Canon 24-70L - Lack of focus at wide-aperture and very fast shutter

[mcmahong]

I recently took some photos for a friend in a very brightly lit environment. The images were using a 24-70L with my 50D and of a sunlight wall with my friend leaning against it. The camera was on AV priority at f/2.8 (accidentally, I must admit). As a result, shutter speeds were in the order of 1/2500. I'm also very careful with holding a steady hand.

Because my friend was right of centre in the images, I used one of the off-centre focus points.

Every photo was slightly out of focus. As I scan the wall and the ground, I realise that no part of the image is in focus. So it doesn't seem to be a front or back focus issue. The fast shutter speed means it shouldn't be motion blur.

I can't work out why these images didn't work (well they do, just not when scrutinised at pixel level).

Could it be:
- using one of the diagonal focus points?
- shooting wide-open in conditions that are too bright?
- Anything else?

Any comments appreciated.

[andylo]

Someone gonna say it anyway.... post the problem pic and show us, and we might able to tell you a thing or 2?

[reaction]

flare could stuff up AF and give you an impression of nothing in focus

[mcmahong]

_MG_0254.jpg_MG_0254-2.jpg

I've attached the photos. They don't look bad at full scale, but the zoomed in image shows the fuzziness I am referring to. I should mention these are taken at ISO100 as well.

Ged

[reaction]

that, my dear friend, looks like lens softness to me.
it is in focus

do a few tests on a brick wall, see if it's maybe decentered

[Dylan & Marianne]

Ive found that mine does something similar - even with centre focus and recomposing as well as using different focus points - not sure if i can offer any advice apart from share the same problem

[andylo]

It does look a little blurry.... but there could be a few reasons and the following are my thoughts.

1) You have mentioned you use 1 of the off center focus point because you friend is NOT exactly in the middle, that's good! But did you just have the point set on any part of her body? Or specifically into her eye?

2) To me - it does looks like a camera shake.... I would like to be corrected if I am wrong (and I know it's at 1/2500), but I see a bit of "ghost shadow" around the subject which leads to my suspicious on camera shake.

3) Is it your own lens? Or someone lend it to you? If it is yours, does it had the problem like this before? Have you accidentally dropped it? (even just from a small height)

4) Sort of related to #1, what is the focal length you are using? (if you are using 70mm from 10 meters away at 2.8, it gives you 2.18 DoF.... but still should be enough in your case)

5) I always feel the 24-70L is a sharp lens BUT not a tad sharp lens (I have one myself). If I want a tad sharp image I usually use the 17-40 or the 70-200.

Just throw out a feel thoughts of mine, I am sure other people who knows more can comment better than I do

p.s. I believe the 50D has MFA (Micro Focus Adjustment) maybe you should try that. And you can try my copy of 24-70 and compare to see if anything different.

[EdZz]

I'm really glad that i'm not in the same boat. if you google canon 24-70L focus issues.. results a plenty..

i bought the 24-70L for its reputation of being tact sharp and it being a like a work horse lens. I thought awesome, i'll get this over the 24-105L... but now.. hrmmm

This lens is nortorious for being inconsistent with sharpness, i've ready multiple posts where one day its the greatest lens on earth and then the next day it would compare to a kit len

- The main issue with micro-adjustment is that this feature targets prime lenses where there is no zoom. If you do the adjustments at say 24mm, as soon as you zoom in to say 50mm (or watever).. i'm lead to believe that the micro-adjustments are now irrelevant..
- If you bought it brand new recently and if there a Canon repair centre in ur state, try and take it there for calibration. I live in perth.. so i'd have to send it to the other side of the country.. grrr
- Have you tried a focus test? Use the link below to download the focus chart (page 18 of the PDF) and follow the instructions. You'll need a tripod that can measure an angle of 45degrees (i used a friends one which this feature) and a completely flat surface i.e. tiles, floor boards
http://www.focustestchart.com/focus21.pdf
- To me, its extremely inconsistent with image sharpness... =/


good luck!

[reaction]

That's a good point. micro-adjustments sound like the end-all, but if the micro-adjustment needed at 24 and 70 are different, or different depending on subject distance, then you can't do it.
Send it in for warranty.

