Whats your autofocus success rate below f2.8?

Messages
410
Name
Andy Reed
Edit My Images
No
In a similar vein to the percentage of keepers thread. I was photographing my godchildren on the beach over the weekend and wanted to capture them with a very close depth of field. So shooting at f2.8 at a high shutter speed (greater than 1/500th) when I viewed my results I was fairly disappointed in the number of keepers. A huge number of these shots I found to be poorly focused and this led me to wonder whether this is typical or I'm doing something wrong. I was shooting in centre point AI Servo with the 70-200 2.8L and IS on. The shutter speeds lead me to think its focusing rather than shake and the variety of shot types rules out motion blur too. Ok the kids were moving around a lot but even some of those with very little movement were out whilst the keepers were spot-on pin sharp.

I've also been experimenting with the 'focus on *' custom function enabled which I really like. It has improved my hit rate but I feel as though it could be better.

So my question is when shooting at low f-stops (i.e. f2.8 or less) what kind of success rate do you guys expect from your auto focus?

Much appreciated,
Andy.
 
Why were you using AI servo, were they moving much? If so the 5D is notorioulsy slow for tracking moving subjects.
 
I've been trailing the 'focus on * button' custom function whereby you leave the camera in AI Servo. A single tap of the * button gives you a one-shot auto focus and holding the button down gives you tracking.

To answer your question they were moving around quite a bit but were static at times. This method of focusing certainly seems to make it faster to switch between one-shot for static subjects and AI Servo when they start moving all of a sudden.

I assume you don't use AI Servo that often then? Do you find one-shot accurate enough when shooting people with very short DOF?

Andy.
 
Never had a problem with the 5d for moving stuff (racing cars etc.), but I would definitely use the nearest focusing point if I'm shooting narrow apertures. Certainly found with the 50mm 1.8 and 85mm 1.8 that's the case, rather than focusing and recomposing.
 
Fair point too. I've got a 5D and the 9 point AF is just not quick enough for anything moving quickly. I got really frustrated trying to take pics of the dogs running. Subjects are ok if you are shooting them side on and panning but anything head on, it will struggle. I also find that you really don't have time to choose the focus spot.

The best compromise that I've found is to shoot more when they are slightly less active or moving more side on. The ultimate solution is a faster AF camera, when I got the 1Ds I could not believe the difference. I took the 5D with me last weekend and I was back to virtually nil keepers.

Did you have the ISO turned up? I'm thinking that possibly a higher ISO would have allowed perhaps f5.6 and easier focussing?
 
Never had a problem with the 5d for moving stuff (racing cars etc.), but I would definitely use the nearest focusing point if I'm shooting narrow apertures. Certainly found with the 50mm 1.8 and 85mm 1.8 that's the case, rather than focusing and recomposing.

What focus mode would you use most often there Mike? Do you find much difference in accuracy and hit rate between one-shot and AI Servo?

Andy.
 
Never had a problem with the 5d for moving stuff (racing cars etc.),

Were the side on (panning) or head on Mike? Head on over a short distance close to the shooter is always going to be much more of a challenge for the AF, hence asking what everyone means by "moving" :lol:
 
By all active I guess you mean letting the camera chose the focus point, I never do that. I find that I have got manually choosing a focus point and always choose the nearest one to the point I want in focus.
So, in effect, you are changing the focus point for each shot? :shrug:
 
Did you have the ISO turned up? I'm thinking that possibly a higher ISO would have allowed perhaps f5.6 and easier focussing?

Hi Ali,

No I purposely kept it low on this occasion as I wanted a nice short DOF. I realise I made it difficult for myself but I had expected more keepers than I got.

Andy.
 
not sure it's the same with the 5D but with the 30D the focus area is 3X the size of the markings. If something more attractive to the AF lies within this area it may lock onto it, rather than what you intended to focus on. With f2.8 and less and at closer distances, this reduced depth of field can cause issues.
 
