Fast Glass focussing technique.

I keep reading how some people have problems focussing wide open with fast primes like the Sigma 50 1.4. I can't help wondering how much of this is the fault of the lens or faulty technique?

There are some pro's who shoot portraits/weddings wide open using one shot AF successfully. I can't do this reliably. I shoot with the 5D3, mostly using AI servo and now have a really high success rate. Particularly since I've MA'd every lens using Reikan Focal.

When I had the 5D2, using centre point, one shot focus/recompose, the success rate was poor. I had the EG-s precision screen, but still couldn't manually focus well. Liveview static subject was ok, but for people shots it just didn't cut it for me. I have no problems shooting motorsport with slow shutter and panning techniques, but one shot AF wide open I can't master.

I'm sorely tempted by the new Sigma 35 1.4 and the 85 1.4 because shooting wide open or thereabouts is where I'm heading. But I keep reading about some peoples toubles.

My guess is that your improvement has more to do with the vastly better AF engine in the 5D3 vs. the 5D2, more than anything else. You now have the AF engine from the flagship 1D pro series ... the best Canon has ever made. Prior, you had a much more consumer/prosumer variety.

AF on the 5D3 blows my mind.
Unlike others I haven't noticed significant improvement with the 85L. Its a great lens (and will always be) but changing systems didn't change much for me.
I use the 85L for a lot of shots but nothing where I will be changing constantly from close to infinity. Guess it wasn't designed for that purpose.

convergent wrote in post #15267397
My guess is that your improvement has more to do with the vastly better AF engine in the 5D3

Of that there's no doubt. But it makes me wonder regarding the supposed poor AF performance of some of Sigma's fast Primes. How much is to do with technique rather than equipment? I'll probably get an 85 1.4 shortly, it'll be interesting to see whether the AF repeatability test thinks the Sigma is any worse or better than Canon glass.

I think that the problem is many fold.. Not simple...
You've got the thinner DOF from the wider apertures... To many- this is new- especially if they have been using kit lenses prior.
You've got oddities- unique qualities of certain lenses-- for example the Sigma 50mm f/1.4 - which tends to front focus / back focus when too close or too far; it's got a sweet spot- an area where it works best.
You've got tolerances- where the body and lens either works together- or sometimes they don't; this is exacerbated by bodies that don't have MFA to handle this.
You've got people that read forums like this one- and hear about the Sigma lottery- for example- and then go looking for issues that they would not have had otherwise...
You've got people that make things worse- by parroting everything they read- for whatever reasons- even though they have never owned, let alone used the lens model in question.
And then you get the occasional user that DOES have an issue. Some attempt to get the problem fixed.. Some don't. Some just swap till they find something that works.. Some complain.. Some ask for a refund.. Some just move on.

I do think that the bodies architecture ( like software ) comes into play as well.. The newer bodies, from what I've seen, are better at focusing than previous models. Some are more accurate- and repeatedly more accurate. I've used 4 DSLR bodies over the past couple of years. If I had to rate them, from worst to best- I'd have to say - T1i, 7D, T4i, 5D3.. It might be that Canon has been improving things and it's showing over time (same order ). It could be that the tolerances on the bodies and lenses are getting closer to norm.. I don't know which way to point the finger- but those are my observations.. Yes, I have a Sigma lens or two...

We try and compensate for the shortcomings of Canon's AF accuracy by coming up with all sorts of excuses or obscure explanations, but at the end of the day, it's just that the system's tolerances aren't good enough

It isn't rocket science - you should be able to point a AF point at something, click, and have it be in focus.

It's hard to see how a photographer can stuff that up.

OP, ironically your AI Servo method of focusing is actually not recommended by Canon, I have canon documentation which says that using AI servo on stationary objects will lead to AF errors. Who knows if that's right or not - it's probably yet another excuse.

smorter wrote in post #15271196
OP, ironically your AI Servo method of focusing is actually not recommended by Canon, I have canon documentation which says that using AI servo on stationary objects will lead to AF errors. Who knows if that's right or not - it's probably yet another excuse.

Handheld shooting portraits, technically both camera and subject are not stationary. Maybe it's the small movement that allows it to work? I have case 2 AF selected and don't allow shutter to operate unless AF is acheived (focus priority).

the great benefit of shooting MF glass is that you don't have to perseverate on whether it was your technique or your equipment that caused a missed shot.

