At my wits end with Canon!

[Ben Rubinstein]

I've been plagued with back focus for a year or more. Since I bought a 5D and new 24-70L I've been having serious problems.

With that 24-70L I sent it back to canon 6 times and each time it came back with a clean bill of health but would still focus on the background intermittently even when there was no reason to, I don't focus recompose and the focus points cannot have been that much out of alignment.

In the end I found out that in florescent light the lens would focus on the background not the subject, tested it with both my 5D's and other lenses. So I gave up on that 24-70L, trashed it and bought another.

This one was OK until I dropped it and had to send it for repair to canon. It came back and hey ho I have the same problem. Borrowed a 24-105L tonight to do some side by side testing and the 24-105L nailed the focus each time, the 24-70L focused on the background each time. Both were shot at 1/160 f4 iso 100 using flash in a dim room, camera shake was not an option, the background was razor sharp!

This has confirmed to me that my 24-70L has weird focus problems, I'd noticed it when working over the past 3 jobs, indoors and out.

When shooting test charts it's fine, it's the real world which is giving me so much aggrivation.

Then I tried to shoot my 85mm 1.8 for comparison. Now I shot a big job last week and every single frame shot wide open was perfectly focused, indoors and out. Tried it out now and sharp ears, every single shot. Turned off the florescents and switched on the tungsten side lights and every single one was perfectly focused.

I'm at my wits end with this. I've been battling it for over a year now and I'm a pro event photographer, photography is my living, I can't afford soft photographs.

So my 24-70L is focusing each time on the background with two 5D's and in comparison to a 24-105L which focuses fine on both bodies. In conjuncture to that the 5D with 85mm refuses to focus accurately in florescent light but perfectly in tungsten. I test it with the 24-105L at the long end, focuses perfectly in either light.

I can't afford this, I can't afford the endless testing and endless worrying, I can't afford the family group that focused on the tree behind (impossible to tell at WA), I can't afford to send cameras and lenses to canon on a monthly basis, CPS or no.

I would switch to Nikon in a second if it would help me but I need FF, I need 13 megapixels and I need the super clean iso 400-1600 that the 5D gives me. On the other hand I'm tearing my hair out every day because Canon can't get their bloody act together!

 

[MArk Le]

I share your experience

but still the 24-70 shows something "unique" in the pictures. I have both the 24-70 and the 24-105 and I do see back focus with the 24-70 (not really that..more like the focus stays in the back, randomly).
I can't explain when or how but that's the result after many years with the lens.

it doesn't have much to do with the camera since I see the same exact behavior on 1Ds, 1D and 5D. No difference.

that said, I still use the 24-70 not only because of the 2.8 with great results.

P.S.: In my opinion the mechanism is not fast enough, sometimes, or the barrel is too heavy or.. whatever: using that lens the focus locked in the back of the subject can happen..

the 24-105 is a great lens, but the 24-70 has something more (in my opinion)

 

[Ben Rubinstein]

Oh I know that, when the lens works it is mind blowing, I just can't live with it focusing behind the subject when the subject is large family group that I have to print big!

Here are samples, the first is the 24-70L, the 2nd the 24-105L, same focal point (between the eyes), same fstop, same settings, same focal length (35mm), the background was 1.5 meters behind the subject:

 

[MArk Le]

exactly

I hope you don't mind if a sort of "smile" came up on my face because what you post looks so.. familiar

that's the way that lens works..just like a pretty (really pretty) girl with "moods"

LOL


ok, get the 24-105 and keep the good work

 

[Ben Rubinstein]

Aargh! I really don't want the 24-105L, it's slower to focus, can't use cross sensors, has ugly bokeh and very nasty vignetting and distortion wide open and the viewfinder is twice as dark! I actually just borrowed one from a friend so that I could take comparison photos to show to Canon so they couldn't tell me that is was fine like they usually do.

What is interesting is that I've been searching for focusing problems and came up with the Pentax people who have worked out that they have a similar issue but with tungsten not fluorescent on their newer camera bodies. The suggestion for them was to use a Hot Mirror filter as the AF sensors cannot deal with the IR, got me thinking, could it be the different coatings on the lenses which are causing the problems, I don't know if there is an opposite to a Hot Mirror Filter or whether it would just screw up the sensor totally, but if the coatings on certain lenses let through a wavelength that the AF sensor could not deal with, i.e. fluorescent, then that would explain why the 24-105L and my 17-40L are perfect, but the 24-70L has caused me non stop problems with focusing on the background and why the 85mm backfocuses in fluorescent light only, the coatings are different.

