Why Are Larger Pixels on Camera Sensors Better?

[peter ellner]

I understand that larger photosites on a digital camera sensor have benefits such as improved dynamic range and lower noise due to improved light collection...but I can't help but think that maybe having larger photosites isn't better.

Think of it this way, a larger photosite captures more light, yes, but it also captures light from more angles, thus an individual photosite on a large photosite camera sensor will be receiving the light from more places in the scene being photographed than a smaller photosite. Wouldn't having bigger photosites actually lower the resolving power of the sensor then? Do you understand what I'm getting at?

I would think that having very small photosites would be better, as long as they can be efficient, because each photosite will then be more accurate because it will correspond to a more precise part of the scene being photographed. So why are larger photosites better?

Thank you!

 

[Graystar]

It's not like you can go out and buy a camera with the number of megapixels you think is optimal. You can only buy what's available.

The image quality across a range of cameras with the same sensor size is practically the same. It's far more useful to be concern with camera features and ergonomics.

.

 

[BobSC]

I understand that larger photosites on a digital camera sensor have benefits such as improved dynamic range and lower noise due to improved light collection...but I can't help but think that maybe having larger photosites isn't better.

In at least some cases this must be true. Canon decreased the pixel count on their G series camera.

 

[JudyTee]

I would think that having very small photosites would be better, as long as they can be efficient, because each photosite will then be more accurate because it will correspond to a more precise part of the scene being photographed. So why are larger photosites better?

Thank you!

In theory, at least, a larger number of small photosites would give greater definition provided the lens in use could resolve that small. On the other hand small photosites have less sensitivity, less dynamic range and higher noise levels. A larger sensor with a reasonable number of larger photosites will have better all around performance. Make the photosites too large, you lose definition - make them too small and you have insensitivity and high noise. There is no optimum balance.

Cameras with small sensors and high pixel counts generally have poorer overall performance. This is one reason why the DSLR with its larger sensor can tolerate higher pixel counts without performance degradation.

My Nikon D700 has 12 megapixels on a 24 x 36 mm sensor. Each pixel measures about 8 micro-meters on a side - about four times the size of a pixel on the average compact digital camera. I feel this is about optimum. Buy a good quality camera with the largest sensor you can afford and a reasonable number of pixels.

--
Judy

 

[partha_roy]

look, each pixel records a color from the scene and if the pixel is larger it will collect more light..

the color that will be recorded therefore will be more precise

 

[Kris in CT]

I think it has to be taken with a grain of salt... I'm familiar with nikons so pixel size/dxo mark stolen from snapsort...
D40 61.1 um2 dxo raw 56
d90/5000 30.5 dxo raw 72
d700 71.7 um2 dxo raw 80
d7000/d5100 22.9 um2 dxo raw 80

So for whatever that is worth..

 

[AnthonyL]

look, each pixel records a color from the scene and if the pixel is larger it will collect more light..

the color that will be recorded therefore will be more precise

Except that if that had been the boundary between two colours then smaller photosites may have recorded each colour separately whereas the larger photosite ended up with just one or the other.

It's all a matter of making the right compromises isn't it?

 

[Charles2]

Buy a good quality camera with the largest sensor you can afford and a reasonable number of pixels.

In regard to the topic of how pixel size and density affect image quality, this is correct. However, isn't there a further consideration - the viewing?

If you crop your image severely, a bigger sensor can provide pixels for the monitor and for prints. However, a monitor does not use more pixels than it displays. Other than crops, does a full-size sensor provide anything more for the monitor than an APS-C size?

Similarly, a print can only be so many dots per inch. The largest print you want and can afford, in analogy to a sensor, sets an upper bound on the usefulness of sensor size.

These points do not come into play in considering the value of something more than a point-and-shoot with its tiny sensor. But, without going through the arithmetic, I find that an APS-C sensor is enough by these considerations. I don't see what a full-size sensor would provide for my amateur use.

 

[gisle]

Take a look at this page: http://dpanswers.com/content/tech_pixels.php
--
– gisle [ See profile/plan for equipment list ]

 

[NetMage]

You have gotten a lot of seriously wrong or terribly out of date answers.

All things being equal, more MP is better. Even when things aren't so equal, more MP are better.

At a per pixel level, the 50D's pixels aren't really better than the 40D - but, because it has more of them, on an image level, it can have less noise at a given light level, and more resolution, and can make the trade-off between the two.

At the cost of speed and space, having more pixels gives you more resolution, more flexibility, and equal or better dynamic range.

--
-- Please remove the Quote option!
-- Why can't you edit more than once???
-- How about switching to real forum software?

 

[Deleted1929]

It's interesting to see that everyone else is wrong and you're right. Thanks for sharing.

