Sunny Kalara at “Talk Like a Physicist” wants a 576 MegaPixel camera, and while the post talks the physicist talk, it doesn’t walk the physicist walk (if there is such a thing).
And the gauntlet has been thrown down —
Next time a person tells me that I don’t need a digital camera with more than 6-10 mega pixel resolution, I am going to hit him/her on the head with the sharp corner of my camera
it appears I am risking photogricide by saying, “But that might actually be true. More pixels do not necessarily make a better picture.”
A further claim is:
Apparently, if you converted the resolution of what an eye can perceive in to mega pixel, it turns out that an eye can see at 576 Mega pixel. So, I want my camera to be at least 576MP camera; is that too much to ask?
When I look out, I see in stereo; with full depth – is it too much to expect that my camera does the same?
I want to take pictures for the unknown technology that will be available to me in 20 years, not for the 3×4 print that can be printed now!
Both the 6-10 MP claim and the 576 MP claim are based on a few assumptions, and as any good physicist knows, you have to make sure these assumptions are not violated in your analysis. So let’s put the Pentax down and talk about this.
First of all, there is a difference between a compact digital camera and a digital SLR that has interchangeable lenses, and it’s not clear that everyone who is risking a Canon to the cranium here is talking about all cameras. When context is given, the 6-10 MP admonishment is often in terms of the compact cameras, i.e. the ones you can carry around in your pocket.
The assumption here is that the quality of the picture is proportional to pixel count. Picture quality is not limited by just the resolution of the sensor, though that’s one factor. First of all, size matters (c’mon, you knew it all along). Compact digital cameras do not all have the same size sensors, and they are generally smaller than the sensors on a digital SLR, which are smaller than 35mm film. If you add pixel count but keep the sensor the same size, you must have smaller pixels. For the same exposure, you now have fewer photons collected, so your signal/noise ratio and dynamic range have gotten worse. This has implications to the aperture, too, since you need more light but the diffraction limit means you are blurring the image onto more pixels. And there’s more to this than pixel size, there’s the lens: cheap, small cameras have cheap, small lenses, and this limits the quality of the picture that hits the sensor in the first place. Imperfect focusing becomes more of a problem with smaller pixels, too, since the light is spread out over more pixels. So more pixels isn’t necessarily the answer — a small sensor with many pixels is just going to result in a faithfully reproduced crappy image stored on your memory card in a large file. The one redeeming part of all this is that you tend to naturally crop your picture with the small sensor, and the image from the edge of the lens is the lowest quality.
One more thing that will risk a Nikon to my noggin: 576 MP. The cited article lists this as the equivalent of the human eye, but it also assumes a 120º field of view, and guess what — most pocket digital cameras don’t have anywhere near this field of view! Even a 35mm camera with a 50mm lens on it has a field of view that is about 1/15 of that area — that brings us down below 40 MP — and a compact digital camera with a much smaller sensor has a smaller area still, though you can compensate for this by using a shorter focal length lens to replicate the angle of view, but these generally aren’t as high a quality. You’d have to be using a very wide wide-angle lens to be needing such a large pixel count.
Pingback: How much money do I need » Blog Archive » Don’t Waste Your Money
Pingback: IRA and 401k » Don’t Waste Your Money
Let’s not confuse what is available now with what one wants to see! Also the argument about the sensor size needs a little caveat too; you either need a large sensor or you need a fast bus; you can have a one pixel camera and a fast bus and make in to an effective giga pixel camera. And I don’t like having a narrow field of vision for my camera either; why should I be forced to choose the part of the scene I want to capture because of the limitation of the technology? Casio just came out with 500fps camera last week and a point and shoot camera with 60fps has been available for a while. Multi lenses cameras have been around for a long time and of course higher resolution camera with better light couplers and sharp optics are plentiful. There is no physics to be worked out here; its all engineering. Not the next Christmas but in a couple of years from now, I fully expect to see my wish fulfilled.
These arguments about more pixel means more noise are spurious; all they are saying is that CURRENTLY they do not know how to remove the noise. Do you think it will stay that way in 10 years? I have seen sharp pictures coming out of really noisy old footage. As I said, I don’t take pictures for current offerings; I want to take pictures for future technology. Every 6MP advocate talks about compromises between pixel size, lens size and the potential use. I want them to overcome technological hurdles so we don’t have to compromise. My requirements are modest; I want my camera to capture as much as my eyes do.
Don’t waster your money if you are happy with your JPG image, 2D world and with the field of vision of 40 degrees, but if you want FULL view with the potential to convert your image data in to a JPG-HD and 3D scenes, push the Sonys and Canons of the world to give you a better engineered image capture devices.
Claiming 576 Mpixels for the eye is ridiculous. The eye achieves high resolution in a very limited field (center of the foeva), and does not efficiently form a panoramic image due to eye motion. You can argue that if you want to replace, say, the entire scene visible thru a picture window with a photograph, and ensure that someone can focus on any part of the image with no discernable loss of resolution due to pixelation, you need that many pixels — but the way you do that photographically is with a panoramic camera or scanner taking multiple frames. There are a few applications (like astronomical sky surveys and aereal photogrammetry) where >100 Mpixel focal planes are useful; otherwise even the highest performance imaging systems make do with a few tens of Mpixels.
You have to improve all of the limiting factors to get a better picture, but the pixel count isn’t currently the limiting factor. And I don’t think the comparison to sharp pictures from noisy footage is apt if you are referring to analog images, especially in film. If you have more than one image, you have more information from which to draw to improve the noise. But a single digital photo does not represent the same situation — AFAIK the current methods rely on interpolation, but this relies on good data in adjacent pixels, and you start losing information when you add more noise. And if the information isn’t there, it isn’t there. And no, I don’t think that will change in ten years.
Having said that, you do have the option of taking multiple images, and have more information (similar to that used by some astronomers , the so-called lucky imaging method)