[EdZz]

I'm also lead to believe that if you do not focus with the centre AF point, no matter which camera, or at least not with the 5DII and below, sharpness, clarity, even focus is not has good/guaranteed had you focussed with the centre focus point

and is it me or is it a little grainy even at ISO100.. or it could just be camera shake

Oh, and another very important thing to point out is that, even tho ur using a extremely capable lens, the 100% sized picture you took would typically not to be used as a portrait shot like how you zoomed in on the 2nd shot. Like you said, at full size clarity is not an issue until you zoom in, but there would typically be no reason for you to use a picture at full size as this will reveal all imperfections. If you were taking a portrait, of course you would stand much closer, etc etc

[reaction]

actually most cameras have plenty of cross points not just with the centre AF point. 7D is all cross points. Not sure with the case of the 50D.
And he's using an f2.8 lens, which is a prereq for Canon cross points to be cross points.

on a crop body it does look a bit too soft

[mcmahong]

Thanks everyone, a lot to consider. I had to do a bit of googling to think some of this through. I'll respond progressively...

Andylo
1. Firstly, it shouldn't be camera shake as the problem recurs through a dozen similar photos at the same place. Unless I've got a permanent jitter!
2. It is my own lens, but it did once go through a slight knock when my 2yo son got to it (good lesson learned).
3. The focal length was 70mm at f/2.8 (it was an accident to go that wide), but like you say should have enough DOF. As I mentioned, none of the ground is in focus, so if anything it could be back-focus, but I don't get how this could be possible.
4. I won't do any micro-adjustments on a zoom lens - just something I'm not comfortable with

Edzz
I'm seeing the same thing. Some shots are perfect, but then others (like these) aren't. I'm starting to wonder if it loses it at certain zooms, focal distances, focal points, or at different settings... I'll probably need to do an extensive test to see how it performs under different setups.
You're right about the centre focus point. On the 50D, only the centre point has cross-point focus. The rest are just single-plane. A point I found interesting is that the focus points, are actually 2-3 times bigger than the little square that you use to aim with. This means that the most contrasting edges within that zone will be the focal point, causing focus shifts.
By the way, I'd never crop an image that far for a portrait. I try to compose my images from the start, but point taken.

I think the bottom line here, is that I've stumbled upon a shortcoming of the 24-70L lens that I assumed would not be there. Ultimately there's nothing wrong with the images, but I just felt that an L lens would be sharper right through. When I find the time, I'm going to put it through a focus test, looking at all focus points, at different distances, and different focal lengths, to see if I can find a pattern. If I do, I'll post it.

Oh, and maybe send it for calibration.

cheers all

Ged

[arthurking83]

Looking at this image, and then looking at the performance stats for this lens at the settings you used, this image looks pretty much par for course I reckon.
There could be a case of slight misfocus, but there is also enough detail in the image to lead to the conclusion that the lens looks pretty much ok.

FWIW: according to photozone's data, they got a the setting used for the lens in this shot, thy got 75% of the ultimate resolution this lens is capable of at 70mm(which is at f/5.6). So considering the settings, this one doesn't look too bad(have you used any sharpening at all in the image, in camera or in PP?).

I can't see any data in the exif for the focus mode used in the camera at the time of shooting, but this is also a possible angle for exploration too.

Stuff like .. if you've used single shot af mode, have you(or are you prone too) move fore/aft of the correct plane in the short time between focus confirmation and exposure? Has the subject?(unlikely in this instance, but still possible).

If there is no focus point(or sharp points) in the image at all, this doesn't necessarily mean that front/back focus is not an issue either!
Along the ground and close to the woman's shoes, look for a plane of focus. This may be sharply rendered, or just a zone where there is a gradual increase in detail, and then tapers off into blur again.
It's too far to the edge of the frame to be of any real use, as the lens may also be susceptible to field curvature too.(ie. it may not be indicative of the real plane of focus in the image.

[mcmahong]

Thanks Arthur,

Good point, I didn't mention that this image is unsharpened, but I believe is set on +1 in-camera. I just put it in as-is, because other photos I took on the day (close-ups, etc) were beautifully sharp without post-processing.