So, in effect, you are changing the focus point for each shot? :shrug:

Yup, but it's a lot easier on bodies with the thumbstick as you can set that up to shift focus points around very quickly, harder work in portrait though :(
 
Were the side on (panning) or head on Mike? Head on over a short distance close to the shooter is always going to be much more of a challenge for the AF, hence asking what everyone means by "moving" :lol:

Head on, front three quarters. Panning is easy, you can do it easily enough by manually focusing.

Also I've done some car to car stuff, and not had problems with that.

Not saying it isn't a problem, I've not used the 5d THAT often for action, but the times I have it's always been good.
 
Yup, but it's a lot easier on bodies with the thumbstick as you can set that up to shift focus points around very quickly, harder work in portrait though :(
Interesting that it appears to give sharper results than using the 'centre spot and recompose' method though.

*runs off to set it up* :exit:
 
Interesting that it appears to give sharper results than using the 'centre spot and recompose' method though.

*runs off to set it up* :exit:

With a narrow DoF focus and recompose can be enough to change the distance between the camera and the subject that the point you focused on is no longer in sharp focus. The time between focus and recomposure can also be enough for you and/or the subject to move OOF as well.
 
as said it sound like the focus point has moved by the time shutter button has been pressed. You did say that the kids were moving so that might be why.
@1/500s i would have stopped down a tad so the DOF wasnt so shallow.
 
Can you grab some software for the Canon, I know Nikon has some that would show you where the focus point was, I find it useful to see why I may have missed a shot that I thought was ok, not sure if Lightroom can do this?
 
as said it sound like the focus point has moved by the time shutter button has been pressed. You did say that the kids were moving so that might be why.
@1/500s i would have stopped down a tad so the DOF wasnt so shallow.

Hmm... as in the original post though that's what I'm trying to achieve. Accurate auto focus at short DOF. It seems obvious that I could have stopped down to f8 and reduced the chance of focusing error.

I can accept that when shooting fast moving subjects at short DOF in the very brief time between the focus being set and the shutter being opened the subject can alter their distance. What I was hoping to get a feel for was how often this affects people. Ok, maybe I need to describe a theoretical scenario to gauge your success rates.

You're photographing a group of adults standing at an event from a few meters away using a 70-200 @ f2.8. Using a single focus point and a fast shutter speed (~1/500th). Assuming there is movement in the group but people are essentially standing still. You take 20 shots attempting to get your selected focus point on the eyes of someone in the group each time. How many of the shots would you expect to have nice sharp eyes?

Is it possible to expect them all to? I can't help but feel that I ought to expect a higher number of keepers. Otherwise why are faster lenses like the 70-200 2.8 so popular over their slightly slower f4 versions for things like indoor sports?

Andy.
 
In that situation I wouldn't use AI Servo, I'd focus then recompose, although I can't really see how (*) focussing would make a difference there, but I agree I'd expect most of them to be in focus.
 
I shot a wedding on the 28th Dec and for the candids and some of the ceremony I shot using a 70-200mm. It's a Sigma and has no IS. Other than a few that I rejected due to some very strong sidelighting producing unflattering lighting I'd say I had over 90% technical keepers and I was shooting well below 1/500 sec too. ISO 800-1000 was the order of the day and I was often down to 1/60 to 1/100 sec.

I also shoot on a canon f4 and I use that one when I know I'm going to be using flash and/or shooting f4 and above. I use the f2.8 for lower lighting conditions and for better depth of field on fairly static objects.

Hope that helps explain why I do it that way.
 
Yup, but it's a lot easier on bodies with the thumbstick as you can set that up to shift focus points around very quickly, harder work in portrait though :(

:agree:

On the Alpha 700, the autofocus points are controlled with a thumbstick and it is the exactly the same on the vertical grip so I change the focus point without even lowering the camera.

