Does pixel SIZE matter?

[crispy]

Hi everyone!

I work for a retailer of cameras, and I've been selling digital cameras now since thier inception. I like to consider myself fairly knowledgable about them, and I like keep up on the technology as a matter of professional pride.

Now, every once in a while I get a customer who, in the process of researching his first digicam purchase, digs up some obsure tidbit of information that I've never heard before. Usually a quick bit of internet research is all it takes to find the answer, but I've had no luck so far; so here I am!

This customer started going on about pixel size, telling me that bigger was better. I'd never heard this before, and I can't for the life of me figure how it could possibly matter.

Little help?

[BillDrew]

It very well can make a difference. A larger pixel size means it can gather more light - hopefully with the same noise level as a smaller pixel. That would mean a higher signal/noise ratio and thus a higher useable ISO.

In addition, pixels act somewhat like a detector at the bottom of a pipe so light that is not in line with the "pipe" doesn't reach the detector. That can cause vignetting which might be less pronounced with larger pixels.

Of course if you are comparing cameras with different pixel sizes, there are going to be other differences as well that could negate those effects.

[Rodney]

Hi Chris,

I believe the customer is referring to number of pixels rather than the size of a pixel. Anymore, consumers are misguided to think a higher pixel count means better quality images. Most users of digital camersa just want 4x6 prints with a "feeling" they can get 8x10 if they want and they rather not use software to edit their pictures so a good 2mp - 3mp digital will do just fine. The storage media needed for 4, 5 and 6 mp digitals can get rather expensive.

Rodney

[crispy]

Originally Posted by BillDrew

It very well can make a difference. A larger pixel size means it can gather more light - hopefully with the same noise level as a smaller pixel. That would mean a higher signal/noise ratio and thus a higher useable ISO.

This makes some sense to me. (although, I have to admit I find the whole ISO concept as it relates to digital cameras a little gray) However, in my mind a 2MP camera with an f/2.0 lense and a (lets just say, for the sake of argument) 2cm CCD, would be more-or-less equivalent to a 2MP camera with an f/2.0 lense and a 1cm CCD.

Originally Posted by BillDrew

In addition, pixels act somewhat like a detector at the bottom of a pipe so light that is not in line with the "pipe" doesn't reach the detector. That can cause vignetting which might be less pronounced with larger pixels.

Huh? Are you talking about the pixels that reside around the edges of the CCD? I don't follow. (Yes, I know what 'vignetting' means)

Originally Posted by BillDrew

Of course if you are comparing cameras with different pixel sizes, there are going to be other differences as well that could negate those effects.

Well, this is really what it comes down to for me; I mean, if pixel size were really that signifantly different amongst consumer-level digicams, why wouldn't the manufacturers list the CCD dimensions in their cameras specifications?

[crispy]

Originally Posted by Rodney

Hi Chris,

I believe the customer is referring to number of pixels rather than the size of a pixel. Anymore, consumers are misguided to think a higher pixel count means better quality images. Most users of digital camersa just want 4x6 prints with a "feeling" they can get 8x10 if they want and they rather not use software to edit their pictures so a good 2mp - 3mp digital will do just fine. The storage media needed for 4, 5 and 6 mp digitals can get rather expensive.

Rodney

Thanks for your insight, but believe me: the customer really was concentrating on pixel size; I was quite sure to clarify that with him.

My research thus far is leading me to conclude that this was an example of a person who knew just enough about the subject to be dangerous. It's a phenomonon that I've commonly encountered throughout my experience in various technical sales fields

[BillDrew]

Originally Posted by crispy

This makes some sense to me. (although, I have to admit I find the whole ISO concept as it relates to digital cameras a little gray) However, in my mind a 2MP camera with an f/2.0 lense and a (lets just say, for the sake of argument) 2cm CCD, would be more-or-less equivalent to a 2MP camera with an f/2.0 lense and a 1cm CCD....

A 2cm sensor would collect four times as many photons as a 1cm sensor with the same aspect ratio because it would have four times the surface area. Four times the signal in each pixel. Likely nowhere near four times as much noise.

ISO in digital cameras is a bit murky (at least to me), but less noise means a higher ISO. I think of it as a finer grain film being able to be "pushed" futher to get higher ISO. Certainly ISO and noise are tightly related in digicams.

Originally Posted by crispy

Huh? Are you talking about the pixels that reside around the edges of the CCD? I don't follow. (Yes, I know what 'vignetting' means)

As I understand it, if a photon strikes the sensor at an angle much differernt than 90degrees, it doesn't "register". How close to perpendicular it has to be depends on the size of the pixel. Since the light near the edge of the sensor is at a steeper angle than that at the center, there is some loss, or vignetting. That should be less with larger pixels.

Originally Posted by crispy

Well, this is really what it comes down to for me; I mean, if pixel size were really that signifantly different amongst consumer-level digicams, why wouldn't the manufacturers list the CCD dimensions in their cameras specifications?

I suspect three reasons:
1) There aren't really very many different sensors out there. They don't really want to advertise that (e.g.) the guts of a Canon are the same as a Casio.
2) To talk about those issues highlights the problems with digital cameras - never mind that there are exactly analogous problems with chemical cameras. Nothing is sold by talking about the problems with the product. Not much is sold by pointing at problems in your compeditor's products either.
3) The gain in signal/noise and vignetting reduction is fairly small in consumer-level cameras because there isn't much difference in sensor size. To see a big change in sensor size, there is a big change in price.

[crispy]

Originally Posted by BillDrew

A 2cm sensor would collect four times as many photons as a 1cm sensor with the same aspect ratio because it would have four times the surface area. Four times the signal in each pixel. Likely nowhere near four times as much noise.