Folks, the difficulty of achieving accurate focus -- even with manual focus film SLRs -- is nothing new!!! Film SLRs had a split image or microprism focus center focus aids. I know, with certainty (I just confirmed this fact only a week ago for another discussion on POTN), that I can focus an old Topcon Super D SLR and 58mm f/1.4 lens with precision if I use the center focus split image or microprism ring; but if I use the general ground glass surround, my focus precision is NOT as good...with a target 5-10' away I might find that I think something is 'in focus' but there is a range of error of several inches in repeatability of focus to the same distance!

In the dSLR, the focusing screen has been made brighter to the eye, but at a loss of focus precision. Screens like the Ex-S improve upon that with fast aperture lenses, but it is NOT sufficient for accuracy of focus precision -- any more than the ground glass surrounds of traditional manual focus screens, like I mentioned in the above paragraph!

I find that AF accuracy is good, and agree with Smorter, "It isn't rocket science - you should be able to point a AF point at something, click, and have it be in focus." And due to the added imprecision of the dSLR focusing screen optimized for brightness -- I don't feel that MF adjustment attempts have much merit and merely is a waste of my time...MF adjustment does not improve upon the AF result. Live View manual focus might reveal a bit of improvement when done at max viewer magnification. But even without LV manual focus the focus precision obtained with AF seems to be on par with the best film SLR focusing aids using a large aperture lens.

Methinks there are a lot of compulsive photographers out there who worry more than they need to worry. Maybe I am lucky...I haven't even found the need to MA; in fact neither my 40D nor my 5D can do MA.

I just conducted an AF accuracy and repeatability test. I set a camera on a tripod, aimed at a barcode which is NOT perfectly parallel to the focal plane but at a slight angle (<30 degrees). The target was 56" from the focal plane (measured with measuring tape). Here is a shot of the overall scene...
shot on 40D, with 17-55mm f/2.8 lens set at 50mm f/2.8. The DOF zone is 1.25" deep.

I think it is pretty well known that prior to the 5D3 and the 1Dx, Canon autofocus systems pretty much sucked. I know I was very frustrated trying to use my 5D classic with wide open primes. No matter what technique I used, there was a substantial randomness in where the point of focus ended up. I had the body calibrated by Canon for a bad front focus issue when I first got it, but even then it sucked. I even bought a $200 split image focus screen to assist my manual focus technique like I did in the old days with film cameras, but the results were a joke. I basically gave up on using any aperture wider than about F 3.5.

I am so much happier with my 5D3!!! It makes all my old prime lenses work like they are supposed to. My keeper rate has gone from about 40% to about 90%.

Many lens reviews that are critical of focus performance were based on older camera bodies. Whenever reading a lens review, be sure to check which body was used.

First check if your lens's manual adjustment is good. If that's okey dokey then check your technique. With my f/1.2 85mm I cannot do recompose ever. Even a slight shake in my hand when I press the shutter button during AI Servo can cause blur. I use the shutter half press button as my focus. I need my subject nearly still since the 85mm is like a elderly grandparent in terms of focus speed. I use center focus if I want the shot.

ride5000 wrote in post #15271503
the great benefit of shooting MF glass is that you don't have to perseverate on whether it was your technique or your equipment that caused a missed shot.

you can't MF the canon 50L? the great benefit of canon lenses is you have the choice.

I think that the big problem is that people are too focused on using their outer focus points when focusing vs just using the center focus point and developing a good recompose technique.

I know that with my 450D my keeper rate went from about 40% when using the outer focus points to about 90% when I learned to just use the center af point only and use a proper recompose technique (not moving the camera inward/outward and instead pulling it sideways and/or down/up to keep everything in focus.

I'm sure that some of the top pros that still use the 5dclassic have pretty much mastered using center af point only and recomposing withing moving the camera forward or backwards and hitting that shutter button fast.

Even on the newer cameras where the outer points use horiz/vert contrast detection too the center point is still better and I'd still only use the center af point as it's more sensitive in lower light (down to f2.8) vs the outer points.

Sovern wrote in post #15273386
I think that the big problem is that people are too focused on using their outer focus points when focusing vs just using the center focus point and developing a good recompose technique.

Is that using fast glass wide open though? I can't reliably nail the 5D3/135L combo at f2 below 10-12 feet. 135L @f3.5+ is ok but at f2 ... The DOF is just too narrow for me to succeed. With static subjects I can just about manage manual focus below 8 feet at f2 in good light with decent support i.e. elbows rested or a door pillar to lean against. Using the 5D2 was worse. The larger viewfinder has helped even though there is no precision focussing screen.

What I'm finding with portraits is that I have to bias the AFMA by around 3-5 and let the AF point rest on the eyebrow(s) with some people. Any cross type AF point on the 5D3 it doesn't matter.