to say that the high resolutions put forward on recent cameras, be it 32MP, 48MP, 64MP or even 108MP, are worthless. These are 8-16MP sensors at heart, and they only excel when used in their low resolution mode.
| More to see more.
There is an advantage to the large, high-end sensors used in recent cameras, including Samsung's 108MP.
Taken in winter and at night, the level of detail available is excellent. This is because these 108MP sensors are 12MP sensors with huge pixels. In the standard resolution mode, each pixel gets 3x the area (108/9=12, 9=3^2). The previous sensors had 1.4µm pixels and the new ones have 0.8µm pixels in 108MP mode. So in 12MP mode we have... 2.4µm! This is where more is more. The picture would have been better if the sensor had a native resolution of 12MP but never mind.
| Sensor types
After the main sensor, there are the auxiliary sensors. There are 5 types of auxiliary sensors, as follows:
-Telephoto lens (colloquially called "Tele")
-Ultra Wide Angle
The last two types are very popular with manufacturers as they do not require high resolution sensors or too much complexity. It's a cheap option to give the impression of versatility. In reality, these sensors are largely useless. For depth, all the other sensors can support each other to get this information, if not already done through the autofocus system. For macros, the 2 to 5MP sensors are of little use. The main sensor will produce a similar or better result. That's why they don't exist on high-end devices.
The difference between a telephoto lens and a zoom lens is that a telephoto lens has a fixed focal length, whereas a zoom lens has a variable focal length. In practical terms, the telephoto lens is always at, say, x5 while the zoom lens can be freely configured anywhere between 3x and 5x. This is more flexible, but the trade-off is that the zoom lens is likely to have a smaller variable focal length aperture, which we will see in the next section.
Finally, the Ultra Wide Angle lens allows the camera's field of view to be considerably wider. However, this comes at a price. This is because these sensors are more vulnerable to chromatic aberration and distortion.
Ultra-Wide-Angle sensor versus the main sensor of the Note10+. Note the chromatic aberration issues, especially on the left.
Oh, and the best sensors are equipped with OIS, which stands for Optical Image Stabilization. This enables the sensor to compensate for very light movements, greatly improving image quality. It's a feature that I consider indispensable for mobile photography.
The aperture is the size of the opening from which light can enter the lens. Let's keep it simple: the larger the aperture, the more light will enter the lens. As we saw earlier, this is a good thing in most cases. The aperture is counted in an f/x format, where X is the size. Since X is the denominator, a smaller number is preferable. A larger number identifies a smaller aperture. You would rather have ½ of your company’s profits rather than 1/1000000, wouldn’t you?
Some cameras are equipped with variable apertures, which offer several advantages. By controlling the amount of light, we can avoid touching other capture options such as exposure time or sensor sensitivity (ISO). This is good for avoiding overexposure and burnt out pixels during the day, or blur with too long an exposure at night, all while reducing electronic noise. The apertures are too small to create a strong enough depth effect, which is another advantage available on dedicated cameras with variable focal lengths.
But as this is no longer a fashionable component, a larger aperture is preferable. You can always reduce the shutter speed to reduce the amount of light.
| Software processing
Finally, there is the software part of the camera. The result of your picture is quite different from what the sensor has captured. Each manufacturer decides what to prioritize, such as brighter colours, sharper lines, or smoother skin. Samsung also offers its "scene optimizer", which basically chooses filters automatically depending on the captured scene. I am not fond of the result this function provides and keep it disabled to get a more realistic result. It is also best to leave it disabled if you intend to edit the picture later, in a post-processing workload. This should apply to other manufacturers’ “AI enhancement” solutions.
Otherwise, there is a handy mode that makes some techniques easier to perform for everyone, the Night mode. There are functions like bracketing that require time, stability, and compatible software, which is not a good sell for those who simply want to get a good shot the first time. Night mode does this and more automatically to increase the level of light and detail in night shots. The result is not the most faithful to reality either but is at least usable:
It truly was a dark night. The photo taken in night mode, in the centre, is a little too bright but more attention-grabbing. At least the overly bright details are reasonably accurate and not the result of a fantasy attributable to the ubiquitous “Artificial Intelligence”. There's also the fact that the first photo was produced using the Pro mode manually and the last photo is from a standard capture. Night mode has a clear advantage over standard night shots. And it's easy to use, too. Just point, press, and hold for one to five seconds. Simplicity itself.
For users wanting to take their photography to the next level, there's the option to shoot in Pro mode and produce RAW files. These files allow advanced control over the post-processing of the images. Specialized software is preferable, such as Adobe Lightroom or Google Snapseed.
Now you know what features to look for in the best mobile camera and what pitfalls to avoid. In general, higher-end phones from previous generations have better sensors than current-generation mid-range phones, plus other benefits such as a full Pro mode. Check out the detailed specifications for each phone you're considering to ensure you're really getting the best camera you can get.