What to Fight Over After Megapixels?

NewScientist has a quick look at where the digital image crowd is headed now that the megapixel wars are drawing to a close. Looks like an emphasis on low-light performance and color accuracy in addition to fun software tools are the new hotness. "For years, consumers have been sold digital cameras largely on the basis of one number - the megapixels crammed onto its image sensor. But recently an industry bigwig admitted that squeezing in ever more resolution has become meaningless. Akira Watanabe, head of Olympus' SLR planning department, said that 12 megapixels is plenty for most photography purposes and that his company will henceforth be focusing on improving color accuracy and low-light performance."

Comments

Maybe not.

[fyngyrz]

The megapixel wars may be drawing to a close, but they sure aren't doing it at 12 mp. Canon's 50D prvides 15mp in an APS-C sensor size, which is pretty tight, but users are achieving excellent results at that density... it just takes decent lenses, of which there are plenty in the Canon line.

15mp in APS-C format is a square sensel of about 4.6 m.

Canon's 5DmkII, on the other hand, is a full frame sensor, and it sports a whopping 21 mp... and does so by only going to 6.4 m, so there's quite a bit of room left there.

The 50D's got some noise issues, but the 5DmkII is a quiet design and they've clearly got some room to go.

So I think Olympus is actually saying that they can't, or don't want to, compete in the remaining space in the megapixel wars; withdrawal, if you will, rather than an actual end.

Re:Maybe not.

[cabjf]

This isn't much different than the Megahertz wars. Chips were getting faster, but at a cost of ignoring just about everything else (plus other bottlenecks prevented the speed from being effective past a point). Now we have plenty of Megapixels (at least enough to be better than consumer grade film was) but the demand has shifted to actually being capable of taking decent pictures.

Re:Maybe not.

[fyngyrz]

You need to print pretty large or crop pretty aggressively to get a significant benefit from extra pixels.

Sometimes, yes, you do. But there's no problem or waste associated with this, and the extra magnification you get with small sensels and good enough glass to resolve to them results in a perfectly usable and quantifiable benefit; resolution. The term "pixel peeper" is a silly one coined by people who can only imagine using the full result of the camera. There's nothing wrong with using the full result, but there's nothing wrong with using a cropped region, either.

For instance, this image of the Orion nebula of mine, taken with a 50D, is a crop that you can't really get much past; I took it at f/2.8 and 200mm, using ISO 6400 and multiple stacked exposures of one second duration. It is only 416 pixels square -- not large at all -- and a lower resolution sensor than the 15mp one in the 50D would have resulted in an even smaller usable crop.

Getting closer - that is, using a longer lens - is problematic, for several reasons. First, as lenses get longer in the same approximate price range, they get slower, so I'd lose my f/2.8 option pretty quickly, or else end up spending a *lot* more for the lens. Secondly, exposure time is limited, as the stars move, or else again, I end up spending more money on a tracking mount (or time building a barnyard door or other homebrew tracker.)

As it stands, the 15 mp of the 50D is directly useful to me in that it gets me a more detailed, closer, image than I would get with, for instance, the 40D, which is 10 mp. I like that.

Basically, any situation where you can't really get any closer to the thing you want to shoot, and you're not filling the frame with the subject, higher resolution sensors help by giving you more detail; you can either use that detail directly, as I do for the nebula shot I linked above, or you can opt to average regions and reduce the noise if the number of pixels really seems to be too many to you, or the noise level seems to demand such treatment.

As with most photography issues, for every person you can find who uses a camera one way, there's someone else who uses it another. Various kinds of noise, spatial resolution, color depth, speed... these are all trade offs with any given sensor technology, and I honestly think there's plenty of room left for manufacturers to push any one at the expense of the others. Olympus wants to go for low noise, I'm all for it -- there's going to be a lot of people who want that above all - but I'm not giving up my 50D's resolution (or my investment in Canon mount lenses) to get it. Plus, it's always entertaining to see what comes next in any one camera's product line. I don't think Canon, the manufacturer of my camera, is likely to be out of places to go quite yet. I'm hoping for a "60D" model that is still 15mp, but lower noise and/or goes beyond the current pushed ISO 12800 limit. If they pull that off, I'll buy.

I'm not sure if I'd buy to go past 15 mp... I've got some good lenses, and 15 mp is really quite a challenge to use well. Plus diffraction blurring affects higher density sensors ability to achieve per pixel sharpness; 15 mp already strongly compromises (via diffraction effects) shots taken at f/11, going past 15 mp is just going to make deeper DOF shots less able to take advantage of the higher densities.

