Focus help for shallowDOF portraits of kids

[ashleynaugust]

I'm trying to improve my shooting. I have been using outdoor, cloudy/indirect lighting, center (or selected) point AI/Servo autofocus; shooting at f/2.8-f/4. Shooting fast moving kids I can't imagine being able to properly manually focus.

My problem is that despite my carefully making sure the focus point is exactly on the eyes, the camera only seems to properly focus on the eyes about 1/3 of the time, the rest of the time it seems to catch their hair/shoulders/something else and their eyes/face is softer & OOF.

I am using the XTi/400D and Tamron17-50 lens. I had the same problem when I used my kit 18-55 lens as well. For older kids I can perhaps freeze them long enough to manually focus, but what can I do to help toddler shots? Do you think it is user error, or more likely a body or lens issue?

[JeffreyG]

Servo focus on the 5D and XXD cameras is worse than on the 1D, and the Rebels are a step down from the XXDs. I found that my old Rebel XT was not reliable in shallow DOF situations (like closer than 15 feet and faster than f/2.8 ) with some lenses.

I'd also point out that the Tamron lens has a pretty slow AF too. Not in the same league as anything with a ring USM motor.

Try posting some examples, but the kicker is usually if the AF is dead on for static targets but misses on moving that it is simply not keeping up. That usually results in just slightly OOF subjects.

[c71clark]

You shooting in AI Servo mode? That might help if you aren't.

[ashleynaugust]

Thanks for explaining that, Jeffrey. As much as I hate to think about needing to upgrade, if it made me get even 1/3 better keepers I think it would be worth it ten times over. If Servo mode isn't up to par on the rebel series, should I choose another mode or they would be equally inefficient?

I am really considering getting the 40D-I have a relative who may want to buy my XTi so it would lessen the cost. Do you think that should make a noticeable difference in focus ability? I don't think I'm going to be able to afford the Canon17-55mm for awhile, I need to prioritize a good telephoto zoom for sports first.

You may be onto something about it just not keeping up-but I noticed a few shots where they really weren't moving, and it focused on the hat/forehead instead. Let me see if I can dig one up to show you what I am experiencing.

[ashleynaugust]

You shooting in AI Servo mode? That might help if you aren't.

I almost always shoot in Servo. I tried switching to the others to see if that would help-it didn't-so I switched back. Thanks for the suggestion though.

[AndreaBFS]

I'm going to go out on a limb and say that from my own experience, I'd have to place blame on the Tamron 17-50 way before I'd place it on the XTi. I can say this because I do own one as my primary everyday lens.

When I still had my XTi, I rented the Canon 17-55 f2.8IS and had a much better keeper rate than with my 40D and Tamron 17-50. In this case I think it's just the consequence of shaving $500 off the price of the lens. The Tamron 17-50 just can't do what the Canon 17-55 can do.

When I switch to my Canon 50mm 1.4, I get a keeper rate that is much closer to the Canon 17-55, but I don't use it as my primary because I really miss the wide end. It's a trade off. I've been working more on adding a bit of fill flash -- it really helps get the eyes sharper.

I am not sure how much longer I'll hold out with the Tamron, which is really a very excellent lens for the money. It just has some limitations that are understandable given the price. It does a great job with my 5 year old because he's willing to pause even briefly, but not for the 2 year old... at all.

[ashleynaugust]

Sorry, I ended up crashing last night. :| I hope it really isn't all the lens, b/c I just wasted some $ I should have saved for the upgrade if that is the case. Here are two examples. In both, somehow it seemed to focus on his clothing, but I had the eye exactly centered in the AF point, taking up the entire AF point in my viewfinder, and I know it wasn't because he was moving b/c the pictures immediately before and after had him in the same spot.

It isn't horribly obvious, but his face/eyes/eyelashes are just soft. Ick.

http://farm4.static.flickr.com/3005/2521841387_81d0737540_o.jpg

http://farm4.static.flickr.com/3047/2521841381_fca758ac0a_o.jpg

[ashleynaugust]

Oh, no sharpness or PP done to these. I think I may have adjusted the exposure .17 of a stop in DPP, and resized and converted to jpeg, if it makes any difference. In the full size version, you can see the detail in the clothing threads nicely sharp, but the face (where focus should have been) without crisp detail. :(

[olly_k]

The xti is suppose to have the same focus system as the 20/30d cameras. I would definately be looking at the lens in your situation. If you are considering a 40d why not look at a 450d?
Lets be honest, SLR cameras and cheaper lenses don't mix as well as one would like. This is a fact that those of us who are on a budget have to live with.

If I were you I would consider a Canon 50mm 1.8 to see how things works out? I would suggest a 17-70 but not sure of it's performance compared to the tAmron and besides I have a 350d which is worse than the XTi!