OK, that sounds logical.

Originally Posted by BillDrew

ISO in digital cameras is a bit murky (at least to me), but less noise means a higher ISO. I think of it as a finer grain film being able to be "pushed" futher to get higher ISO. Certainly ISO and noise are tightly related in digicams.

Perhaps we need to start another thread to get to the bottom of the ISO mystery. I've never understood how a digital camera can have a selection for different ISOs. As you likely know, a higher ISO film requires less light in order to make the exposure, but the properties of the CCD sensor don't really change. All I can figure is that the difference lies in the voodoo that happens when the camera processes the CCD data into an image. As you also know, noise is a problem that becomes more and more pronounces the longer you expose, so a higher 'ISO' should result in less noise, should it not?

Originally Posted by BillDrew

As I understand it, if a photon strikes the sensor at an angle much differernt than 90degrees, it doesn't "register". How close to perpendicular it has to be depends on the size of the pixel. Since the light near the edge of the sensor is at a steeper angle than that at the center, there is some loss, or vignetting. That should be less with larger pixels.

Following this reasoning though, wouldn't a larger sensor require a more optically precise lense? I know that with large-format film cameras the lenses are much more expensive because of the difficulty of designing/manufacturing a lense that can produce a flat image over the larger surface.

Originally Posted by crispy

Well, this is really what it comes down to for me; I mean, if pixel size were really that signifantly different amongst consumer-level digicams, why wouldn't the manufacturers list the CCD dimensions in their cameras specifications?

Originally Posted by BillDrew

I suspect three reasons:
1) There aren't really very many different sensors out there. They don't really want to advertise that (e.g.) the guts of a Canon are the same as a Casio.
2) To talk about those issues highlights the problems with digital cameras - never mind that there are exactly analogous problems with chemical cameras. Nothing is sold by talking about the problems with the product. Not much is sold by pointing at problems in your compeditor's products either.
3) The gain in signal/noise and vignetting reduction is fairly small in consumer-level cameras because there isn't much difference in sensor size. To see a big change in sensor size, there is a big change in price.

Sounds valid to me. Thanks, Bill.

As an aside, something else that occured to me when I was considering the possible advantages of large vs. small sensors was their respective heat (and therefore noise) generation. I know from circuitry that larger chips require higher voltages and thus generate more heat. This is the primary reason behind most (computer) chip manufacturers steady drive to reduce thier core sizes. Would you consider this much of a factor in CCD design?

[Lin Evans]

The customer is right, but really not using the correct terminology. A pixel is the unit of display produced from the output of the photo diode (photosite) on the sensor.

Professional level cameras have larger sensors with larger and deeper photo diodes. With electronic noise being somewhat consistent at a given temperature, signal is relative. The larger and deeper "wells" of the pro-level photodiode collect more signal than their smaller counterparts on consumer cameras. Thus the signal to noise ratio is favorable on the pro cameras.

In practical terms what this means is more information detail and less noise in the image. There are only a few manufacturers of consumer level sensors. The vast majority excepting Fuji are made by Sony and purchased by other manufacturers. For example, all 5 megapixel sensors in consumer cameras are made by Sony. Yes, the Minolta, Olympus, Sony, etc., all have identical sensors. The differences in their respective performance comes from differences in their electronics, lenses and designs.

Over the years Sony has used several small sensors, sometimes crowding more photosites on the same size sensor. This has contributed to adjacent sensor voltage "leakage" which causes the purple/blue fringe called "blooming" and commonly mis-identified as "chromatic aberration" which is usually red/green and caused by lens design.

The actual photo diode size is relatively greater on the lower resolution cameras. Perhaps this is one reason why the Olympus C2100UZ has such an outstanding image for a 2 megapixel camera with an incredible 10x optical zoom lens. It represents the zenith of 2 megapixel iterative design. The same size sensor was used in subsequent 3 and even 4 megapixel cameras - crowding more and more photodiodes onto a relatively small space.

Professional cameras and removable lens dSLR models like the Canon D30/D60 - Nikon D100, Fuji FinePix Pro S1/S2, etc., use much larger sensors and larger photodiodes which allow much lower noise. Lower noise at low amplification (low ISO) allows using higher amplification of signal (high ISO) while keeping noise to a low enough level to permit useful images. Consumer cameras start with much more noise and the more the signal is amplified, the more the noise becomes an issue. So "pixel size" (in the customer's terms) or more properly photoreceptor size, does play a major role in the quality of the image. But because the consumer cameras excepting Fuji all share primarily Sony sensors, there is little relevant difference in photoreceptor (photodiode) size at a given pixel count.

Lin

[crispy]

Originally Posted by Lin Evans

The customer is right, but really not using the correct terminology. . .

Thank you, Lin. This is exactly the explanation I was looking for.

It also explains ISO in digital camera terms. Let me see if I have this right: The photoreceptor's properties (as I suspected) do not change, the ability to increase ISO is really the ability to increase the amount of signal amplification, which is limited by the signal-to-noise ratio of the photosensor; a higher noise photosensor allows for less amplification than a lower noise photosensor.

By Jove, I think I have it.

Thanks again.

[eric s]

Crispy & others,

Actually, the topic of what ISO is has been talked about recently. Here is the thread:

http://www.stevesforums.com/phpBB2/viewtopic.php?t=7078

And thanks to everyone in thread for helping me to understand more about CCDs and photoreceptors. I seem to remember a press release about an interesting idea in CCD design. The company was suggesting that there were better layouts/patterns for the photoreceptors. Specifically, they tried modling the eyes of some animals and found that they got less noise and better overall pictures. Did anyone see this? Is this just the usual marketing spin/garbage or is there something to that?

Eric