Not surprising

[Anonymous Coward]

Akira Watanabe, head of Olympus' SLR planning department, said that 12 megapixels is plenty for most photography purposes and that his company will henceforth be focusing on improving color accuracy and low-light performance."

That not surprising. Look at the Amazon reviews for any camera with a huge megapixel count, like the Canon G-10, and you'll see dozens of people complaining that, yes, the megapixels are nice, but the sensor may be noisy or the colours may be off. Too bad the industry didn't give more attention to accuracy earlier. I'd be happy to have a mere 7 megapixels if noise is seriously minimized.

Quoted in history

[hansamurai]

"12 megapixels should be enough for anybody." - Akira Watanabe

16 Megapixels is point of diminishing returns

[Locke2005]

The accuracy of the human eye is such that you can only distinguish ~4000 pixels in a line while still being able to see the whole picture. 4000x4000=16 megapixels for a square image, or 12 megapixels for a 4:3 aspect ratio picture. Having more resolution than that is only useful if you are going to take part of the image and blow it up or otherwise focus on just a part of the image. So yes, once they achieve 12 megapixels CCDs, they should focus on something else, like speed for example. I have several pictures of "the couch where my daughter was a second ago" because my Nikon Coolpix inserts a huge delay between the time I push the button and the time the picture is actually recorded. Color accuracy would be nice too, or perhaps doing something about the graininess the CCDs seem to exhibit in low light conditions.

It's all about the optics again.

[Dr. Spork]

As was the case in the 35mm film days, the cameras that are best are the ones with the good lenses and good auto focus mechanisms. Secondary are good light meters. The pixel density is definitely high enough at 12M. At the start of digital photography, the CCD was definitely the primary bottleneck for picture quality. But those days are definitively over.

Re:HDR? Depth channel? Optical SVG?

[morgan_greywolf]

Optical SVG - the ultimate! Forget pixels. Have cameras sketch accurate SVGs of a scene with the ability to show or print at any resolution.

Good luck with that one. It's a lot harder than it sounds. Try tracing a simple 2-color bitmap in Inkscape sometime and zoom in real close. Now try tracing a full-color, full page photograph in the maximum number of colors possible.

Oh, BTW, hope you got lots of RAM and time to wait....

What else?

[Tarlus]

What to Fight Over After Megapixels?

Simple. Gigapixels.

Re:Low Light

[hankwang]

The performance at low light is ultimately limited by fundamental physics. If you want to take a picture of a scene with brightness L (cd/m2), exposure time t, aperture numebr N, the amount of light reaching the sensor is H = L t /N^2. For example, at 5 cd/m2 (twilight?) and t=0.03 s, N=2.0, we find H = 0.04 lux seconds at the sensor.

That is about 10^16 photons per square meter. Of you cram 10 MP on a 5x5 mm sensor, that is 3000 photons per pixel. Each pixel has a color filter that on the average transmits 25% of the photons, which means 750 photons per pixel. Simple Poisson statistics means that you get a noise that is 1/sqrt(3000) = 4% for these numbers. That is if the sensor has 100% effectivity and no electronic noise.

Cropping is a big win for pixel count

[davidwr]

I can't tell you how many times I've looked at a picture, analog or digital, and wished I could turn a small part into a poster.

Back in the day, that's what they used large-format film for.

Imagine if a movie editor could decide, in post-production, to "zoom 10x closer" in on a subject?

Just because you can't benefit from them when viewing "full frame" doesn't make extra pixels worthless.

Besides, even if you aren't cropping, to emulate the resolution of 35mm at 100 a typical line-pair-per-mm resolution, you'll nominally need 5000 dots per inch, or 33 megapixels. I'd prefer 4x as many to handle worst-case situations. Now, in practice, how many of our photographs wind up as 2' x 3' posters? Not many. If the biggest you will ever enlarge is 1/3 of that, then your resolution can drop to 1/9th. Depending on how demanding you are, you won't need more than 4-16 megapixels to produce a nice 8"x12" print. Consumer- and pro- 16 megapixel cameras are here today.

Re:Compression

[Flaggday]

e.g. when I went from a 4 mega pixel camera to an 8 mega pixel camera my file sizes became 4 times larger.

This is normal. When you double the resolution, you double it in 2 dimensions. (Height and Width) This results in a four-fold increase in data size.

But 4 megapixels to 8 megapixels isn't doubling the image size, it's doubling the number of pixels. So it is reasonable to expect the file size to double, not quadruple.