Added to that there are different types of fluorescent light, with the more modern screw in type bulbs the problem is much decreased, and with muti layer fluorescent lighting mixed with tungsten my 85mm seemed to do fine. It's probably the single or double long tubes without any type of mix that are the real problem. It's worth looking into though as at least I could understand the problem and try to start dealing with it.

Oh and PS, my brother is anything but a pretty girl You say the post is familiar, is this a problem that has come up before? When I asked on FM they seemed to think I was very strange to uncover a problem that has never been seen or heard of...

 

[MArk Le]

Well, I had the same problem

many times, I still remember when I first got the lens and another photographer (Senior AP) told me about it right there. So, if one actually uses the lens it will be a matter of time before you see the focus behind the subject. The only way to deal with it will be shooting more than one shot. That's what I do if the shot is important.

So, it's not a "strange" behavior for that lens, as far as I know.


I'm quite happy with the 24-105 and the difference between 2.8 and 4 is not that big (in my opinion) but if you really need a standard 2.8 zoom then look at the 28-75 tamron. I did a wedding with it and the AF was spot on. All the time. Just the colors are not the same (of the canon I mean). One more thing about the Tamron : if you use the flash then expect random red faces (for no reason) at the corners in particular. Beside that the tamron is as good as the canon.





P.S.: the "girl" thing was referred to the lens
the "familiar" thing was also referred to the lens

 

[Ben Rubinstein]

I know, it was just an opportunity to have a dig at my brother who is about as feminine as a brick wall!

 

[John_Nevill]

You guys have got me worried now, my 24-70L works fine on the 1DMkIIN, but as I'm imminently considering a 5D....would this be a trusty combination?

I also prefer the 24-70 to the 24-105, for similar reasons.

I agree with Mark, the Tamron SP AF 28-75mm F2.8 is pretty damn good but is quite a bit softer and exhibits significant CA wide open at the 28mm end. Although its an extremely compact lens.

 

[Ray West]

Which focus point is being used? Sometimes the background is more interesting. ;-)

 

[dhphoto]

You guys have got me worried now, my 24-70L works fine on the 1DMkIIN, but as I'm imminently considering a 5D....would this be a trusty combination?

I also prefer the 24-70 to the 24-105, for similar reasons.

I agree with Mark, the Tamron SP AF 28-75mm F2.8 is pretty damn good but is quite a bit softer and exhibits significant CA wide open at the 28mm end. Although its an extremely compact lens.

As I have already said on one of the other sites this guy insists on posting this thread on, I have considerable experience with several cameras, 5D's 1DIIN's etc and 24-70L lenses with no problems whatsoever

This is a complete non issue IMHO

David

 

[Ben Rubinstein]

David, that you don't have an issue doesn't mean that I don't, your tone is extremely insulting, I have two 5D's and have owned 3 24-70L lenses with them. The pictures up there are not made up neither are the examples I've been coresponding with Chuck Westfall about. You need to get your attitude sorted out!

 

[dhphoto]

David, that you don't have an issue doesn't mean that I don't, your tone is extremely insulting, I have two 5D's and have owned 3 24-70L lenses with them. The pictures up there are not made up neither are the examples I've been coresponding with Chuck Westfall about. You need to get your attitude sorted out!

Lets just say I've read enough of your posts in various forums on various topics to draw my own conclusions and leave it at that

 

[Ray West]

Hi Ben, David, et al,

I spent some time last night (for another thread), trying to locate an image of the focus point 'red squares' and the actual camera focus points, which someone had created - it was not produced by canon, and iirc it was for a 20d. It showed the latitude/error between what you thought you were focussing on, and what the camera was actually focussing on. I did find this http://www.fredmiranda.com/forum/topic2/241524 which gives an indication of the design difficulties that canon face, in order to perform the magic we crave. Again - can I stress 'tolerance', (pun clearly intended)

I also found many other articles re. focussing issues, some mentioned older lenses did not focus as quickly as newer ones.

I personally do not see many of the problems that folk get. In no way does that mean that the problems do not exist. I, and others do try to help, but often more information is required. It is also very difficult sorting out the facts from some of the pseudo science babble that goes on.

I know Ben has a problem with the type of photography he is concerned with. I am not certain if the errors lie within the tolerance zone of the lens/camera/subject combination. (subject, because if it was, say, pictures of trees, it may never be noticeable.)


Best wishes,

Ray

 

[Ben Rubinstein]

I've posted samples up on the other forum which demonstrate the problem clearly, David seems to have a personal vendetta which enables him to ignore EVIDENCE so that he can have a dig at me.

 

[René Damkot]

Hi Ben, David, et al,

I spent some time last night (for another thread), trying to locate an image of the focus point 'red squares' and the actual camera focus points, which someone had created

Here you go: This is a 10D:

Edit: Image doesn't load ???

