Re: 18 megapixels on a 1.6x crop camera - Has Canon gone too far?

In article <bscba5pnbmv88ja03frbek0np49b21ih14@xxxxxxx>, John A. <john@xxxxxxxxxxxxxxx> writes

The image taken by a camera is static. As noted elsewhere in the
thread, the eye is more dynamic.

That makes little difference to resolution and, if the AA filter in the camera is designed properly, none at all.

I used to make thermal imaging cameras which had sensors with only 8 pixels in them. The output picture was assembled by scanning an infrared image past those 8 pixels in a raster manner, from top to bottom. That was considerably more dynamic image acquisition than the eye, yet resolution was measured to be exactly as predicted by conventional methods.

There is nothing special about motion that enhances resolution - in fact, motion degrades resolution as everyone knows only too well with camera shake.

Incidentally, I used to design thermal imaging cameras which would detect no signal at all if the image was static - just like the photoreceptors in the eye.

I'm not sure it's a perfect analogy, but have you ever looked in
through the window in a closed microwave door as it cooks? You're
looking through an array of small holes. If you keep your head still
you can't resolve much detail of what's inside. If you sway or turn
your head a bit side to side while watching your food the holes sweep
through your field of vision and allow you to make out more detail.

It isn't, but its close - unfortunately the analogy misses the vital element that differentiates it from the camera and thus gives the illusion that you are getting more from movement than you do with a static image. When you look through the aperture grille on the front of the microwave window you are viewing a spatially sampled image, just like the digital camera or, indeed, an infinitely short instant on your eye. Each hole on the aperture grille is equivalent to the sensitive area of a pixel on the camera - or a cone on your retina. The grille itself represents the gap between pixels or cones. Just as with cameras, some aperture grilles have small holes and a thick grille while others have bigger holes and a thin grille - just as some pixels have large gaps between them while, more recently, pixel microlenses have progressed to small gaps and close to gapless.

So far, so good, but that's where it goes wrong. If you built a camera like that then it would suffer terribly from aliasing and moiré - just as the original digital cameras indeed did. That occurs because the resolution of each pixel, or hole in the aperture grille, exceeds the ability of the array of all pixels to unambiguously reproduce the information that each pixel detects. The critical difference therefore between your analogy of the microwave and the camera is therefore the AA features of the camera, both in the lens and the AA filter itself. If you were to put something like tracing paper, or a *** of glass smeared with grease or petroleum jelly, between the subject in the microwave and the aperture grille, thus forming an AA filter in the microwave door then you would not see any additional information in the subject no matter how much you move your head.

Other thermal imagers that I have designed have used motion to increase resolution, using a technique known as microscan. Instead of using an AA filter, we move the image on the array of pixels by half a pixel in each axis, vertical and horizontal, in successive frames. This has the effect of doubling the sampling density of the array of pixels whilst still retaining the same size of pixel. Consequently, the ability of the array to unambiguously reproduce information is doubled, enough to reproduce all of the information that each pixel can detect without aliasing. The effective resolution of the thermal imaging camera can be almost doubled using microscan, depending on the other MTF limiting elements in the system, such as the lens. Your microwave analogy is closer to microscan than to a conventional camera. Unfortunately, it isn't applicable to still image cameras.

The resolution of the camera is determined with the AA filter in place. The resolution of the eye is measured with the lens in place and the normal eye motion occurring. The resolution of the thermal imager is measured with microscan operating. In neither case can you increase the resolution of the system by introducing further motion.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
.

Loading