To many Pixels is counter-productive!

This subject has been beat'n almost to death, but there has to be a point where a sensor can have too many pixels

According to this website it actually begins at 6mp in compact cameras. Any more then that and quality goes down.

http://6mpixel.org/en/

"Time went by and at the Photokina in 2004, 6 megapixels for digital cameras were introduced. Experts agree that a photo with 6 megapixels can hold up to an average picture on 35mm film under typical exposure conditions. Unfortunately the quality of pictures has consistently decreased since then.

When there are more pixels than there are photons reaching the image plane this might perhaps be true.

For a very peculiar definition of "quality". It is not defined explicitly, but implicitly understood to be pixel level noise.

Which is a very weird way to look at "quality", since if all photons were individually recorded the image at 1:1 would look like a noisy mess with zero detail (worst possible quality according to that site) while the photo viewed on a screen would have marvelous, fantastic, today unheard of detail, resolution and lack of noise (best possible quality).

Thus something must be amiss in the premises that site uses. Perhaps they know not of what they speak? SIWOTI!

Jesper

Have you read and understood the relevant discussions ?

Why ? The article that you cite was authored around 5.5 years ago (September 2007). Nobody commented. The author(s) make erroneous claims, and provide no images as their evidence.

How about if you, yourself make specific assertions (and provide evidence) as to why such is so.

Its a classic case of an arguement being taken way too far making some incorrect claims.

If the point was left at "some high megapixel compacts offer little IQ advanatge for there larger file sizes" then it would make alot more sense.

There are (at least) two paths this discussion could take:

1. What are the theoretic limits for when you gain exactly nothing (or the gain is negative) by increasing sensel density for a given process technology. How does this "sweetspot" change as technology improves? Is there any fundamental limit?

2. What might my D800 have offered me in practical photography had they limited sensor density to 30MP or 24 MP, everything else equal?

In both cases, outside of the R&D labs of Canon, Nikon, Sony etc, it is hard to really know. We can speculate, apply theory and read papers on the subject. Or we can try to arrange semi-fair practical tests and try to learn from those. My impression is that a few forum members do that quite well, while most base their opinions on intuition. It can be hard to distinguish one from the other, but after a while you get to know the usernames.

-h

I see that you started an interesting thread on this forum a while back regarding these subjects:

http://forums.dpreview.com/forums/thread/2922787

DM...

In the case of the D800 I'd say we can get a pretty good idea from the D600 sensor, really the only disadvanatges of the D800 seem to be slower FPS and larger file sizes.

I think theres a clear difference between larger sensors and compact though, the former has always ensured theres a clear benefit from extra megapixels were as the latter has IMHO often used megapixels as a marketing gimmick.  The fact that compacts aimed at more serious users have not pushed megapixels as far dispite generally having larger sensors really is a bit of a giveaway.

Or it could be that the "serious compact users" have heard rumors that "small pixels have more noise", and Canikon market research shows that a 10MP compact will sell better than a 15 MP apart from any actual difference in image quality.

-h

I don't think you can say that more megapixels means less quality. Not only because sensors evolve, but also because of downsampling.

I think you could question any possible benefit, of course. Particularly in connection with RAM, hard drive, and other equipment needed. Sometimes I miss the small and fast-to-copy files of my 6 MP D40...

--
Check my blog for reviews, tips & tutorials:
Amateur Nikon
Follow me on Twitter:
@amateurnikon

The website do have images for comparison

http://6mpixel.org/en/sensitivity/

Technology have evolved. More megapixel didn't result in degraded image quality.

However, the increase file size takes longer time to process and affect productivity.

Most people don't print the image and just upload them to flickr or other image sharing website.

Therefore, 6 MP is actually enough for normal point and shoot camera.

I remember my old 6 MP - Fujifilm F31d can produce very good images before it went to heaven.

I hope the camera makers lie FujiFilm would focus on increasing the dynamic range rather than the megapixel.

My new computer has 32GB RAM and a few terabyte hard disk. I'm not bothered by a few tens of megabytes per image.

You do realize that goes hand in hand, don't you? Photon counting (for maximum dynamic range) requires small photosites.

Jesper

Perhaps to some degree but such users are also much more likely to check reviews and samples.

I remember my handy 6 Mpixel Panasonic LZ5 fondly. Things were great viewed on my 15 Inch Sony CRT monitor (800x600 pixel-resolution). But when I moved up to the 8 Mpixel FZ30 and 10 Mpixel FZ50 and LX3, my 20 Inch TN-screen TFT-LCD (1600x1200 pixel-resolution) became essential to being able to much better resolve the higher spatial-frequency detal information present in those higher resolution camera images - and my 6 Mpixel LZ5 images showed their limitations.

Now using a 16 Mpixel GH2, my newer 24 Inch H-IPS-screen TFT-LCD (1920x1200 pixel-resolution), and it's higher contrast-ratio and true 8-bit per color-channel is equally important to be able to present my GH2 images - and the 10, 8, and 6 Mpixel images show their limitations.

We never comprehend what is necessary on the "capture end" of things until we obtain and view using higher quality "spectacles" on the "presentation end" of things. It is as simple as that ...

Unless you have a GPU card, your image processing will not be faster with 32GB RAM.

Putting few layers of  high mega pixels pictures together will easily eat up the RAM.

More megapixels doesn't mean more dynamic range. The camera maker need to design the processing engine to transform the extra megapixel into dynamic range.

