Megapixel theory

I've been pondering the enigma of the question "how many pixels does my camera need".

Most writers on the subject base their answer on the size of the print one wants.

But consider this:

When one views a picture one usually stands at what might be described as 'a comfortable distance' from it, and I would suggest that that distance increases with the size of the image.

For example, when viewing a 6x4 print one usually holds it in one's hand at a distance of, maybe, 12 to 18". When one views a wall mural one stands much further away. If one tried to view a 10' mural at 12" one would only see a small part of it, probably in unnecessary detail - the overall picture wouldn't be seen.

So, that being the case, is there an optimal viewing angle for any given image?

For the purposes of my argument, let's say that there is.

Now, if a 3 MP camera can produce an acceptable image at a size of 6x4" when viewed at the optimal angle, with no pixels being discernable, it follows that the same image could be equally well viewed enlarged to 12"x8" if viewed from twice the distance, maintaining the same, optimal viewing angle. The pixels in the image would subjectively be the same size as in the first instance.

Similarly if enlarged to 24" x 16" and again viewed from further away the pixels would still not come into view provided the optimal viewing angle is maintained.

And so on.

So my basic theory is that, unless one wishes to view an image at an unnecessarily close distance, there is an optimum number of megapixels which a camera needs to offer, beyond which extra pixels may be an unnecessary and expensive luxury.

Can anyone tell me what that optimum number of pixels might be, I wonder?

My only input to this discussion would be the ability to seriously crop an image without it becoming grainy. In that case, the more the better.

Cheers

Reasonable response - but then you get into how good your optics are. Most optics are outperformed by the CCD.

The critical limiting factor is to be able to achieve a resolution on the print of about 4 lpmm. This equates very nicely to the resolution you can get if you enlarge a 35mm negative to 16" x 12", which means that smaller prints than that can be very comfortably achieved.

From a practical viewpoint this can be achieved with a 6MP chip. 10MP enhnaces the ability to view large prints from unusually close distances or to crop.

As stated, there is an optimum distance to view a print from depending on its size.

Thanks for that John.

So why are so many people being drawn in to the 'must have more pixels' trap I wonder?

I like to have as many pixels as practicable even though I now rarely use all that I have. I want the possibility left open for large high-res prints or to crop or zoom a lot.

Also I want to be able to view the pictures I take today if/when I'm an old geezer tomorrow with computer and television monitors with resolution high enough to make what we have today look like a joke. Gone will be jagged lines and crude sharpening as people view screens with many times the number of pixels per inch we have now.

I don't know what to vote in the poll because I don't like the 'N' word. I don't need a very good camera at all. But I want one... and pixels coming out my ears.

The opposite is suggested sometimes, and optics will continue to improve.

The K20D captures more detail than the K10D with the same lenses.

same reason people are always drawn into whatever trap it is this week.

The MHz myth in computing lasted a very long time (although it stayed true longer than the MP myth too).

The problem is there are too many things which people want to use for them to understand them all. As a substitute for true understanding, people look for a single overriding performance measure. With computers it was the mighty Mhz clock speed until the AMD Athlon vs Intel P4 race finally demonstrated once and for all that clock speed alone is an invalid measure of performance.

With the digital camera the measure which has been latched onto is Megapixels. For a while back in the late 90s it was true that higher MP=better. Back then 2MP was pretty decent and even compact cameras sported lenses capable of out resolving their sensors with ease. As we here all know, MP are irrelavent unless you have the glass to back them up and the sensor technology to avoid excessive noise (see my response to the "not sharp" thread in this forum for a discussion of noise vs pixel size). For a while it was noticable that the good camera brands held the same view. Pentax, Nikon, Canon, Olympus and friends sat at about 4-6MP for ages in their compact cameras. Alas it seems market forces have now prevailed, the companies must give the consumer what he wants or go out of business, and what the consumer wants is megapixels.

With personal computers it took an undeniable demonstration (the Athlon was more than 50% faster than the P4 at the same clock speed), plus a concerted effort by a number of leading manufacturers to convince the populus that a 3GHz machine was not automatically "better" than a 2GHz machine. It will take the same with digi-cams, but no one seems very interested in trying at the moment.

Joe Bloggs on the street is still being sold cameras by John Smith the minimum wage lacky in a big electronics store, and John Smith doesn't have a clue. All he can go one is the numbers in front of him, and bigger looks better.

That is reasonable for most viewings of most images, but sometimes I like to study small fine details in wide landscapes, for example, without having to look at a series of telephoto shots instead.

Also I expect that the dot-per-inch count of monitors and prints will increase in future, so we we won't necessarily need larger pictures to enjoy the greater number of pixels - which will result in more naturalistic or fantastic images.

There is the "bragging right" that is so important for quite a few persons.

Can you direct me to the test/review that proves that is the case please? I'd love to be convinced. Whenever I zoom in on images shot only on 6 MP cameras I notice softness, aberrations, etc before I see pixels, and that's with OEM lenses too.[/b]

What worries me is that picture libraries will be taken in by the hype as well and start specifying unrealistic pixel dimensions for submissions.

Not at the moment. It has been my impression from viewing full-size examples that the K20D captures more detail than I get with my K10D. The level of sharpness when viewed at 100% is similar, but the images are larger. (Also the noise levels are lower).

Edges of objects appear soft when viewing at 100% with 6, 10 and 14 MP images, but that does not mean there is not more information with the 10 and 14 MP ones.

Of course noise is a factor and tends to be higher with higher MP cameras, but you don't necessarily have to compare images taken at the same ISO setting. Personally, I'm prepared to use lower ISO settings with my K10D than I did with my *ist DS in order to have more pixels (along with the other advantages of the camera).

The K20D apparently though has lower noise than the K10D despite the higher pixel count.

However, I do agree that you don't currently need more than 6 MP when not cropping very much and either resizing to view on a computer screen, or making small prints, or making medium size prints to view at a normal distance. I only like to have more pixels for the reasons I mentioned previously.

I actually prefer the results from my ds2 for low light/available light situations to the results from the k10.
now bring both cameras into the studio, and setup powerful strobes, expose to the right of the histograms (ie don't blow the highlights, but slightly overexpose the shadows) and the k10 wins hands down.

remember that the lcd is displaying a jpeg that already has been processed in camera, and therefore doesn't accurately show what data is in the raw file...

recovering the shadows in aperture gives you shadows with less noise, and more detial in the midtones, than you would get otherwise.

a good raw file on my ds2 actaully looks ruined on the lcd screen!
so exposure affects noise in practice, more than pixel count....
but the ds2 does create cleaner results than the k10 (low light), and since I often sell 20x30 inch prints off my files, there's no complaints.
now in the studio, the k10 really shines you can see the difference in detial at 20x30!.

the end results from both camera are great as long as you remember which camera ya got in your hands and use each to it's strengths.

Sorry to labour this, but isn't this argument akin to the audio argument that sampling at 96 kHz produces better results than sampling at 44.1 kHz, even though the human ear is physically incapable of hearing the difference?

There may be more information with a 14 MP chip, but if you can't see it, what's the point of having it?