Paris

by Jon

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.

Commonly I hear of landscape photographer who rely on live view to focus now. The process I use is:

  • Tripod mount the camera
  • Switch lens to manual focus
  • Enter Live View
  • Live View zoom to maximum level(10x?)
  • Manually focus the best I can
  • Capture image, etc.

Can it be proven that this is in fact more accurate then using Phase Detect AF in my standard mode? I know many variables exist such as subject brightness, contrast, body AF performance, lens, and of course how accurate the human controlled adjustments are. But is it possible to more accurately or at least equally focus using live view even if conditions are optimal? Should I always live view focus landscape images if possible?

Related Questions:

share|improve this question
    
I am also unsure if my Canon 6D live view magnification is a true 1:1 display of the image when zoomed in all the way. I think it says 10x? So is it actually 10x the size of 1:1? –  dpollitt Mar 14 '13 at 14:57
    
Are you talking about Phase detection or contrast detection AF? –  Matt Grum Mar 14 '13 at 14:59
    
You're asking if contrast detection AF is as good a manual? –  Håkon K. Olafsen Mar 14 '13 at 15:16
2  
@MattGrum - Sorry, I am talking about AF in standard Phase Detect Viewfinder mode, over MF in Live View. No contrast Detection AF involved. I've updated the question appropriately I believe. –  dpollitt Mar 14 '13 at 15:26
2  
This article really spells out all of the combinations with nice tests - lensrentals.com/blog/2012/07/… Thank you to @HåkonK.Olafsen for pointing that out in chat. –  dpollitt Mar 14 '13 at 16:06

7 Answers 7

up vote 11 down vote accepted

Yes, manual focusing is more accurate than phase detect AF (except for the combination of very recent Canon camera + lens). Over at LensRentals blog Roger performed AF tests back in July / August 2012. For almost any combination of camera and lens, manual focus can (given enough time) be better than phase detect AF. Read the whole blog series if you want to figure out why some Canon cameras with some Canon lenses perform equally good with phase detect AF as manual focus.

I even have graphs to back me up on this one. First a small excerpt of the blog post to explain the graph.

Then we took eight repeated shots using LiveView and manual focus (represented by red squares), spinning the focus ring between each shot to either infinity or absolute close up. The red squares demonstrate Roger Units (how well Roger can manually focus given all the time in the world on a tripod with a perfect test target.)

Finally we did the same thing but let the camera autofocus in LiveView (represented by green triangles), resulting in contrast-detection autofocus. In theory this should be as accurate as Roger is, perhaps more so.

The graph below shows the results of those shots. For those of you who are not familiar with our Imatest graphs, the numbers reflect the sharpness of the image in Line Pairs / Image Height. The sharpness in the center is shown on the X-axis and average sharpness on the Y. Higher is better and, in this test, better focus equals higher sharpness.

(...)

Now let’s throw away the repeated shot results from above since we’ve made that point, and replace them with standard (phase-detection) autofocus shots. These are taken in exactly the same way as the live view AF shots: take the image, spin the focus ring to one extreme, let the camera refocus, save the image.

Graph that shows the inaccuracy of phase detect AF, compared to manual focus and contrast AF

It's quite clear from the graph that phase detect AF is not as good as manual, and that the performance varies a lot more. You might be lucky and get good focus with phase detect AF, but far from always.

share|improve this answer

This depends on the camera. On most digital cameras the Live-View image is generated by sub-sampling the sensor. Most cannot read the entire pixels fast enough to resfresh the Live-View fast enough. This means the 1:1 view which is usually around 8-10X magnification is interpolated.

When the view is interpolated it becomes very difficult to focus there are no sharp pixels even when in perfect focus. On cameras that stop the MF Assist to show the resolution at which the sensor is read, focusing manually is easier.

