This arrived in my inbox, and having read this web site for about a month, I think people really are fooled, and this might potentially help people understand why and how not to be one of them.

---------------------------------------------------------------------------

The Megapixel Myth - You’re getting ripped off.

Megapixels don’t matter. There, I’ve said it. Throw your stones, rotten vegetables, or what-have-you.
David Pogue has an article in the New York Times today about the Megapixel Myth. His heart is in the right place, and he comes to the correct conclusion, but he’s fairly sparse on the details and the proof is a little hard to grok without actually being there.
News.com ran a story this week quoting Chris MacAskill, my father and SmugMug’s President. He comments on how the word megapixel is a marketers’ dream, and he’s right. Anyone marketing a camera based solely on megapixels is ripping you off. But their article wasn’t clear enough either. So here it is, as clear as I can make it:
What you *really* want are “better” pixels, not more of them. Contrary to popular belief, adding more pixels to the same size sensor isn’t going to help you very much. In fact, it may hurt. Why? It’s reallly quite simple: A limited amount of light gets through the lens and hits the sensor in your camera. It’s really quite small. At some point, if you cram more pixels into the same tiny space, those pixels aren’t picking up enough light to be useful. Instead, they’re making your photo noisier and reducing the quality.
The real way to get better images is to have a bigger sensor (so it can capture more light) and a better lens (so more light gets to the sensor). It’s that simple.
Don’t believe me? NASA’s Spirit Rover has a 1 megapixel camera and it takes better photos than any 8 megapixel camera you can buy at Best Buy. MSNBC has a great article on how it works and sample images. See for yourself.
Finally, SmugMug has printed more than 3 million photos for very discerning customers. We publish the reasons why photos get returned. The number of returns for “not enough megapixels” is at or near zero.
So how do you figure out what camera to buy since megapixels don’t matter? Pogue is right on the money with this one: read reviews at sites like dpreview.com.

Wow! very interesting read. We all know this feeling when you buy a top product on the market, and next month - in the same shop - you can spot the next model which has more MP and is even cheaper. This article makes the pain easier Now I won't fear for the next years of my camera, thanks mate!

kriso

Hmmm. You may have saved me $1,000. Thanks!

Well you know I charge a 10 % savings fee. Just kidding.

I'm always saying that to firends and family who want to buy a digicam. The size of the lense is important, bigger means more light comming into the sensor.

I personally have a 3MP Fuhifilm camera and I'm perfectly satesfied.

Higher resolution means:

buying larger and more expensive memory cards.
slow transfere rate to your PC
large image files that might not be editable in Photoshop due to memory restrictions.

I read an important article about camera sensors that basically said :
Current camera sensors take pictures in BW and the camera circuitry conevrts it to colour which causes innacuracies in colour reproduction.

Interesting read

Saying that the number of samples ( ie pixels, the measuring devices on the chip behind the lens ) is completely meaningless is as silly as saying it's all important. Like almost everything else in life, the truth is somewhere in the middle, and depends more on the individual than anything else.

A couple of glaring misunderstandings: the size of the lens in general has nothing to do with how much light it transmits. This is what aperture is all about. What's important is the ratio between how wide the frontal element is compared to how long the lens is. An f/2.8 lens is an f/2.8 lens whether it's 6 mm or 60 mm - both transmit exactly the same amount of light to the sensor. What's important in digital cameras is how big each pixel is. The pixels on the chip are wells that capture photons - light - being projected by the lens. Smaller pixels collect less light, and have to amplify the signal, which introduces noise; also the charge can leak out to nearby wells. At the same size and with the same technology, more resolution means smaller, lower quality pixels. But the solution is to make the sensors bigger, not to have less pixels.

Also, taking a 10 million pixel input and resampling it to 6 million pixels is very different from comparing a native 10 mpx image to a native 6 mpx one.

It's true that most digital cameras "see" in monochrome, or they see luminosity but not color. Each pixel has a color filter over it, so instead of b/w, it can see one color. There are twice as many greens as reds or blues, arranged as RGBG. A green pixel on the chip translates into a pixel in your jpeg file with the green channel as measured, but the red channel as the average of the four red neighbors, and blue as the average of the four blue neighbors. This is very much like how the human eye works - we can see differences in brightness much better than in color, and we can see green better than anything else. Also, for the problems it does create, more pixels help, as long as you're using quality optics.

It's an interesting read, but it oversimplifies things to the point that what it says is only half true.

FC, I was awaiting Your post here

thanks

Uh oh ... hopefully I'm not something like the blind leading the blind. I'm sort of against over simplifications in general. The article is right that megapixels aren't a very good indicator of quality, but that goes both ways - few of them don't mean low quality, but lots of them don't mean low quality, either.

