Author Archives: Wim Van Gestel

“An increase of megapixels is useless and only increases the digital noise in imagery”

“Of course landscape photographs taken with such a high megapixel camera have a higher image quality”

These are two remarks I hear most frequently regarding a high megapixel camera. Both comments I find myself disagreeing with and they annoy me enough to make me write a post about.

A well known trade off in digital photography is the resolution of the sensor and its ISO performance: as the number of pixels on a certain sensor size increases, the amount of light, and thus information, the receptor receives diminishes (this holds for a constant level of technology, of course, because advances such as gapless microlenses can effectively increase the light gathering efficiency of sensors). On the other hand, and quite trivially, having too few pixels is undesirable as well. Who would buy a camera with one pixel, however magnificent its signal to noise ratio is.

This made some people think that there is such a thing as an optimal level of megapixels. Funnily enough, this “optimal level” went historically from 6 to 8 and now 12-14 megapixels. Anything beyond this optimum is a waste of light per pixel, the argument goes. This I do not agree with. The optimal level of megapixels is contingent on the the photographers application in practice. Indeed, for a photojournalist light gathering abilities may outweigh the camera’s pixel count, not in the least because of the low resolution demands of printed media such as newspapers. However, for those photographers making a living by doing studio work, by taking magazine photographs or by selling fine art landscapes, the low ISO noise performance and pixel count are vital features in their photographic equipment.

Additionally, amateur photographers tell me “they don’t print big”, so the high pixel count is wasted on them. In my living room I have large photographs printed on forex plates, that could not have been made with lower resolution cameras, so maybe one does not know what one is missing. However, suppose that it is true that they do not desire to print that large. Even then, a proper picture made with a 21 megapixel camera that is downsized to 12 megapixels in photoshop and properly sharpened afterward, will contain much more fine detail than a picture taken at 12 megapixels. The question regarding the optimal level of megapixels is personal and should concern whether one wants high detailed images or low light performance. Technological improvements may move the efficiency frontier between those two factors outward, but at the edge of those frontiers, a similar trade off can and must be considered.

Note that I say that a properly made high resolution picture contains more detail than a low resolution one. This brings me to my second source of annoyance regarding high resolution sensors, namely that the resulting high image quality is self-evident. I argue that for a fixed sensor size, increased pixel count shows all the errors the photographer makes and the exposes all the flaws in other equipment such as lenses and auto-focus modules. Indeed, looking at a picture taken in 21 megapixels at a 100% view (actual pixel level), equals looking at a picture with much higher magnification than looking at a 12 megapixel photograph at actual pixel level. When a photographers technique is not adequate, say he uses a too large aperture on a landscape, this will show up much more easily. Moreover, even photographers that do use proper technique are bounded by the laws of physics. In photography there is an optical effect called diffraction. It pertains to the loss of contrast and resolving power of lenses as the lens aperture closes down. More practically: the areas that are in focus with aperture f5.6 are much crispier than the areas that are into focus at aperture f16. With a high megapixel camera (given a fixed sensor size), this effect becomes much more pronounces. This means that a lens that peaks at f8 and is usable until f16 on a 12 megapixel sensor could already reach its peak at f5.6 and be usable only until f11 on a 21 megapixel camera. This is especially bothersome considering that depth of field requirements are also different of low resolution sensors. Whereas a scene would potentially require aperture f8 on a lower resolution sensor, it could require f13 on a high resolution one, simply because what is outside of the depth of field by f8 is magnified to the extent that one can no longer label it as within the zone of focus. But since then diffraction kicks in, this might not be the optimal choice after all.

Another point is that high megapixel cameras need high quality lenses and autofocus units as well. It is no use to have pixels capture blur because of optical misalignment, focus errors or wavy focal planes. Buying a high resolution camera exerts considerable pressure on the set of lenses one might already own. Also the autofocus performance seems to take a hit with high resolution sensors. Again, the same autofocus unit would be labeled as good on a lower resolution sensor, whereas with the higher magnification of the high resolution sensor it displays its errors.

Of course, these arguments are moot if one always resizes to 12 megapixels or if one compares pictures in actual print. For photographers that use lower print sizes or for that resize anyway, focus errors, depth of field considerations and lens requirements are no different for high resolution sensors than lower ones. But those photographers are not using their camera’s sensor to its full potential. And I’ve always been told that’s a shame.