For example, the Nokia 808 high pixel count 41 MP is to allow the 808 to produce better quality lower-resolution images (3MP, 5MP or 8MP).

If the camera maker didn't transform the additonal pixels into dynamic range, what you get is just a large picture with same dynamic range of a lower megapixel picture.

Than with, say, 16MB? It definitely will. And it will also be faster than with, say, 1GB. Or 2GB.

32GB nets me more than just "a few layers". I can do dozens without GIMP breaking a sweat or busting the 10GB barrier.

So you're saying the megapixels are required for the dynamic range but do not automatically provide it. You do understand that much. Then what is the objection to the megapixels?

And why can't I be allowed to do that in my own post processing instead of it being forced on me in the camera?

It's not up to the camera maker what I do with my 41MP image. It's up to me. If he'd only give it to me!

Jesper

Not "enough", gross overkill. If the entire purpose of your photography is to upload images to photo-sharing sites and see images on your TV, then you don't need more than 3 Mpx; 2 Mpx if the sensor were 16:9.

Regrettably, some of us still make prints.

Not nearly as good as that of the 12 Mpx compacts like the Samsung EX2F and Olympus X-Z2, at any ISO.

--
http://www.flickr.com/photos/draek
http://500px.com/draek

Obviously, it depends on the size of the sensor, i.e. noise.

Diffraction provides a hard limit dependant on aperture, but creating a lens that's diffraction-limited wide open is pretty hard, even a small one like in most compacts. Dunno if there'd be another natural limit but, well, the diffraction issue highlights an important point: often the *real* limit is well below the theoretical one due to the cost (not just monetary) of approaching it much further, so I question the value such a figure would hold.

Tautologically, less resolution, everything else equal. Still, you do have the Nikon D600 to look at, for a more realistic what-if scenario.

The problem of 'practical tests' is that, unless you follow a strict scientific discipline, they can suffer from... well, all the problems that strict scientific discipline was designed to tackle.

For instance, some years ago a forum member went out to determine whether one could tell apart 6 Mpx images from 24 Mpx ones on a 4x6" print, and thus made the obvious practical test: grabbed shots from two different cameras, made prints, and looked at them. He saw he could tell the higher-resolution photos apart with ease, so he mentioned it in the forums and a few other members made the same test, arriving at the same conclusion.

The problem? well, someone else had the idea of conducting the test a bit differently, and not only selected 'similar' images but specifically went out to create ones which were as close to each other as possible, also made prints of each file, but instead of seeing if he himself could tell them apart, made a blind test with his family members. Still not quite as accurate as a double-blind test, but then there's limits to what a lone researcher can do on a budget and well, his results were what he expected and science predicted it happen: results no better than chance, thus no meaningful difference between a very-high-resolution print and an obscenely-high-resolution print, the original posters being victim to the phenomena known as "confirmation bias"; you expect the larger image to look better to you, so it does.

Most of the alleged benefits of full-frame on this forum follow the same pattern, as do the performance of cameras 'alleged' to perform well in low-light (the Fuji F30 and F200 are prime beneficiaries of this, being long since beaten by similar cameras in rigurous tests), as well as the alleged benefits of low pixel counts for high-ISO shooting, the very topic of this thread. There's a reason why fame holds this forum is populated by pseudo-scientists, so take *anything* you read here more technical than a DOF chart with a large grain of salt.

--
http://www.flickr.com/photos/draek
http://500px.com/draek

Kind of. You will notice a slight improvement in details if you scale a 6MP photo to TV resolution. That said, you won't notice any further improvement at higher MP everything else being equal (which it of course never is).

Back in the 1MP camera days we'd stitch photos together to get details resolved, even if the end image was only 640x480 the effort resulted in much better details.

And soon we will have four times as many pixels in our TV's. By the time those who today are teens have grandkids the low resolution 48MP images (which are yet to be made and have scorn heaped upon them for their horrid characteristics while they outperform anything we have today at any ISO) will look like they're stitched from Legos on the wall displays.

Jesper

We're already at a 1.1 micron pixel pitch for the ubiquitous 1/2.3" and 1/3.2" sensors on compact cameras and camera phones. But if you look at the imaging-resources and dpreview studio comparisons for 12MP, 14MP, 16MP, and 18MP cameras that have 1/2.3" sensors, you'll notice no improvement in actual resolvable detail. The reasons for this are two-fold.

First, the smaller the pixel pitch, the sharper the glass has to be to utilize those additional pixels (given the same sensor size). But all practical evidence points to the fact that we were already at the limits of what these lenses could resolve at 12MP.

Then there's the issue of diffraction. With some of these higher MP small sensor cameras, it's already being diffraction limited near the widest angles.

None of this implies that higher MP counts will yield worse image quality, but these small sensor cameras have reached their limit. And the OP is correct about smaller pixels having less dynamic range (at the pixel level). This doesn't imply that the sensor itself will have less dynamic range with more MPs, but the OP is correct that the processing has to become more advanced to maintain high dynamic ranges with smaller pixel. There's no reason to believe that can't happen, but I also see no point to going below the 1 micron pixel pitch, since it won't achieve anything IMO.

Of course, there's a long way before larger sensor cameras hit these limits. u4/3 sensor would have to hit the 120MP mark before they have a 1.4 micron pixel pitch, so they've got a long way to go.