Well, here is the link to the page: Click

 

[Ben Rubinstein]

Just an update, tried it out with a new 24-70L and a 50mm 1.4, exactly the same problem. Tried it with an FLD filter, didn't help. I'm really panicking now, can't think of what to do.

 

[Ray West]

Hi Ben,

I know nothing about any other forum, I have enough trouble watching what goes on here. ;-) If we are only getting part of the problem definition, you can not expect to get any sensible solution, even if there is one. I asked earlier which focus points were being used, I may have missed that somewhere, and the whole issue may just be tolerances adding together, as opposed to often cancelling each other out. I believe canon will do their best to adjust the lenses and the camera, but you have to send it all to them, but you know that already.

Test charts are not necessarily a test for your problem, as you found out. There is no background for the sensors to be misled by. The image you show, the face, between the eyes. Where is the contrast that the af sensor needs? There is plenty of contrast pattern in the wallpaper/curtains behind the subject.

If you want an answer, you must give more detail. I would think, if canon keeps returning the lens as OK, it is within their tolerances. Maybe try another canon repair centre - it may be the same test guy, who is not awre of what he should be doing.

Best wishes,

Ray

 

[Ray West]

Hi Rene,

Thanks for the image. I believe canon has them for other cameras too, but that is not the image to which I was referring. Maybe the light goes on in my brain cell, and I realise a) I was imagining it, or b) I can remember the url.

However, the image you posted shows how, if you have a sensor, as viewed in the eyepiece, centred on a small, but flat part of the image, it will focus on a more interesting scene, maybe behind your desired position. The image which I was looking for, was iirc, showing the misalignment between the rectangles as shown by canon, compared to as they are in the real world. It is not too straightforward, the light path taken to the AF sensors.

Best wishes,

Ray

 

[Ray West]

wrt flourescent lights - my take - in crude pseudo science babble -

In UK, mains electricity is at 50 Hz (cycles per second). That means, in effect, it is switching on and off at 100 times per second. A filament lamp, will stay hot and glowing for longer than 1/100th of a second, and will, to all intents and purposes give an even light output, and it will usually have a high red content.

For a florescent light, the original long tube variety, without an electronic control system, depending on the type of coating, it will turn on an off at 100 times/second. They are not used where machinery is used, due to the stroboscopic effect that makes rotating parts of machinary appear stationary. (They also cause eye strain and other problems - disposal - arsenic, etc. ) Now your camera has a micro processor, etc. quite fast enough to pick up this on and off light variation. It will be almost random - i.e. you can not physically press a button at exactly the right instant in 1/100ths of a second. The light output does not have much red content, so, if on an af system, sensitive to red, it is quite likely to mis focus, due to when it measures the light, there is not enough at that split second of time for af to work.

The more modern, small florescent tube/filament lamp long-life replacement type, have electronic controls, and the lamp is switched at a far higher frequency than 50Hz. The same for the 12 or 24V tubes used in vehicles. Many are also colour balanced, to give a more filament lamp type colour output. It is thus less likely that the AF system is fooled.

If you want to talk exposure, hopefully the meter is slow enough to average the light output, but when you press the button, on a fast exposure, who knows. (similar to taking a picture of a scanning tv image.)

Best wishes,

Ray

 

[René Damkot]

wrt flourescent lights

Well, all that will (and does) explain a color cast and different exposures from frame to frame when a fast shutterspeed is used.
I fail to see how it explains differences in AF between a 24-70L and a 24-105L...

That's my biggest problem with this: If it wasn't for the 24-105, my first guess would be "user error": The camera focussed on the BG, because the camera found that an easier target. (more contrast etc).
Since the 24-105 however *is* focussing on the subject, that isn't the case...

Wild guess (Since I don't have, nor have ever used a 5D): The camera has some 'extra' AF points around the center one. Could it be that these don't work with the 24-105, but are used (to focus on the BG) with the 24-70?

 

[Ray West]

Hi Rene,

The 324-70 originated over 2 years before the 24-105, afaik, and was sort of an upgrade to the 28-70, so its fundamental design is much older. I suspect it does not respond as well as the later lenses. 2 years is a lot of development time in electronics. Maybe the camera is able to control the newer lens more accurately, its af drive being more accurate and responsive, what has to be moved is lighter, and so on.

If you spent a week or so, taking a number of sort of measured shots, you may get an answer - maybe canon know and will say, if you ask, or the answer is already out there somewhere. (perhaps look at the differences between the lens construction, and it seems that canon say different things on their different www sites.)

I also do not have a 5d, I don't rember what the exact function of the extra hidden af sensors are.

