Front/Back Focus - when is it important?

[xpatUSA]

I'm starting to do more IR when my used Sigma SD10 shows up from Germany.

IR is a bit more fussy about focusing and I'm after sharp images. I've never worried much about Front/Back Focus but today (in preparation) I tried a quick slanted ruler test:



Sigma SD9 with 18-50mm zoom, f/2.8 about 35cm from ruler which is at about 30 degrees although I realize the de facto standard is 45 deg. I focused on the 150mm mark best I could. Looks like it focused on 160mm in this particular shot (I find that the actual point of focus with that lens varies a lot with the slightest touch on the focusing ring).

Point being that, the focus was en avance by about 5mm in 350mm. So. if my intent had been to shoot the whole 300mm ruler, the shot would have been a success, most likely. If, however, I was only concerned about the 150mm mark itself, the shot was a failure.

In other words, the importance of front/back focus could depend on the intent of the shot, eh?

My question for this thread is: "For Real World shots, when is Front/Back Focus important to you?"

[george013]

I'm starting to do more IR when my used Sigma SD10 shows up from Germany.

IR is a bit more fussy about focusing and I'm after sharp images. I've never worried much about Front/Back Focus but today I did the slanted ruler test:



Sigma SD9 with 18-50mm zoom, f/2.8 about 35cm from ruler which is at about 30 degrees although I realize the de facto standard is 45 deg. I focused on the 150mm mark best I could. Looks like it focused on 160mm in this go (I find that the actual point of focus varies a lot with the slightest touch on the focusing ring).

Point being that, the focus was en avance by about 5mm in 350mm. If my intent had been to shoot the whole 300mm ruler, the shot would have been a success most likely. If, however, I was only concerned about the 150mm mark itself, the shot was a failure.

For Real World shots, when is Front/Back Focus important to you?

I don't think this setup wil help you. Turn the measuring rot 90 degrees, take care the corresponding marks are parallel to the sensor and aim to the middle.

George

[Stagecoach]

I focused on the 150mm mark best I could. Looks like it focused on 160mm in this go

Ted,

This suggests you are talking about 'manual focusing' not 'AF' (and supported by the below comment of yours) so if the results are consistent with repeated tests it would suggest the viewfinder is indicating different than the sensor is seeing?

(I find that the actual point of focus varies a lot with the slightest touch on the focusing ring)[/I].

Is this simply the often talked about problem with modern lenses designed for AF that have a shorter rotation than the older manual lenses?

For Real World shots, when is Front/Back Focus important to you?

I have only ever tested one lens, just out of curiosity and taking repeatability into account found it showed nothing untoward.

[Shadowman]

Sometimes what we perceive to be front/back focusing can indeed ruin an otherwise great shot. The problem I had was my focus indicator was dead center of the composition, yet the image was incredibly sharp at the edge of the frame and only modestly sharp on my intended subject. There were a lot of other variables at play, center subject was moving, subject at peripheral was not moving, I was moving (camera shake). I don't think my problem was due to front.back focusing but as you inquired; it could be an issue if tack sharpness is your goal and you believe you have a front/back focus issue. The first step is ruling out all other possibilities before you start worrying about front/back focusing.

[xpatUSA]

Ted,

This suggests you are talking about 'manual focusing' not 'AF' (and supported by the below comment of yours) so if the results are consistent with repeated tests it would suggest the viewfinder is indicating different than the sensor is seeing?

Hi Grahame,

Definitely not consistent - I was just trying it out - took maybe 5 shots manually - 3 focus bracketed by hand. The shot I'm showing is statistically insignificant, not even sure which one it was but it illustrates what I did quite well.

Is this simply the often talked about problem with modern lenses designed for AF that have a shorter rotation than the older manual lenses?

Absolutely. "Manual everything" dinosaurs like myself are no longer catered for. That lens is about a quarter-turn total range (sux). I focused on this occasion by slowly turning until the "focused" beep went off then back and forth until the beep kept a-beepin. So it was sorta AF but with me as the motor

Hoping this thread will not turn into me having to justify my test or apparently being in need of education on front/back focus and how to avoid it, etc, etc. Just sayin' . . .

[IzzieK]

........Hoping this thread will not turn into me having to justify my test or apparently being in need of education on front/back focus and how to avoid it, etc, etc. Just sayin' . . .

No...it is OK...I am interested to know the answer to your question too, so this will be a good topic...

[DanK]

Is this simply the often talked about problem with modern lenses designed for AF that have a shorter rotation than the older manual lenses?
Absolutely. "Manual everything" dinosaurs like myself are no longer catered for. That lens is about a quarter-turn total range (sux). I focused on this occasion by slowly turning until the "focused" beep went off then back and forth until the beep kept a-beepin. So it was sorta AF but with me as the motor

Hoping this thread will not turn into me having to justify my test or apparently being in need of education on front/back focus and how to avoid it, etc, etc. Just sayin' . . .

Not at all. It's just not what most people use the terms front- and back-focusing to mean. They are usually used to mean that AF is systematically off in one direction or the other.

Like Grahame, I have never found a problem. I tested all of my lenses with my old 50D, and only one showed any noticeable bias--and the bias was in different directions at the two ends of the zoom. Moreover, it was too small to matter. I have actually not gotten around to testing any of my lenses with my newer body in the year or so I have had it. I suppose I ought to.

[NorthernFocus]

If you're going to shoot with really narrow DOF, this can be an issue. In my experience it is also worse at longer distances. The focus error seems to compound more than DOF increases at longer range. If you manually focus but use the focus indicator to tell you when focus is achieved, it's the same net effect as using AF.

