Options
fuzzy images with a fancy lens: what am I doing wrong?
I'm getting fuzzy images with my 70-200mm f2.8 lens when I'm NOT zoomed all the way, whereas shots taken with the same settings are totally snappy at 200mm. Is it me or my lens?
Here's the scenario. I was at a 1920's movie theater that's also enjoying a new life as a concert hall. I was shooting kids auditioning for a country music contest. I was sitting in the 2nd row, perhaps 35 feet from the microphone. I took a few test shots in manual mode, and found a camera setting that worked for me:
ISO-3200, f2.8, 1/250 sec
Then I flipped my settings over to Aperture Priority, left my aperture pegged at 2.8 and zoomed in and out with abandon (though being careful not to let the black back-curtain throw off my meter).
Later, when I downloaded, I noticed that my 200mm zooms are crystal clear, and my wider full-body shots (say in the 80mm range) are all smushy, even when the shutter speed is still up to 250 or higher (wider shots included more back-curtain).
Technique-wise, I thought I was doing everything right. I had 1 of my 9 AF points selected for each shot, which I tried to center on something contrasty like an eye for every frame.
I might have expected my tighter shots to be fuzzy, because at 1/250 sec at 200mm, I would have been encroaching on the rule of thumb of "shutter speed faster than the inverse of focal length." The problem looks more like a mussed focus than camera shake or motion blur, but I couldn't say for sure. Objects slightly closer to me than the singer look perhaps a little clearer, but not significantly different.
I'm not extra knowledgeable about autofocus zones, but I didn't think any of that would come into play so long as I was picking a point.
Here's more details:
Canon 40d. Vertical grip.
monopod mounted at the lens collar.
Sigma 70-200 f2.8 with original mfr hood
Aperture priority (AV)
ISO of 3200 (H)
Drive setting on single frames (not servo)
No flash (contestants are nervous enough already)
Here's the scenario. I was at a 1920's movie theater that's also enjoying a new life as a concert hall. I was shooting kids auditioning for a country music contest. I was sitting in the 2nd row, perhaps 35 feet from the microphone. I took a few test shots in manual mode, and found a camera setting that worked for me:
ISO-3200, f2.8, 1/250 sec
Then I flipped my settings over to Aperture Priority, left my aperture pegged at 2.8 and zoomed in and out with abandon (though being careful not to let the black back-curtain throw off my meter).
Later, when I downloaded, I noticed that my 200mm zooms are crystal clear, and my wider full-body shots (say in the 80mm range) are all smushy, even when the shutter speed is still up to 250 or higher (wider shots included more back-curtain).
Technique-wise, I thought I was doing everything right. I had 1 of my 9 AF points selected for each shot, which I tried to center on something contrasty like an eye for every frame.
I might have expected my tighter shots to be fuzzy, because at 1/250 sec at 200mm, I would have been encroaching on the rule of thumb of "shutter speed faster than the inverse of focal length." The problem looks more like a mussed focus than camera shake or motion blur, but I couldn't say for sure. Objects slightly closer to me than the singer look perhaps a little clearer, but not significantly different.
I'm not extra knowledgeable about autofocus zones, but I didn't think any of that would come into play so long as I was picking a point.
Here's more details:
Canon 40d. Vertical grip.
monopod mounted at the lens collar.
Sigma 70-200 f2.8 with original mfr hood
Aperture priority (AV)
ISO of 3200 (H)
Drive setting on single frames (not servo)
No flash (contestants are nervous enough already)
0
Comments
Moderator of the Cameras and Accessories forums
Thanks for taking an interest, Ziggy.
Okay, I posted two photos to my smugmug. There are two shots for direct comparison. I converted them to jpeg, but didn't do any post-processing other than that (though I was tempted to do a noise reduction for reasons that will be obvious).
http://vocaleyes.smugmug.com/gallery/8325100_vBeez#545472055_gaFRg
From what i can tell, the focus point is not on the face but on the speaker box or at the base of the mic.
Most of my bad photos are caused by user error not camera error.
i'm far from an expert so let's see if Ziggy can give us a better technical answer.
Beauty is in the eye of the beer holder.
Kinky Friedman
GreyLeaf PhotoGraphy
The light looks to be pretty low so you might also have to use the center focus dot, focus, and recompose.
The center focus dot is both more sensitive and more accurate on the 40D, especially using lenses of f2.8 or better.
As it is, it looks like the camera used the center focus dot and it was positioned on the forearm, so that's what is in focus.
The Canon 40D also has a "Live View" and you may use that to focus in difficult situations. Turn on the live view and magnify the view. Now move the view around until you are positioned on the area of the scene you wish to be in focus and use manual focus to visually focus that area.
The second image is slightly less sharp, it appears to me, but the face, shirt and belt buckle are all equally sharp. It looks like the camera focussed on a button on the shirt or close to that, and that is in the same plane as the face.