I think, its like upgrading your pc and software, to run the new lens you need the new camera, although it may sort of run on the old camera, but not the best it can do - and vice versa.

All I know, is my 24-70 is superb, on the 20d, with just the centre focus point (which is more sensitive than the others, in particular on f2.8 lens) except in dim lighting.

Best wishes,

Ray

 

[MArk Le]

we are just talking

nobody should get offended about the opinion Ben has about the 24-70 (which is the same opinion I have as a matter of fact)
It is a great lens but sometimes the focus stays in the back, regardless of where we put it when we press the shutter .

I know the lens limitation in that regard but I keep using it. That's really the end of the story (for me). Of course when I hear a fellow photographer moving remarks that I can share I feel the need to show my sympathy. After all we are all photographers and we do have the duty to tell the truth at least for what we know or see (every day), when we talk to other photographers at least.

this is a 100% crop unprocessed of a recent shot demonstrating the beautiful auto focus performance of this lens. The sad part is that I needed to get the people in the front in focus, not the car behind them. LOL
To be honest the next shot (after re-focus) came out great. That's all this lens (sometimes) needs us to do. Again, we are photographers and we must find the way to make our gear work (whatever we use).

 

[MArk Le]

The image was too big:
here's the link


http://imagingphotographics.com/crop100_g.jpg

 

[Roger Lambert]

Wild guess (Since I don't have, nor have ever used a 5D): The camera has some 'extra' AF points around the center one. Could it be that these don't work with the 24-105, but are used (to focus on the BG) with the 24-70?

I think Rene has a good suggestion here. I remember reading a post at dpr where a high school basketball player reported better results when he eliminated the extra hidden focus points with his 5D.

If one has ONLY the central focus point working with a 2.8 lens, and yet one still gets misfocus - I would have doubts that even manual focus would work under those conditions?

 

[Ben Rubinstein]

The problem is that under tungsten light the camera gets the focus every single time, this is actually about 1-2 EV darker than the fluorescent. Under fluorescent it misfocuses every time. This only seems to be with the lenses faster than f4, the 24-105L, 17-40L and 70-200 f4L seem to be fine. Nor is it the cross sensor that is a problem, this occurs with both the center and off center focus points.

I'm sending both 5D's, the 24-70L and 85mm back to Canon today with a note explaining the problem and telling them to sort it out, whether it be camera, body or whatever. I just wish I had the confidence that they would bother....

 

[John_Schwaller]

Well, all that will (and does) explain a color cast and different exposures from frame to frame when a fast shutterspeed is used.
I fail to see how it explains differences in AF between a 24-70L and a 24-105L...

That's my biggest problem with this: If it wasn't for the 24-105, my first guess would be "user error": The camera focussed on the BG, because the camera found that an easier target. (more contrast etc).
Since the 24-105 however *is* focussing on the subject, that isn't the case...

Wild guess (Since I don't have, nor have ever used a 5D): The camera has some 'extra' AF points around the center one. Could it be that these don't work with the 24-105, but are used (to focus on the BG) with the 24-70?

Rene...

I think there is one difference...on non-1D cameras, I believe the 24-70 will use the high precision AF points whereas the 24-105 will not. Could that be a diiference?

John

 

[Don Finch]

I have had the 5D for more than 1 year with the 24-70mmL as my first lens. I find none of the experiences you have related to be true for me. The lens is my fallback favorite .

 

[Theo Wallis]

Ben after 3 pages why don't you just buy a Nikon! ;o)

 

[MArk Le]

Nikon? well over there Ben will have to rethink about the focals and bokeh and distances all over again because there is no 24-70 but a 17-55 (which is not really 17 to 55) , and the 70-200 becomes something else.. the 85 will be a telephoto lens, the ..

oh well...

LOL

 

[Ben Rubinstein]

Both cameras and both lenses are with CPS at present, I'll update when they send them back having done absolutely nothing (I assume!).

I upgraded my 20D to a 5D because when shooting weddings my 2nd body is almost permenantly set to iso 1600 with a 85mm 1.8 wide open. The only camera to give me the same resolution is the D2X and it is far more expensive and iso 1600.....not sure I want to think about it. A 50mm 1.8 wide open isn't going to look as good either.

If Nikon had been the answer I would have changed, got a lot of friends shooting Nikon professionally but not at higher than iso 400. Thing is that even my 24-70L isn't used stopped down more than f4 other than family groups, I need FF DOF and the amazing resolution and noise of the 5D. Don't get me wrong, the 24-70L is a prime killer. When it focuses right. I don't want to switch to anything else, I really don't, that is why I'm sweating this so much.