[AxelFoley]

Before you spend money on a new lens or camera I would suggest eliminating some other possibilities.

Your symptoms sound very similar to a friend of mine. He has a 400D with an expensive L lens and was having similar issues. It was only after a year of blaming himself for poor technique that he read about back focusing. He found that some cameras suffer from back focus problems where they focus slightly behind (or in front of) the target. He tried a focus test - see http://photo.net/learn/focustest/ for an example and discovered that his camera / lens combination was focusing about an inch behind the target at the closest focusing distance which was enough to make the eye soft and fur or feathers further back on the head sharp with his bird and wildlife photography. He sent his camera and lens to canon for a simple recalibration and it is amazing what a difference it has made. All his images are now fantastically sharp exactly where he wants them to be. We did a lot of research on the subject of back-focusing at the time so if you want more information just shout.

[tdodd]

Oh, no sharpness or PP done to these. I think I may have adjusted the exposure .17 of a stop in DPP, and resized and converted to jpeg, if it makes any difference. In the full size version, you can see the detail in the clothing threads nicely sharp, but the face (where focus should have been) without crisp detail. :(
If you're shooting raw and not sharpening at all you will get soft results. The raw file is soft because the AA filter deliberately blurs the image. Try setting sharpening to 3 in DPP and you should see a substantial improvement. 3 is my default sharpening amount in DPP. It is very rare that I use less than 3. If a photo is OOF I may need to bump things up to maybe 5-6. If I feel the need to go beyond 6 I figure the photo is junk.

[adam LC]

That first image looks like back focusing to me, especially if the FP was on his eye. I don't know if it's just me but the wood behind the boys head look sharper.

If it was me, I would print off copies of this and send the lot to Canon for calibration.

[Deckham]

Perhaps a silly question - what focal points are you using?

[airfrogusmc]

I have photographed kids allot for a couple of my clients that have services directly related to children. These shots wind up in annual reports, brochures and other type print and wed media supporting their services. Shooting with shallow DoF is kinda my style. I live in the world between 1.2-4 and I have an 851.2mkI. The slow focuser :rolleyes:.

OK OK I'll get to my point. When I photograph these kids its usually as they are in their world in fast moving situations and I shoot with three lenses, all prime, all fast, 24L, 35L and 85L.

If theres not a problem with body focus on your camera or a lens issue my advice is get comfortable with your gear. Most of the problems with focus are usually a problem of not knowing how to get the best out of your stuff that you have . You need to spend allot of time shooting wide open or near it to get good at it. Theres no book or easy way to get you there. Good glass is only part of the answer.

So to rap it all up I can get good consistent shallow DoF from fast moving kids without allot of FPS and a so called slow focusing lens because I've spent allot of time with my gear and I know how to use it to get the images I need. My advice is spend hours a day with you equipment and if your not having an equipment problem you will learn to get the kind of shots you desire. You will learn where to be because you've done it so much it becomes second nature Focus and exposure will also be second nature.

[olly_k]

That first image looks like back focusing to me, especially if the FP was on his eye. I don't know if it's just me but the wood behind the boys head look sharper.

If it was me, I would print off copies of this and send the lot to Canon for calibration.

I suspect you would get a more helpful response by sending the lot to Tamron?

[AndreaBFS]

I've been tempted to blame my camera or lens, but isn't the more important question whether you are able to get sharp images at certain times and not others?

the camera only seems to properly focus on the eyes about 1/3 of the time

If the camera is focusing properly when the focus point is placed on the eye 1/3 of the time, that's 1/3 too much of the time for it to be a problem with the equipment, no? It's my understanding that those types of problems don't just come and go as the situation changes.

[AxelFoley]

It may be that all the photos are out of focus but the 1/3 that look good have a large enough depth of field to achieve acceptable focus, I wouldn't discount equipment issues without testing.

Test all your lens and camera combinations. Swap camera body or lens with a friend.

Another tip I was given is sometimes it's better to refocus on something far away from your subject and then back on the subject forcing the focus motors to move in and out substantially. Sort of like a reset, camera gets a new grip on the focus point and focuses sharply.

[ashleynaugust]

Thank you for all the tips. I'm going to try the focus tests with both of my lenses (Canon 18-55 and Tam17-50) this afternoon to determine if a back focus issue is occurring, and if it is, if it is with one lens or both.

[ashleynaugust]

Ugh. Just finished shooting. From the LCD preview it is looking like both lenses are showing a slight but noticeable backfocus. Will post examples once I get them cropped & uploaded

[DDCSD]

Are you selecting the focus point or letting the camera choose?

[adam LC]

If you use Zoombrowser it will superimpose the focus points and highlight the FP used, take a screen grab of that and show us all.

[ashleynaugust]

I just processed them all, I didn't know that about Zoombrowser, I always use DPP. If this isn't sufficient JLMK and I'll go reprocess them with ZB.