Here's an interesting statistic. A lot of people post photos online at 640x480; this is 0.3 megapixels. The photos on my site, at least the big ones, are 800x533, which is 0.4 megapixels. Whether it's a photo from my old camera at 6 mpx or my new one at 13 megapixesl, I'm still going to shrink the heck out of it. For poster sized prints, 13 m does much better than 6 m. Pixels are about size, not quality, and most people rarely go over 8x10 inches.

I just think it's a better approach to check things out and make up your own mind based on your own situation, than to tell people they're getting ripped off. And I sort of feel obligated to post a photo along with my ideas on photography ... they're only as good as the results.

Thank you for the detailed and technical explanation ForrestCroce. Your macro picture is great because of the sharpness and clarity of the insects and flower and the shallow dof that has thrown the background out of focus.

I still think the size of the lense is important. A larger lense will logically have more surface area to catch more incomming light. It will also be more prone to dust and scratches.

Another issue that I find more important than resolution is camera sensitivity (ISO) and speed. Higher shutter speeds is useful in freezing fast moving objects. Low shutter speeds is useful in shooting low light situations where a flash cannot be used.

Great read. Opened up my eyes.
Had made up my mind about buying a camera yesterday and was going to buy it on monday. Think i will do some more research now.

Originally Posted by Mohab

I still think the size of the lense is important. A larger lense will logically have more surface area to catch more incomming light. It will also be more prone to dust and scratches.

It's a little more complicated than that, but not much, and you're on the right track. A lens gathers light being reflected from your subject, and focuses it onto a film or a chip with pixels. How far the "light path" is determines the "angle of view" - whether you're zoomed way out and get a lot of foreground, or zoomed way in, and make distant subjects look close. But a longer path means less of the light gathered makes it to the sensor. It's the relative size of the lens ( how wide the iris is divided by how long the lens is ), not the absolute size. Example: a 24 mm lens ( popular in landscape ) with a 3 mm iris ( f/8 ) transmits more light to the chip than a 300 mm lens ( popular in sports ) with a 27 mm iris ( f/11 ).

Originally Posted by Mohab

Another issue that I find more important than resolution is camera sensitivity (ISO) and speed. Higher shutter speeds is useful in freezing fast moving objects. Low shutter speeds is useful in shooting low light situations where a flash cannot be used.

This is where we prove you're right about wanting a lens that gathers much light. You need a certain amount of the stuff to make a proper exposure, and there are three ways to achieve that: using a higher ISO to make your camera more sensitive, using a slower shutter speed which can mean a blurry picture, or opening the lens up so it transmits more light.

The good news is they're getting better in the ISO department. This one, below, was ISO 3200, because I was using a 300 mm lens to get close, it only opens to f/4, and it was dark and cloudy, so there wasn't much light available:

Another important facor I forgot to mention is noise which usually increases with higher ISO and low light situations. The more advanced sensors produce less noise.

For readers who are not into digital photography, noise is random dots of random colours that appear in your image

I can not see any noise on the dog picture... My camera at iso3200 makes pretty much of noise. Do You use some filters, or is this less because of the big MPix thing too?

The number of pixels doesn't really have much to do with noise ... it's their size. Just like how a lens with a wider light and a shorter light path is able to gather more light, bigger pixels are able to absorb more of it. ( Light is a wave, but it's also a particle - photons. ) After the pixels catch a bunch of light during your exposure, the shutter closes, the photons are converted into an electrical current ( which is amplified as you turn the ISO up ) and then run through an analogue to digital converter ( ADC ).

Noise is when you get little specs all over your photo, because the signal coming out of the pixels is being amplified, and noise in the signal is also amplified. The pixels read out the wrong values, but randomly wrong ones, so it breaks up a smooth background, or a blue sky. We're especially good at seeing noise in blue.

Another problem is that as you turn the ISO up, you loose out on dynamic/tonal range. You can't see the deep shadows, because there's no signal there, only noise, and you the highlights blow out, or turn pure white.

I should write an article about all of this and see if any photography sites or blogs would publish it.

So part of why this doesn't have noise is that the exposure was right, so I didn't have to bring up ugly shadows, and also because the chip in my camera is 36x24 mm and the pixel resolution is 4368x2912 ( 13 mpx ). Each pixel is around 9 um wide. Here's another one ( Mary ) at 1600 ASA.

I thought everyone knew about this? Good topic for those that are new to this, shocked the heck outta me when I first found out as well.

As i've said before, generally speaking there is no reason to ever go over a 3.0mp camera.

I'm not new to this but I learned much from this thread. In general I think it is not exactly correct to set the border on 3 mp camreras - I would not advise a friend to buy a 3mp camera
New cameras are more technological advanced and provide new functions, better chips, power efficiency, etc. Megapixels are just following their evolution (I suspect at least for the good brand)

Originally Posted by auron

As i've said before, generally speaking there is no reason to ever go over a 3.0mp camera.

EXACTLY!! Just like no one will ever need more than 640 K of RAM.