Alley in Pisa, Italy

by Lars Kotthoff

submit your photo


Hall of Fame
View past winners from this year

Please participate in Meta
and help us grow.

Take the 2-minute tour ×
Photography Stack Exchange is a question and answer site for professional, enthusiast and amateur photographers. It's 100% free, no registration required.

From this answer I understand that the reflex mirror of a DSLR is actually not reflecting all the light, but it passes a certain amount on to the AF sensor.

So if the reflex mirror can pass light, why not use the main sensor (which is directly behind the mirror) for focussing?

Note:
In a comment below the linked answer it is noted that the AF sensor requires a lens to focus beams of light on the appropriate location of the sensor as it is smaller than the main sensor (italics is my own assumption). If the main sensor would be used, would this additional lens still be necessary?

share|improve this question
    
When in Live View mode, DSLR's do use the main sensor for focus. It's generally quite a slow process. –  James Snell Mar 6 '13 at 22:24
1  
I'm aware of that, but I refer to the fast phase detection AF, for which normally a separate sensor is used. Maybe I should make this more clear in the title. –  Bart Arondson Mar 6 '13 at 22:25
1  
Curiously, I've just binged it and the Sony A99 has TWO phase detect sensors and one is part of the main sensor array. I didn't see any detail on why but it might give you a starting point. –  James Snell Mar 6 '13 at 22:45
    
Strickly speaking, the Sony A99 lacks the "Reflex" that would make it a DLS*R*, since it is an SLT design. The reason it has two phase detect systems: 1) The independent sensor is faster and can continuously focus while recording video 2) The AF sensor incorporated into the imaging sensor provides coverage for moving targets when they are in between the focus points on the independent focus array. Even so, DP review and most others say the Sony can't keep up with the 5DIII's keeper rate for shooting action/sports, much less the D4 or the 1D X. –  Michael Clark Mar 7 '13 at 4:38

6 Answers 6

up vote 15 down vote accepted

Phase detect autofocus works by measuring the horizontal displacement between brightness patterns projected onto the AF sensor. To measure the displacement, pairs of 1-dimensional arrays of monochrome pixels are used. This is what the AF sensor in the Canon 5D mkIII looks like:

You can see lots of different lines of pixels used by different user selectable AF points. In principle you could use lines of pixels on the main image sensor to do exactly the same job.

This approach has a few advantages:

  • You don't run into any problems if the main image sensor and AF sensor are misaligned, since they're one and the same.

  • You avoid the complexity of the secondary mirrors and the cost of the AF chip itself.

There are drawbacks to using the main sensor.

In a comment below the linked answer it is noted that the AF sensor requires a lens to focus beams of light on the appropriate location of the sensor as it is smaller than the main sensor (italics is my own assumption)

Your assumption is not quite right. It's not to do with having a smaller AF sensor, the AF "lenses" are actually one lens with a wavey 'B' shaped profile. This lens focusses light coming from either side of the lens onto different parts of the AF sensor.

You would still need some sort of lens to do this job when using the main image sensor, and it would have to swing out of the way when taking a photograph along with the reflex mirror, requiring a complicated mechanical arrangement within the camera. This is the main drawback with this approach, though there are other obstacles:

  • The image sensor pixels are behind colour filter arrays which reduce the amount of light that reaches them by up to two thirds. This could potentially reduce performance in low light however it would enable you to perform the phase measurement matching in colour for fewer false results (you're less likely to mistake one piece of detail from the foreground with a piece of detail from the background for example. Also colour can be used to aid tracking).

  • The size, spacing and sensitivity of pixels will be different between the two sensors, so doing both with one sensor means compromises have to be made.

  • The main sensor would have to be on for much longer periods, causing more energy to drain from the batteries. As Stan points out the shutter would also have to be open during AF so closing it prior to making the exposure would introduce a delay.

  • Finally phase detect AF pre-dates digital image sensors so all the of the technology and tooling to perform AF using a separate sensor already exists and is well developed.

