As some of you may know there are mainly two types of auto-focusing: contrast detect and phase detect. Both of them use a technology that detects for the maximum contrast, the technical approach, however, is slightly different. In this post we shall be looking at a specific type of phase detection auto-focusing; namely on-chip phase detection.
Phase detection auto-focusing
Phase detection is the faster of the two technologies which is why it is employed on all modern digital SLR cameras. For phase detection to work you need some of the light coming through the lens to be redirected to the phase detection AF sensors located at the bottom of the camera. This is possible only when the mirror on the camera is in its down (normal) position. There’s a semi-transparent area on the main mirror that acts as a beam splitter. It splits the light into two rays and projects them as two images on to the phase detection sensors located at the bottom of the camera. The phase detection sensors then compares the two images and makes corrections until maximum contrast is achieved.
Although this system is fast, it is not accurate at all times. Moreover this system will not work when the mirror moves away from its restive position This happens when you are shooting stills in live-view or recording videos.
Contrast detect auto-focusing
To counter this problem all cameras also have the older, more accurate but slower contrast detect auto-focusing system. Contrast detection was actually the dominant auto-focusing technology before the more recent phase detection auto-focusing technology took over. Contrast detection uses an older more reliable ideology, that of adjusting the contrast of the image by suitably adjusting the focusing elements. Usually, contrast among adjacent pixels is at its highest when they are in sharp focus. The only drawback is the speed at which auto-focusing is locked.
Needless to say, phase detection is a much faster technology, however the problem is, when the mirror flips out, it is no longer functional. There’s got to be a better way to focus, especially in live-view and when recording videos. The answer is on-chip phase detection auto-focusing. Canon, when they launched the EOS 70D, a replacement for the much appreciated EOS 60D, brought with it a new type of autofocusing – dual-pixel CMOS auto-focus. This is nothing more than an iteration of on-chip phase detection.
Dual-pixel CMOS auto-focus
To be fair, on-chip phase detection has been around for some time, even before Canon introduced dual-pixel CMOS auto-focusing. However, they were unreliable and slow. With the 70D and its newly designed sensor, Canon has managed to split each of the effective 20.2 million pixels into pairs, thus converting them into individual phase detection sensors. Going by the technology of phase detection, one half of the dual pixels receive light from one side of the lens and the other half receive light from the other half. The two images are then compared and focus is locked.
The best thing about this technology is that it is always on, even when the reflex mirror inside flips out when shooting in live-view mode and when recording videos. When shooting videos, traditional auto-focusing is perilously slow. The slowness of the auto-focusing performance often results in wasted shots and lengthening of the shoot due to missed opportunities. Videographers, in spite the fact that DSLRs provide better lens choices and much better image quality preferred to use larger, more cumbersome, video cameras. The new dual-pixel CMOS auto-focusing technology and others on-chip auto-focusing systems no doubt is the perfect cue for them to get back into the DSLR fold. Although, personally, I have mostly switched to mirrorless cameras which at this point provide excellent hybrid focusing systems.