ISO 3200 will also play a part in how "crisp" an image looks as the ISO noise competes for image detail.
Those are tough conditions and you did fairly well considering. Just practice a bit more with your focus options and you'll be at the limits of that camera's technology for those conditions.
Moderator of the Cameras and Accessories forums
Ziggy>>For the first image, prime focus is on her left arm and it appears to be sharp. If you wanted prime focus on her face you would probably want to chose a single focus point on top and put that dot on her face.
Vocaleyes>>That's what I did for most of the shots at this shoot. I chose a single dot, and aimed for an eye. Some of the shots, I chose an alternate focus point, but still in the top right of my frame (because my subject was most likely to be facing righ-to-left across the frame). Switching back and forth between horizontal and vertical format meant I needed to be quick on the dial to choose a new point when "top right" moved when I turned the camera over. I'm going to chalk the forearm focus up to a focus point in the top right of this photo as if it were horizontal (top left in this case).
Ziggy>>The light looks to be pretty low so you might also have to use the center focus dot, focus, and recompose.
Vocaleyes>>I used to shoot that way all the time before I figured out that I could choose another point of focus. Recomposing is a nuissance when you're shooting on a tripod (or in this case, a monopod), and is less practical when your subject is moving. I prefer to choose a focus dot that's somewhere near where an important detail (such as an eye) will be when the composition I want finally comes together and I push the shutter the rest of the way.
Ziggy>>The center focus dot is both more sensitive and more accurate on the 40D, especially using lenses of f2.8 or better.
Vocaleyes>>Really? You're saying that the center focus dot is more sensitive than any of the other dots? Weird! Perhaps there's more light available in the middle.
Ziggy>>As it is, it looks like the camera used the center focus dot and it was positioned on the forearm, so that's what is in focus.
Ziggy>>The Canon 40D also has a "Live View" and you may use that to focus in difficult situations. Turn on the live view and magnify the view. Now move the view around until you are positioned on the area of the scene you wish to be in focus and use manual focus to visually focus that area.
Vocaleys>>Interesting suggestion. A bit of an end run. Coincidentally I bought this lens to do HS Football, where live view won't be an option. If I were shooting a coffee-bar act, Live View might be just the ticket.
Ziggy>>The second image is slightly less sharp, it appears to me, but the face, shirt and belt buckle are all equally sharp. It looks like the camera focussed on a button on the shirt or close to that, and that is in the same plane as the face.
Vocaleyes>>This is the only full-body shot from the night that I didn't already throw away. I wish I had another example to show you the difference in the other wide shots. They were unsalvagably fuzzy.
Ziggy>>ISO 3200 will also play a part in how "crisp" an image looks as the ISO noise competes for image detail.
Vocaleyes>> Absolutely right. I'll reduce the noise of my keepers very judiciously in Digital Photo Professional before I convert the rest to jpeg.
Ziggy>>Those are tough conditions and you did fairly well considering. Just practice a bit more with your focus options and you'll be at the limits of that camera's technology for those conditions.
Vocaleyes>> Thanks Ziggy. I actually got a ton of shots with a lot of snap and a good range of emotional expression from the contestants. Technology-wise, for the most part I'm thrilled. There's no way I could get flashless shots like this with a camera that didn't have ISO-3200, or with a lens that didn't open up to f2.8.
For the most part, it sounds like your opinion is that the focus was inaccurate, and not that I'm looking at camera shake or motion blur from too slow a shutter speed. I'll have to give this lens a couple more workouts in extreme conditions and see if I can't get it to cooperate better on wider zooms. I remember seeing software somewhere that showed autofocus points as part of the RAW processing interface, but none of my Canon utilities seem to have that feature. That would certainly be helpful in keeping me from second-guessing my technique and showing me if I was actually picking good target features like I thought I was. Do you know of anything with that feature built in?
Thanks again so much.
I believe that Canon's "ZoomBrowser" has the capability of displaying active focus points.
Moderator of the Cameras and Accessories forums
Shooting at f2.8 can be challenging if the desire is to get a whole person sharp in the frame. Usually you will get eyes, but not nose or ears.
Moderator of the Technique Forum and Finishing School on Dgrin
Ziggy,
Thanks for the tip. I took a look at the photo I posted with the view points option turned on. Here's a screenshot:
http://vocaleyes.smugmug.com/gallery/8325100_vBeez#547081168_KxBUR-A-LB
I feel vindicated. You can see by the screenshot that I WAS actually following the method I said I used (pick a focus point, and center it over the subject's eye, then push the shutter the rest of the way down). The shot of the girl (which I also posted) is even more telling. The part of the photo under the focus point is NOT in the primary focal plane. What's more, it's apparent that I didn't do an AF lock on her face, only to have her step backwards out of focus by the time I finished pushing the shutter -- there's no way her face could have been in front of the microphone.