However manufacturers have developed a slightly different method for phase detect AF which does use the main sensor. It was developed for mirrorless cameras which don't have the option of a dedicated AF sensor and which have traditionally relied on the slower contrast detect method using the main sensor.

Instead of a pair of AF lenses in the light path to direct light from either side of the lens onto different parts of the AF sensor, pairs of regular microlenses with alternate halves blacked out can be used to get a similar effect (pixels with the left half blanked will mostly receive light from the right side of the lens and vice versa).

This enables a hybrid AF approach using a combination of phase (to get close to the right focus) and contrast detection (to finetune the result).

share|improve this answer
    
very nice answer! –  Michael Nielsen Mar 6 '13 at 17:11
    
Not completely sure but I believe the masking done at the pixel level is troublesome and is why there is precision issues with such on-sensor Phase-Detect. With a standard Phase-Detect sensor, you can measure the phase-difference with a greater delta. This may be why some sensors are sensitive up to a different maximum aperture value. –  Itai Mar 6 '13 at 18:06
    
Great answer. Just to be certain, the B shaped lenses are there to "split" the image in two such that the two phases can be analysed, right? (as described in this answer) –  Bart Arondson Mar 6 '13 at 18:47
    
I think another advantage, especially historically (with modern designs supporting HD video this is less of a factor), is that the few dozen pixels on an AF sensor can be read much faster than the millions on the primary sensor allowing for a faster feedback loop. –  Dan Neely Mar 6 '13 at 18:57
1  
There's also the little matter of the shutter to deal with in the mechanics. Using the image sensor for focus while relying on a mechanical, focal plane shutter for exposure duration means that the shutter needs to be closed/cocked between acquiring focus and beginning the image recording (as most DSLRs do in live view). That means a two-fer on shutter actuations even when you can mask the time taken behind the reflex mirror swing, which means either slowing the shutter down or losing reliability. –  user2719 Mar 6 '13 at 20:34

The main sensor in a DSLR has one function and one function only, and that's to record an image and do so very very good.
Compromising on that by building in additional electronics would degrade the quality and performance, so why do it when there's perfectly good dedicated sensors available?
Also, DSLRs evolved from film SLRs where of course there was no electronic gadgetry built into the film that could be perverted for the purpose.

share|improve this answer

Specialized devices work better and more efficiently at what they do.

Autofocus sensors used in DSLRs are highly tuned for performing autofocus quickly and down to very low light levels.

If you used the main sensor to do autofocus, you would have two options:

  • Implement Contrast-Detect Autofocus which is an enormous power-drain but actually more precise that Phase-Detect. However, Contrast-Detect requires back and forth movement of the focus elements in very fine steps which is not optimal for modern DSLR lenses.
  • Implement Phase-Detect on-sensor which is possible but much less precise than using a dedicated sensor since the imaging part of the sensor constrains what you can do.

Reality is that the mirror only lets through light in some areas and that most of it is fully reflective which gives you a bright view.

Sony has SLT cameras which have a truly semi-reflective mirror (30% / 70%) which is their solution to provide both Live-View and Phase-Detect Autofocus at the same time. This lets them use a dedicate Phase-Detect sensor to make Autofocus speed and avoid back-and-forth lens movements which is terrible for video. The sensor does lose some light reaching it so it has to be more sensitive to produce the same effective ISO which is a disadvantage when it comes to image quality. The light reflected upwards is too dim to make a nice optical viewfinder, so they had to fit in an EVF.

share|improve this answer

Simply put, the same reason you don't drive a semi-truck to work in the morning (unless you're a trucker). The AF sensor can do it's job better when it doesn't have to compete with the image sensor for surface area. The sensors are different in nature and while there are some designs that incorporate AF sensors in to the imaging sensor, it means that areas of the imaging sensor are not capturing the visible light hitting those areas as well as it could.