I must admit I wouldn't try to shoot my kids at 2.8 when they are running around though, far easier to light the room with an off camera flash and shooot at between F/5.6 and F/8.
 
I shot a wedding on the 28th Dec and for the candids and some of the ceremony I shot using a 70-200mm. It's a Sigma and has no IS. Other than a few that I rejected due to some very strong sidelighting producing unflattering lighting I'd say I had over 90% technical keepers and I was shooting well below 1/500 sec too. ISO 800-1000 was the order of the day and I was often down to 1/60 to 1/100 sec.

I also shoot on a canon f4 and I use that one when I know I'm going to be using flash and/or shooting f4 and above. I use the f2.8 for lower lighting conditions and for better depth of field on fairly static objects.

Hope that helps explain why I do it that way.

Thats very useful. Would you be shooting that in one-shot focus and recompose?

I must admit I wouldn't try to shoot my kids at 2.8 when they are running around though

Hmm... maybe I'm expecting too much. I just know I've seen shots of indoor sports such as basketball shot at 2.8 with pin sharp focus and assumed its readily repeatable.

Andy.
 
The 5D's Af isn't really up to tracking action such as basketball and it would be servo all the way for something like that. I tend to stick with servo when shooting around f/2 in low light and have a keeper rate of 90% or more but that's with a 1 series body which has much better AF performance.
 
Great stuff. Thanks for your help. It sounds as though I'm probably expecting a bit much from the 5D auto focus on fast moving subjects with such short DOF. Your experiences sound about right on more static subjects though. Ali's use of the 2.8 vs f4 lenses seems to mirror mine pretty well, so thanks for that reassurance.

Best regards,
Andy.
 
Shallow depth of field, moving subject and relatively poor continuous focus mechanism by todays standards = quite a lot of misses.

BTW you probably don't need anything like 1/500th to freeze people, 1/150th is probably good enough unless you are trying to shoot someone being fired from a cannon or some sort of Jamaican sprinter...

PS if you are using f2.8 because of "focusing in low light", remember the AF process always uses the maximum aperture, then closes to what you want for the shot when the mirror flips up...

Also, dropping from 1/500 to 1/150th will give you back several stops of light in the first place...
 
General rule for shooting people running is min. of 1/250s and ideally 1/500s. 1/150s will show subject blur even on a slow runner.
 
Actually, I was joking about sprinters, I did actually mean people standing around :D
 
BTW you probably don't need anything like 1/500th to freeze people, 1/150th is probably good enough unless you are trying to shoot someone being fired from a cannon or some sort of Jamaican sprinter...

Also, dropping from 1/500 to 1/150th will give you back several stops of light in the first place...

Yeah, I'm fine with that. The reason for the 1/500th shutter speed was shooting in bright sunlight already down at ISO 100. As mentioned previously I wanted the very close DOF therefore f2.8. This wasn't a low light situation.

Andy.
 
Aha! Ok, deliberately trying for the shallow DOF...

Its still asking a lot from a wiggling subject though - have you worked out just how shallow the DOF would have been on your subject? From what you are saying we are talking pretty damned close which will result in a razor's edge of DOF... the slightest movement and thats gone somewhere you don't want it.
 
I haven't worked it out mathematically. Just from gut feel really. My guess would be around 5 to 6 inches or so. Maybe that would be worth confirming though. Any recommendation as to best calculate it?

Andy.
 
Did you work it out? Go on, what was it? :D
 
Well they vary but some of the duffers had as little as 4cm DOF which does surprise me and certainly explains why fast head-on movement caused issues.

Ok, so looking at this from the low light point of view. The scenario described earlier with fairly static subjects photographed in low light. You're shooting at high ISO and f2.8 to get a reasonable shutter speed. Presumably your preference is to shoot at the longer length of the focal range from further away to get safer DOF? In my mind closer subjects at wider angles equates to much shorter DOF and therefore the increased risk of missing your focus. Fair to say?

Andy.
 
Back
Top