Now the accuracy of normal Phase-Detect AF can be very good but it can also suffer from front-or-back focusing as well as a focus shift when stopping down the aperture. When this happens you are kind of stuck because DOF Preview produces an image that is rather dim and hard to focus anyways.

With Live-View, on many cameras you can focus with t he aperture stopped down and still get a bright image and can focus at the correct aperture.

share|improve this answer
    
This is the kind of info I was hoping to learn, so thank you very much. I am curious which way current Canon DSLR's operate in Live-View? I've always thought that live view at 10x on my DSLR was kind of hard to see clearly for sharp focus. Am I correct in assuming that if my camera does interpolate the image - you would suggest that MF in live view may NOT be as good as phase detect AF(assuming fairly wide aperture for landscapes)? Sorry for so many follow up questions, just trying to understand so I can put the knowledge to use! –  dpollitt Mar 14 '13 at 15:55
    
10X is way too much. Even if they could read the entire sensor, that would take a 0.2 MP portion of the 20 MP sensor and spread it over a 1 MP LCD which requires interpolation anyways. Unfortunately I have no idea what resolution the read the sensor at but 2 to 5 MP is a safe bet. –  Itai Mar 14 '13 at 16:06
    
@dpollitt - To be clear. Not that is as good, better or worse. It varies. The performance of Phase-Detect does not always nail focus perfectly well but it can and generally does do. However, with an interpolated Live-View, the Phase-Detect AF has better data to work with! –  Itai Mar 14 '13 at 16:21
3  
@dpollitt Canon phase detect is only designed to be accurate to within 1/3 of the (wide open) depth of field with the central point, and 1 depth of field with the outer points. And that's 1 depth of field according to the ancient Zeiss CoC which is far too large for today's sensors. Even taking liveview interpolation into account you ought to be able to manually focus more accurately than the phase detect system. –  Matt Grum Mar 14 '13 at 16:31
    
Live view x10 on Canon 40D is very usable for manual focus. you can really see the hairs and fine textures. –  Michael Nielsen Mar 14 '13 at 23:14

There are many problems with phase detect autofocus that make it inferior to manually focusing using liveview:

  • Misalignment/calibration errors. AF in not performed using the main image sensor, but a separate AF sensor which is supposed to be mounted the same distance behind the lens. This mounting is subject to a tolerance, as are the position of the focusing group in the lens. This is a problem because PDAF is not fully closed loop, i.e. it doesn't continually refine the focus distance. When it believes it's close enough it sends a command to the lens to move and then terminates. Liveview uses the actual image projected onto the actual sensor so lens tolerances have no effect.

  • There is a threshold for what the AF system believes as good enough. As mentioned in the previous point, the AF system sends it's final command when it believes the focus distance is close enough. These thresholds were set in the late 1980s when people didn't print larger than 9"x6" from a 35mm film negative. A Canon 5D mkIII RAW will contain significantly more detail than a 35mm neg, the AF system might be happy with getting "close enough" but viewing the image at 100% or printing a large sizes will reveal errors. A good liveview AF implementation can continue until the focus is dead on, or close enough that no improvement is seen on the image sensor by moving the lens.

  • PDAF performs phase detection by measuring the horizontal offset between brightness patterns detected by two 1-D arrays of pixels. It can easily be confused by repeating patterns which can appear well aligned at different offsets. For subjects with variation principally in one direction (such as stripes) in addition, accuracy suffers, dropping to zero as the angle between the texture and AF sensor approaches zero. Liveview / contrast detect AF and manual focus look at an entire area, not a single line, and is therefore sensitive to detail in any orientation and not as easily confused.

  • PDAF is performed with the lens wide open. This can cause problems with lenses which exhibit focus shift when stopped down. Focussing with liveview can be performed with the lens stopped down to the aperture you plan to use, and therefore also offers a realistic depiction of the depth of field you'll get in your image.


Old answer that applied to contrast detect autofocus:

A human can be expected to perform better than the cameras contrast detect AF system in a few scenarios, for example when the subject is moving slightly (such as trees swaying in the breeze) as a person is able to recognise the content of the image and predict its behaviour better than a computer.