There are hybrid sensors that can do both the AF and the light sensing for the image, but they tend to do neither job as well as a dedicated sensor for either. It might not be by much, but it is enough that if you are trying to get the best AF and the best image performance you can, separate sensors still do a better job.

share|improve this answer
1  
I don't see why the AF sensor would have to compete with the image sensor for area when exactly the same photosites can be used for both purposes. Phase detect AF pixels on the main sensor (as used by some mirrorless cameras) are capturing visible light when taking a photo and do contribute to the image. The reduced sensitivity of these pixels is not noticeable as the AF pixels number a few thousand compared to tens of millions of regular pixels. –  Matt Grum Mar 6 '13 at 17:37
1  
@MattGrum - I never said it is a major deterioration, just a deterioration. When your goal is to capture the absolute best image possible, not having to make that compromise gives slightly better quality. As I understand it the hybrid pixels also don't work as well as dedicated AF. They are jack of all trades, but it is a trade off. –  AJ Henderson Mar 6 '13 at 18:15
    
You said "areas of the imaging sensor are not capturing the visible light hitting those areas" which is untrue, those pixels do receive light, convert it to charge and contribute to the image. It's only 0.1% of pixels anyway, so I would be amazed if you could see the difference. Yes they are not as good for AF, but you don't have the option of a dedicated AF sensor in a mirrorless camera so it's better than nothing. I'd be more likely to buy a mirrorless that had PDAF on the sensor than one that didn't. –  Matt Grum Mar 7 '13 at 9:33
    
@MattGrum - yeah. I admit that I did not know that the hybrid pixels captured useful information. I assume the color filtration is why they end up with reduced effectiveness or is there other issues with it? Is my understanding correct that a dedicated PD sensor does not capture useful information since I would suspect it lacks the color filtering necessary for anything other than metering? –  AJ Henderson Mar 7 '13 at 14:20

Speed

Speed is probably the main reason why the imaging sensor is not used for focusing in most DSLRs. AF was developed near the end of the film era, so using the "sensor" (the film) for focus was not an option. Most phase detection AF systems were "open loop" built for speed more than accuracy. Only recently have the major camera makers begun designing lens/body systems that utilize "closed loop" communication between the body and lens during phase detection AF. This has allowed phase detection systems to approach, and in some cases equal, the accuracy of contrast detection AF. Although contrast detection AF using the main sensor is improving in speed, because it is a multistage process that requires several "move and measure" cycles it is still slower but typically it is also more accurate.

While there may be rare exceptions, all DSLRs I'm aware of still use mechanical shutters. This means that the main image sensor is covered during focusing and metering. There have been a few mirrorless models that only have a second curtain shutter, but technically speaking they are not DSLRs.

To use the main sensor for phase detection AF would require the shutter to be open in order to focus, the first curtain would then close before opening again to expose the image after which the second curtain would close. Even when shooting 8+ frames per second the most advanced DSLRs focus between each shot (if instructed to by the settings the user has selected). Currently DSLRs reset both curtains at the same time while the mirror cycles and the AF focuses. To use the image sensor to focus, the first curtain would have to remain open until the mirror had dropped back down and the camera achieved focus lock, then the rest of the camera would have to wait for the first curtain to close and the energy absorbed by the sensor during focusing cleared before the first curtain could reopen to begin exposing the picture. This would slow down the overall process when the whole point of phase detection AF is speed. Contrast detection AF used during Live View which does use the main imaging sensor to focus, on the other hand, is generally more accurate but slower.

share|improve this answer

One possible reason that has nothing to do with autofocus is that the image sensor must be cleared of all charge before an image is taken. Using the main image sensor for AF/composition, as is done with all liveview cameras (most of mirrorless), does require that the sensor be turned off and cleared of charge before the image is taken, otherwise you'd have some ghost of the liveview image on the taken image.

This is one of the reasons for shutter lag on liveview only cameras. dSLRs don't depend on liveview.

And no, if you use the main sensor for AF, then you do not need to have the separate lightpath/lens for the separate AF array. Mirrorless cameras are much much simpler than dSLRs because there is no mirrorbox assembly and no need for separate AF and exposure sensor arrays and lightpaths.

In his "Demise of the dSLR" blog post, Ming Thein estimates from his personal experience of disassembly, that mirrorless cameras have 60-70% fewer parts than dSLRs.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.