Why doesn’t my camera capture a scene exactly how I see it?

Ways of seeing

Although the human eye is fairly similar in structure to a camera, the retina has absolutely nothing in common with an electronic sensor. Similarly, the brain works in a totally different way to an electronic image processor.

Basically, a camera uses a high-performance sensor to capture images that are pretty much true to reality and then finishes things off with some low-grade image processing. The eye, on the other hand, uses a sensor that’s much more complex but much less accurate, together with image processing that’s incredibly advanced, and which can easily adapt to all kinds of situations.

Cameras use a lot of clever simulation techniques to produce results that are pleasing to the eye, but for a good while yet, you’ll still have to keep adjusting the manual settings to get the best possible results.

A sensor with a blind spot

Let’s start with a simple experiment. This will help you understand that your eye is actually far from perfect and that your brain does some serious image processing to fill in the blanks.


Take a piece of paper and draw two dots on it, about 5 cm apart from one another. Close your right eye, and then look at the right-hand dot with your left eye. Now, start moving the paper towards your face and away again, without taking your eye off the dot. At some point, at around 15 cm from your face, the dot on the left will mysteriously disappear. When this happens you’ve found what’s known as your eye’s ‘blind spot’. This, experts will tell you, corresponds to an area in the eye where the optic nerve connects to the retina.

Note, however, that you can still see the sheet of paper, and what’s even more surprising is that if the paper is lined, (in my test, you can see I used squared paper), you’ll still ‘see’ the lines, as your brain fills in the missing part of the image with flawless accuracy. So, our experiment has shown us that the eye is by no means an accurate sensor and that the brain carries out an incredible kind of image processing which helps us ‘see’ the images our eye wants to see.

What you really see

In order to understand why digital cameras don’t shoot pictures that are identical to the scenes you see, we need to consider how the human eye actually sees things.


The picture on the left is a photo taken by an accurate digital camera. The whole scene is clear, the colours are well rendered and the variations in brightness reproduced in the image are exactly equivalent to those observed at the scene. The picture on the right is what your eye sees before your brain does its magic.

• Only a small area in the middle of the shot is actually seen clearly. This is the part of the scene that’s captured by the fovea, a small section of the retina that’s just 2 mm in diameter. The angle viewed clearly by the eye corresponds to an equivalent focal length of around 350 mm! Because the retina is shaped like a hemisphere but the projected image is almost flat, the edges of the field of vision are completely blurred.
• Over most of the scene, the eye sees colours very poorly. This is because the photoreceptors that detect colour (roughly speaking, indigo blue, green and yellow), known as cones, are concentrated in the macula - an area in the centre of the retina covering just a few millimetres. The small central section of the macula called the fovea contains only cones, and it’s in this small area that colours and detail are best detected. Areas surrounding the fovea contain no cones at all, only rods, responsible for detecting black and white and differences in brightness.
• The eye is much more sensitive to variations in brightness that appear in dimly lit areas of the scene rather than brightly lit areas. In the dark section at the top of the picture, for example, the beam can clearly be distinguished from the ceiling, whereas no difference in brightness is detected on the wall to the right of the picture, even though it is more marked in real terms.
• Our vision in black and white is much lighter and clearer, although much less detailed than colour vision.

Yet in spite of this, we see clear, good-quality pictures over the whole of our field of vision.

Brain Stitcher, the ultimate software application!

Have you ever had a go at making a panoramic picture by joining together several photos? To capture a landscape, for example, you can move your camera round to take a series of photos following on from one another, and then put them back together on a computer with a specialist software application (some digital cameras also have a function that can do this automatically, most often found in compact cameras with CMOS sensors). This technique is sometimes known as ‘stitching’.


This is pretty much what the brain does, instinctively and constantly. The eye is always moving, passing from one direction to another while the brain records the zones of light and darkness and the various colours in each section of image. It then pastes all of these snapshots together to make one complete picture. As a result, the field of vision recorded by the brain is much larger than that of any digital camera, as the brain effectively ‘sees around’ any subject you’re looking at.

Even if somewhat fleeting, this extremely wide viewing angle serves to provide the brain with a wealth of information. It allows the brain to accurately determine the colour of the light and compare it against the colour of the object you’re looking at. A white object under a red light, for example, will still appear white to the human eye, whereas the camera will be tricked into seeing it as red — see our article on white balance to find out more.


Light sensitivity

We’ve already seen that the human eye sees differences in brightness much better in low light rather than brighter conditions. This has two main advantages: first of all, lower sensitivity to bright light stops us being dazzled by the sun; and second, it makes it easier to spot predators hiding in the shadows (don’t forget that we were still hunted prey just a few thousand years ago and our sight has developed accordingly).

A CCD or CMOS sensor is said to be linear. This means that if it receives three times more photons, it translates that as a value three times greater than a standard value.


The result is that the image captured directly by the sensor with no subsequent treatment (left) is dark and undefined compared with the same image seen by the eye (right). The CCD picks up all the clouds, but the hills and the forest just become a dark mass. The opposite is true of the eye, however, as the clouds appear more washed out but the trees can be clearly distinguished from the ground.

To correct this, a simple ‘gamma curve’ can be applied to help the camera simulate the eye’s hypersensitivity in areas of shade. However, this process effectively involves magnifying the darker areas, which in turn increases the digital noise. As a result, where the eye sees a pleasant, normal-looking image, the camera produces a picture that’s much grainier in dark, shadowy areas than in bright parts of the scene.

The limits of technology

Many of the corrections required to produce a ‘natural’ looking photo are made directly in the camera, but technology also has its limits.

A camera, for example, cannot simultaneously capture detail in low-light areas and bright-light areas. For this, the brain uses information from several sources: the rods are much more sensitive to light and the cones pick up fine detail and colour.


The black and white image on the left is not very sharp but very light and clear. This is what would be captured by the rods. The middle image is detailed and in colour, but it lacks definition in darker areas. This is what would be captured by the cones. The image on the left provides information about dark areas of the scene while the middle image provides information on lighter areas and colours. The brain then puts the two pictures together to reproduce all of the various details, in a similar way to making a High Dynamic Range photo (merging two photos with different exposures to create one final image).

So, in order to capture exactly what the human eye sees, a digital camera would have to:

• Detect even the slightest variations in brightness in low-light areas, while neglecting lighter areas.
• Take a colour picture that’s highly detailed and a black and white image at a very high sensitivity setting, then merge the two together.
• Absorb light over 180 degrees, even if it the image captured only represents a tiny section of the total area.
• Have the incredible power and ability of the human brain.

So, it’s pretty unrealistic to expect a camera to behave in exactly the same way as the human eye. In fact, it’s already impressive enough to think that all we have to do is make a few changes to the basic settings — white balance, sensitivity etc. — to get decent-quality photos.