True, I "think" I was just using single-shot mode, and have only just learned in the last day or two exactly what the other modes mean. I probably should be using AI Focus, I guess just in case I was swaying in the breeze!

There's no plane of focus in whatever foreground I can see on the ground. I checked that before I posted, which is what led to my need for some explanations or ideas. But it's interesting that you say it may not help anyway due to curvature.

I'm going to check out the photozone website now...

cheers
Ged

[arthurking83]

......

There's no plane of focus in whatever foreground I can see on the ground. I checked that before I posted, which is what led to my need for some explanations or ideas. But it's interesting that you say it may not help anyway due to curvature.

I'm going to check out the photozone website now...

cheers
Ged

If you look at the PZ figures on how sharp the lens is at the edges at f/2.8, your comment that you see no plane of focus makes perfectly good sense, as detail is not super duper sharp, just sharp enough for some detail to reveal itself. In this instance, the lack of any super sharp detail will appear as tho the plane of focus doesn't exist.

I think the lens focused accurately and f/2.8 was not the ideal setting for these conditions. The details turned to to be just that little bit hazy in these conditions, but good detail is still there.
There is the possibility that, if you try these settings again, but not in such directly lit conditions, the image may appear to look sharper. Oh and if you do try this test to see if there is any difference, also try another shot, and this time with either -0.3 or maybe -0.7 Ev exposure compensation as well. Slight underexposure also produces the illusion of extra saturation as well as slightly more sharpness, in some situations(possibly these).

Also note that the term sharpening is a relative terms, and the amount you either use or don't use may not even be a decision made by you.

(as an example): If you chose to edit a raw image in LR, your setting of +1 is ignored completely and LR's default sharpening routine is applied instead.
Alternatively: If you shot in jpg mode, then your +1 sharpening is applied to the jpg image. +1 is usually a very low value(I think).

[virgal_tracy]

The following is a post made on another forum (wedding related) when people were asked for recommendations of changing from Canon to Nikon due to focusing problems. The link is for custom settings.

http://www.ashtonlamont.co.uk/custom-settings.html

The key factors are 1) I only use the centre focus point, the other focus points should be regarded as focus assist points not focus points 2) I focus and recompose if necessary 3) I always shoot full size RAW so a focus > recompose followed by a crop in post usually has pretty much the same effect as focusing with an outer point would have done 3) if the subject matter is pretty flat I focus on a more contrasty area at a similar distance. Same in low light. I don't use the AF assist beam - its disabled on my various guns including the main two 580 MKII's. If the depth of field is critically thin and the alternative focus object is not the same distance I may rock my body slightly forwards or backwards 4) AF point area expansion is disabled - very important 5) I use the back button to control focus NEVER EVER EVER EVER the shutter button. This is very easy to get used to but practice it at an event which doesn't matter like a non-paying amateur sports event first if it makes you nervous. I use AI Servo in conjunction with this.

There is also a quote that states that Canon considers in focus as being Canon's standard is "sharp when viewed as an 8x12 print from 2 feet away". Something that is blurred over a radius of one pixel will not be visible under those conditions, so that amount of misfocus would be within spec. I don't know if any of this helps but i will keep an eye on the thread and see if i can come up with any other information

[pmack]

lol at the suggestion of camera shake at 1/2500s...

4. I won't do any micro-adjustments on a zoom lens - just something I'm not comfortable with

Why not? work out the best setting at 24mm, 50mm, and 70mm. it might be +3 +5 +6, so just set is somewhere suitable, you'd be silly to leave it at zero.

I'm seeing the same thing. Some shots are perfect, but then others (like these) aren't. I'm starting to wonder if it loses it at certain zooms, focal distances, focal points, or at different settings...

Are you comparing with other shots at f/2.8?
naturally it is going to be a little softer wide open (though i don't know the specific performance of the 24-70).
you also need to consider the fact that viewing 100% on a 15mP crop sensor is the same as viewing at 100% on a theoretical ~40MP full frame


I want to add that i too have found that sometimes when you think that all things are equal, a lens one second will give you a sharp image, and a slightly soft one the next. sure you could put this down to many many things, including user error, but it seems like there is a comletely unknown variable in all this, lets call it the x factor. now with an L lens, you'd hope the x factor is built in though so i feel ya pain