I used the central focus point for all of these, and used autofocus, and handheld (bracing my arms against the table and using extreme care not to move) I know that isn't a foolproof technique but as I don't normally shoot with a tripod (and don't have one) I thought it would be indicative of real-life usage. The dot in the center of the central focus square in the viewfinder was exactly on the middle of the center chart line in each shot. I took a number of shots from each f/stop, making the camera focus across the room and then coming back, and the resulting shots were all consistent.

Tamron17-50 @ f/2.8
http://farm3.static.flickr.com/2371/2525196785_9741a24802_o.jpg

Tamron17-50 @ f/3.2
http://farm3.static.flickr.com/2130/2525196791_d92a383390_o.jpg

Tamron17-50 @ f/3.5
http://farm4.static.flickr.com/3127/2525196799_81289b1df2_o.jpg

Canon18-55mm @ f/3.5 (I had to use it at about 18mm to get the widest aperture, the results at larger f/stops were similar, but with more DOF)
http://farm4.static.flickr.com/3054/2525196805_d2d7f588c2_o.jpg

So, does this look like a body issue or lens issue or both? The thing is, the line I focused on is in focus, but then it falls off in the front almost immediately. So theoretically, if I focused on the eyes they should be in focus, but not anything in front of them. What I'm seeing from my typical shots is that focus seems to be further back the majority of the time, so it is probably user error combined with slight backfocus? Should I send the lens and/or body in?

[ashleynaugust]

:o Sorry for all the crazy WB issues... I was indoors, and didn't take time to compensate in PP.

[adam LC]

Still really looks like back focusing to me. But you really cannot trust your test without using a tripod and the correct angles. The only time I use zoombrowser if for the FP highlighting, it's the only program I know that does it... I posted an example a while ago, I'll see if I can find it.

If it were me, I would get the body calibrated with the Canon lens. I don't think Canon would touch a Tamron lens, idk.

EDIT: Wow... I found it after half an hour! I remember it was front focusing for me, using zoombrowser....

[AxelFoley]

Depth of field is usually 1/3 in front, 2/3rds behind the focus point. You really should have the camera on a tripod or stand to get the best results. Also use really good lighting, preferably outdoors as this will give you the sharpest contrasts and make the test results easier to interpret.

Zoombrowser shows quite clearly, excellent idea.

I agree that it looks like there may be some backfocus there but difficult to say without knowing how the target was positioned.

Would like to see the zoombrowser overlay on those test targets.

If you are not happy with all your test results I would suggest getting the camera calibrated by canon. If it is uncalibrated you will never get good results no matter how good your technique. Money well spent in my opinion. (usually get a free clean thrown in)

[tdodd]

Depth of field is usually 1/3 in front, 2/3rds behind the focus point. You really should have the camera on a tripod or stand to get the best results. Also use really good lighting, preferably outdoors as this will give you the sharpest contrasts and make the test results easier to interpret.
I think you'll find that at close range, such as these shots, the balance fore and aft is nearer to 50:50 rather than 33:67. Taking the first Tamron test as an example, we're at 50mm and f/2.8. I'm guessing at the subject distance but let's say it's 5' or 60" if you prefer. According to the DOF calculator here - http://www.dofmaster.com/dofjs.html - the DOFat this distance would be from 1.8" in front of the focus point to 1.96" behind. That's a ratio of 48:52. So you would expect a little more to be in focus behind the focus point than in front, but not that much more.

Given the absence of a tripod in these tests I don't think the results can be relied upon as conclusive. A half inch wobble between focusing and firing could easily skew the results. In some images I'd say the results look pretty OK to me. I think to be sure of the results will need a tripod and more than one shot at the target, from a fully defocused starting point, for each aperture/focal length tested. A sample size of one shot per FL/Aperture is not much to confirm poor performance beyond doubt.

[ashleynaugust]

tdodd-I took at least 3-4 shots from each f/stop, and the results were just about identical. That is what I meant when I said that the results were consistent-I took a number of shots, forcing the camera to refocus with each one, and only posted one example (as they were all about the same).

I'll try do it again with teh camera stationary, but probably won't be able to get to it today.

[tdodd]

tdodd-I took at least 3-4 shots from each f/stop, and the results were just about identical. That is what I meant when I said that the results were consistent-I took a number of shots, forcing the camera to refocus with each one, and only posted one example (as they were all about the same).

I'll try do it again with teh camera stationary, but probably won't be able to get to it today.
Sorry, I missed that the first time round.

As for testing the 18-55, even though you get a smaller aperture at 55mm, your DOF will be a lot less at that end of the zoom than at 18mm, and thus should give you a more precise clue as to where you are hitting focus. At 5' subject distance your DOF would be ~39.1" wide open at 18mm but only 6.28" at 55mm and f/5.6.