--
StephenG

 

[Steen Bay]

It's a trade-off between resolution and noise. If we double the MP count on a given sensor size, then the linear resolution will increse by 41% (potentially, but less in practice because lenses aren't perfect), and the pixel level shot/photon noise will also increase by 41% (assuming that the efficiency/QE is maintained), because the shot/photon noise = square root of the number of photons captured by the pixel. For example, if the average signal is 100 photons per pixel, then the noise/deviation is 10%. If we halve the pixel size (area), then the average signal will only be 50 photons per pixel, and the noise will increase to 14.1% (of the signal).

 

[sherwoodpete]

It's interesting to see that everyone else is wrong and you're right. Thanks for sharing.

--
StephenG

What is needed is actual evidence, not just hearsay and opinion.

Regards,
Peter

 

[Chris59]

I understand that larger photosites on a digital camera sensor have benefits such as improved dynamic range and lower noise due to improved light collection...but I can't help but think that maybe having larger photosites isn't better.

I can't help but agree with you.

Think of it this way, a larger photosite captures more light, yes, but it also captures light from more angles, thus an individual photosite on a large photosite camera sensor will be receiving the light from more places in the scene being photographed than a smaller photosite. Wouldn't having bigger photosites actually lower the resolving power of the sensor then? Do you understand what I'm getting at?

Correct.

I would think that having very small photosites would be better, as long as they can be efficient, because each photosite will then be more accurate because it will correspond to a more precise part of the scene being photographed. So why are larger photosites better?

They're not. The smaller the photosite (or pixel if you prefer) the more you can fit on a given size sensor, and the more photosites, the better the resolution.

Also, the smaller the photosite the fewer the digital artifacts such as "jaggies" and de-Bayer-ing artifacts, and the weaker the resolution robbing AA filter needs to be.

Thank you!

 

[BobSC]

It's interesting to see that everyone else is wrong and you're right. Thanks for sharing.

--
StephenG

What is needed is actual evidence, not just hearsay and opinion.

Evidence:

Source: http://www.dpreview.com/reviews/canong11/

Quote: Changes compared to G10 -- 10 megapixel 'high sensitivity' sensor, down from 14.7 megapixels

If you go to the "Compared to" pages, you can clearly see that the G10 has more resolution -- About 2500 lines vs 2000 lines.

But if you go to the Higher ISO and look at the sample images, the G11 has a dramatic advantage.

Since these cameras have the same lens and the same size sensor, it is easy to compare the effect of changing the pixel size.

 

[sacundim]

For a constant sensor size and same lens, the best predictor of image quality is how recent the sensor design is; newer sensors are better than older ones. Newer sensors also tend to have more pixels.

A lot of people still mistakenly believe that having more, smaller pixels leads to less noise and dynamic range. You know better, so you may now ignore these people.

 

[sacundim]

In theory, at least, a larger number of small photosites would give greater definition provided the lens in use could resolve that small.

Actually, more pixels give better high-frequency MTF even when the lens is diffraction limited . Diffraction limits the finest detail that can be recorded, but increasing pixel count increases how well detail below this limit is recorded, even when the pixels are smaller than the airy disc.

On the other hand small photosites have less sensitivity, less dynamic range and higher noise levels. A larger sensor with a reasonable number of larger photosites will have better all around performance. Make the photosites too large, you lose definition - make them too small and you have insensitivity and high noise. There is no optimum balance.

You're making it sound like smaller pixels give you better detail at the cost of dynamic range and noise. This is not true. Smaller pixels = more pixels = finer sampling of the same information . Smaller pixels have more noise and less dynamic range, but there are more of them , which at the whole-image level makes up for the disadvantage.

In fact, it exactly makes up for it when looked at mathematically. If at a given output size, one of the sensors is doing worse in terms of noise or dynamic range, then it means that sensor is relatively inefficient compared to the other one. So an intuitive way of putting it is this: a new sensor that has twice as many pixels as an older one won't have more noise or less dynamic range as long as the smaller pixels record at least half as much information as the older, larger pixels.

In even simpler words: smaller pixels are better as long as they're not disproportionately worse than larger pixels .

Cameras with small sensors and high pixel counts generally have poorer overall performance. This is one reason why the DSLR with its larger sensor can tolerate higher pixel counts without performance degradation.

People keep saying this, and I just don't see the evidence.

 

[BobSC]

In fact, it exactly makes up for it when looked at mathematically. If at a given

And the reason Canon reduced the pixel count on the G11 is?

 

[sacundim]

In fact, it exactly makes up for it when looked at mathematically. If at a given

And the reason Canon reduced the pixel count on the G11 is?

I don't know. Do you, really?

 

[glitched]

In fact, it exactly makes up for it when looked at mathematically. If at a given

And the reason Canon reduced the pixel count on the G11 is?

I don't know. Do you, really?

Pixel peeping at 100%, larger files, no real need for 16mp in the first place, complaints from the target audience they sell to.