A human can, during focusing, re-evaluate decisions about the correct object to focus on (e.g. starting to focus on one area, then seeing another detail start to come into focus which is more interesting). A human can also dynamically vary the size of the region to concentrate on based on content, e.g. focussing on an individual branch, whereas most CDAF algorithms will only consider a fixed rectangular region of interest.

Additionally a human can be expected to provide finer adjustments to the lens position using a mechanically linked focus ring than is possible for the camera using the lens's AF motor.

Outside of those situations, given an appropriately flat, detailed area to focus on, a contrast detect AF algorithm can be expected to perform as well or better than a human, as it is easier for a computer to measure contrast than a person.

share|improve this answer
    
Sorry, no I'm not talking about any contrast detect AF here. Phase Detect vs manual in live view. Example of a Canon EOS DSLR. –  dpollitt Mar 14 '13 at 15:29

Live view focusing takes place on the sensor that will be used to take the final image. If it's in focus there the final image will be in focus. Additionally, live view allows you to magnify portions of the image for finer focus, and to see the DOF effects of aperture settings.

AF on the other hand depends on a separate AF sensor within the camera, and a different image path. If there are any misalignments in the AF path then the image will be out of focus on the main imaging sensor.

The D800 had reports of focus problems when it was first introduced. Live view focussing was fine, but AF was misaligned. The affected cameras had to be re-aligned by Nikon to correct the error.

share|improve this answer
    
So your argument is that manual focus in live view can be more accurate because of potential for the AF system to be misaligned. So you are saying that you would recommend sticking to manual focusing in live view. What if I have calibrated my AF system and I know it is working excellently? –  dpollitt Mar 14 '13 at 15:50
    
Live view also gives you the advantage of knowing -precisely- what you're focussing on. AF gives you a little box. In a confused or ambiguous image (e.g. flower in a field of tall grass) live view lets you nail the focus in an unambiguous way. –  BobT Mar 14 '13 at 17:01
2  
Accuracy vs precision. You can have a perfectly accurate AF but still the precision has a distribution with a deviation, which means most of those shots will be outside that perfect center. The alignment trouble moves the center of that distribution. –  Michael Nielsen Mar 14 '13 at 23:16

I use Canon DSLR with manual lenses and I have the efs focusing screen and sometimes I do use live view. So I don't know about others that might handle live view differently, but I do know Canon. Indeed, Canon x5 and x10 views make it tack sharp for accurate focusing.

Actually, in live view it is sharper than the preview of the taken image itself, when zooming in on it on the LCD screen. Sometimes I get confused that after I take the image it seems not sharp in the preview while the live view was tack sharp. Then I check on the computer and it really IS tack sharp.

And I do find manual focus more reliable than any AF in any of my AF lenses. Certainly, the AF can be faster than myself, if it gets lucky. This is why I prefer AF with FTM.

share|improve this answer

I'm using great manual lenses (zeiss ze) + focus screen on a full frame DSLR (5DMKII) and yes, I definitely obtain much better results when I focus manually with live view.

Moreover I've been using manual focusing lenses for quite a long time and I've always got the best results when I had the opportunity to use live view. No contest.

share|improve this answer

Getting straight to the answer...... yes indeed using liveview to focus is more accurate but i will have to disagree it is that important for landscape.

It just takes a little more seconds to shoot compared to auto focus. setting up a decent aperture will make sure your picture is sharp enough when shooting landscapes.

Where i use the liveview for focusing is when shooting moon (extreme tele shots) or when shooting smallest subjects (macro shots).Where i would require very accurate focusing as I have to open the aperture to the largest to compensate for my shutter speed, and acute focusing mistakes are a big blunder.

:) that's my opinion and how I shoot. If you can sort out my mistake if any i would be happy to know :)

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.