Okay, now that we're actually looking at how I use my autofocus, instead of trusting my memory, perhaps you can see why I consulted this forum. I'm still fairly confident that I was doing EVERYTHING right, and still coming up with weird results.
That leads me back to my original question, I need to figure out if it's me or my camera.
I had a brainstorm about it this morning during my shower. If there were perhaps a smudge on my secondary mirror (or the back of my primary mirror), it could throw off the autofocus, and the effect might be more pronounced at wider angles IF the light comes into the camera at angles. If the light comes out of the last element parallel, that would throw off my hypothesis.
Can you think of anything else that might throw off autofocus at wider angles?
Yes, you were using good technique.
Besides the mirror there is also the AF module itself. If a stray hair should land in the AF module itself it's almost impossible to see without specialized tools and yet it's possible for the hair to disrupt the image going into the AF module and that might affect the ability of the module to do its job.
I still think it was just too dark for the AF sensitivity. You needed to use ISO 3200 just to get an acceptable exposure. Those are not easy circumstances for autofocus. Even a 1D/1Ds body might have difficulty in those conditions.
Just shoot more images and hope that the increased numbers of images will produce an acceptable "crop" of sharp images. Count on some images being out-of-focus and count on deleting them.
Moderator of the Cameras and Accessories forums
Ziggy,
I really hate to argue with you when you've tried so hard to help me, but I have a much more obvious solution. You said something about how I should try using AF lock (center subject, half-shutter to autofocus, reframe, full shutter). I'm pretty sure that what I was doing was a variation on that.
I think I was overeager about half-shutter, and was treating the shutter release like a rifle trigger... (like none of the squeeze counts until the final click). Chances are really high that I was doing the half-shutter, getting a focal point, and then zooming out from 200mm to 80mm WITHOUT LIFTING MY SHUTTER FINGER, and then letting the shutter release all the way down. Even though my focus point light was still illuminated on a good target, it doesn't mean I was still focusing on that point, because the ACTUAL focus migrated with the zoom.
AF Lock: too much of a good thing. Doh!
Thanks again for working on my problem, Ziggy. I wouldn't have arrived at this solution without your input.
I have another audition to shoot on Thursday, so I'll get to try lifting my finger between zooms in exactly the same conditions. I hope I haven't been a pain in the ass.
Alas, I thought I was soooo smart! My repeat shoot gave me the same exact results, even when I made sure not to overuse AF Lock. There's a pattern I'm starting to see. All the fuzzy shots appear to have an actual focus point somewhere nearer to the camera than the point I'm using for my AF target. In my mind, that sort of rules out the cat dander idea Ziggy suggested, which would be less predictable.
My new hypothesis: my AF is out of calibration. Perhaps there's some camera assembly tolerances that are out of adjustment, and the light bouncing around in my camera travels a different distance to get to the AF points (when both mirrors are down) than it travels to get straight to the sensor during the exposure. That would always throw focus off in the same direction.
Perhaps I'm not seeing this with my other lenses because none of them have this wide an aperture (f2.8) in the zoom range where I'm experiencing this problem. The smaller aperture of my other lenses could lengthen the depth of field enough to make the miscalibration less apparent or a non-issue.
Anyone have thought on this? Ziggy? Anyone?
I have a plan of action. I'm going to test my lens on another camera. I also plan to poke around the internet and see if there is anything about AF miscalibration that will help confirm my hypothesis. Stay tuned.
above in arial black and size 4
Art,
Who said anything about hunting? I'm a former marksmanship guy who knows that an indelicate trigger pull can throw off your group, so you're preaching to the choir. It's still a useful analogy because lots of shooters of both kinds are taught that at the end of a gentle squeeze, the actual click should be something of a surprise, since that helps you keep from wiggling (flinching) right at the wrong moment.
This phenomenon of having objects in focus that are nearer than my focus target has me baffled. I wrote to Canon, and they said it might be a lens communication issue and to clean my contacts with an eraser. I didn't tell them I was using a Sigma lens. I'm sure that when I tell them it didn't help they'll tell me to buy a Canon L lens. I think I'll try to get a second opinion from Sigma support (assuming there is such a thing). Sounds to me like you swear by Sigma lenses.
If it only happens on a single lens then it is a lens issue and you may need to send the lens in for repair and/or recalibration.
It's awfully difficult to diagnose with a single lens but most repair shops will be able to test against several lenses.
Another thing you can do just to verify a problem exists is to put the camera on a tripod and lock it down pointed at a suitable target and disstance. Use AF to focus on the target, switch the lens to manual focus, and then switch the camera to Live View and see if the target is in focus or if it needs focus tweaking. Live View can be used this way to diagnose an AF error but it will not necessarily identify the cause.
Moderator of the Cameras and Accessories forums