Like so many of these topics, this is one that tends to invite all of the arguments about "real world" vs theory, etc. One thing you'll learn if you start delving into this is just how much variability there is with AF. We all like to think it is perfect. But if you set up and do some bench testing you will quickly realize there is a good bit of variability from one shot to the next. So if you have a lens/camera combo that has front/back focus issues, it can manifest itself by unpredictable results. Some shots in focus, others not due to the inherent variability in the AF function. With a front/back focusing system you are operating closer to the error band so more shots end up unacceptable. Tuning to center simply reduces the cull ratio by keeping more shots within the error band.

Many well experienced people are quite oblivious to focus issues and get by shooting in burst mode. They just accept a certain number of OOF shots as par for the course and get around it with numbers. Nothing wrong with that. But ironically those are some of the same folks who will argue that the phenomenon doesn't exist and argue that every lens/camera they've ever owned has been perfect. Which may be true if one defines perfect as falling within an individual's tolerance band. That's where the issues crop up. Based on how/what we shoot we all have different tolerance bands and therefore different definitions of "perfect"

[DanK]

To add to what (the other) Dan said: I tested several lenses with one body using FoCal. It focuses repeatedly, using the identical settings and a fixed target. It was sobering to see the variation in results. In fact, in at least on case, it gave me a warning saying that I should have limited confidence in its estimate because of the magnitude of the variance.

I'd put this in slightly different terms than Dan did. He wrote:

So if you have a lens/camera combo that has front/back focus issues, it can manifest itself by unpredictable results.

I agree, but I would add: Even if you have a combo that does not have front/back focus issues, it can show unpredictable results.

Statistically, I think of back/front focus as bias, which is independent of error (variance). You could have a combo that shows a lot of variation but that gives you a long-run average that is spot on. That would be a combo that is free of bias. Conversely, you could have a combo that shows tiny variation shot-to-shot but that nonetheless consistently front- or back focuses.

When I have concluded that I didn't have a problem with most of the combos I have tested, I was referring to bias--with one exception, I found no evidence that a combo was consistently erring on one direction or another by a substantial amount, even though at least one combination showed a disturbingly large variance.

[NorthernFocus]

...Statistically, I think of back/front focus as bias, which is independent of error (variance)...

Exactly so. With zero bias(i.e. NO front/back focus) every shot appears to be in focus as long as the variability is less than the DOF. As the bias shifts the center of the variability curve one way or the other, shots begin to drift out of one side of the DOF and you end up with a percentage of OOF images. The lower the variability, the more bias can be tolerated and vice versa.

That's what I referred to as "unprecictable results". No doubt over time they are statistically repeatable, but at any given moment you don't know where you are within the variability/error band. Which is infuriating

I got by for years with the burst shooting philosophy and was probably the happier for not worrying about my gear. Then I had to one issue where only about 20 percent of my shots were in focus and I had to figure out what was going on. I noticed that on the OOF shots the focus was consistently in front of the subject and on the in-focus shots the majority of the scene within the DOF was in front of the subject. That led to focus tuning targets, software, etc. And once the genie is out of the bottle... well you know how it can be

[Inkanyezi]

I don't know that particular camera, I hope it has live view. I did a deal of IR shooting back in the days of film. It was a little guesswork, although there was a mark on the lens where to move after focusing, when there was IR film in the camera. The raindance was focusing without filter first, move the focus ring a tiny bit, the distance between focus line and red R mark, put filter on and shoot. Focal length increases for IR, but when using live view, that should not be a problem, as you would see the IR image from the sensor and can have the filter on. With a little bit of luck, AF might work.

I almost always focus manually when taking still objects, particularly with mostly flat objects that I want with much depth. I guess tilt lenses might come with AF for triple points in the future, but as it is, manual is the way. It cannot be done in a DSLR viewfinder, but works well with mirror-free systems. The image is taken with a Chinon 35 mm lens on µ4/3.

[dje]

Ted to answer your specific question, I think a high level of AF accuracy is important for shots with shallow DOF which you need to examine closely ! (Similar answer to Dan).

The front/back focus issue can be avoided with Live View mode if Sigma cameras have it. In this mode focus is based on what is actually reaching the main sensor. It's only suitable for static subjects though.

Just to add to some of the comments from the two Dans.

Front/back focus is really a calibration issue because with PDAF systems using separate sensors, the optical path is different to the main one for the actual capture sensor. Precision in AF performance comes down to the design of the AF system which is fundamentally an electro-mechanical control system with feedback. Factors such as loop gain and acceptable error signal affect precision but there must invariably be trade-offs with these parameters to ensure system stability (ie no hunting). Different lenses with different focus characteristics must also be allowed for. So I guess it is inevitable that high precision may not always be achieved. By and large though, things work pretty well for my shooting anyway !

Dave

[xpatUSA]

Thanks Urban, Dave et al,

No, the Sigma SD10 design is some 12 years old - so no live view which we have grown to know and love, especially like on my GH1 with it's 10X manual focusing, CDAF and 3" LCD.

With my lower-wavelength IR pass filters (680nm, 700nm, 720nm) the SD10 will usually auto-focus or beep "in focus" in manual. Not always though (it is a Sigma, after all). And they do have variance, successive tries can be different, just enough to irritate. In fact, a nicely focused shot often comes as quite a surprise!

[NorthernFocus]

...By and large though, things work pretty well...

Things work amazingly well. Much better than my old tired eyes. But like most things associated with modern tech, we get spoiled and always want more, better, AND cheaper.

Considering the topic of focus and how difficult it can still be to get right makes me stand in awe every time I now see an old film action photo taken in the days of manual focus. There is a reason so few people had those skills back then.