Canon 70d Focus Issue - Am I going mad?

[EnthusiastOwned]

Hello All,

I've been looking at my camera/images so much no i'm getting nob eyes so need your input. I'm either going mad or I'm right.

My Canon 70d has started with what seems to me, a focus issue. Most of my recent photo's appear to be out of focus ever so slightly. The camera is around 14 months old but I have a 2 year warranty from the supplying shop. Sent the camera off to the supplying shop and the camera was sent to Cannon where the only issue found was with the calibration. I quote - "They [Canon] have calibrated the camera autofocus to specification and have also calibrated the lens’s focusing unit to specification."

Got the Camera back yesterday and the kit lens has a micro adjustment W +2 (adjust by lens). I can't see any other changes to settings. According to the shop, the autofocus points can be calibrated by anyone, although a google search has only brought up micro adjustment - is there another focus calibration setting for the 70d? Call me sceptical, but feel like I've been mugged off by Canon.

Anyway, I've tested the camera today where it still seems off. I have a calibration card. I've set it up in a fixed location, single centre point AF, used a prime lens (so I don't have to worry about zoom), on a sturdy tripod with a remote and delayed shutter. I'm seeing inconsistent results. These 6 images were literally taken shot after shot, looks to me like I have a +/- 3 variation.

IMG_9678.jpg by Rich Perkins, on Flickr
IMG_9677.jpg by Rich Perkins, on Flickr
IMG_9676.jpg by Rich Perkins, on Flickr
IMG_9675.jpg by Rich Perkins, on Flickr
IMG_9674.jpg by Rich Perkins, on Flickr
IMG_9673.jpg by Rich Perkins, on Flickr

Is what I'm seeing normal or a potential issue? This to me doesn't seem right - If i'm getting 3cm focus differences on a tripod, I have no chance of getting anything whilst on the move.

Any advice or pointers to put me on the right track please?

[tenohfive]

Bookmarking this thread. Same body, same issue (minus the trip to Canon.) I've even got those same cards but haven't got around to using it yet - and IIRC they only work when using Liveview rather than through the viewfinder?

[LongQ]

Have you seen this:

http://www.learn.usa.canon.com/app/pdfs/articles/A...

[LongQ]

I note the sample shots were taken with a 50mm f1.4 wide open.

An interesting lens but historically such fast lenses have never really been sharp wide open. They can be very soft wide open.

All of the samples seem much the same.

Nowhere do I see sharpness although the eye tends to discern more possible sharpness on the left of the target.

I'm not sure if this is a faux effect - is the target more BOLD on that side?

It could be that the lens or mount is decentred touch. With a very shallow DoF and a soft lens setting any margin alignment problem could simply result in nothing specific being selected as the point of focus.

Can you get a successful sharpness manually focusing?

I have experimented with micro-adjustments with a few lenses on my 1d3 - mainly telephoto zooms.

Extremely inconclusive and not always consistent results shot to shot. After a short time ones judgement becomes suspect, even when shooting tethered and immediately processing the RAW file.

Stay sane!

[EnthusiastOwned]

tenohfive said:

Bookmarking this thread. Same body, same issue (minus the trip to Canon.) I've even got those same cards but haven't got around to using it yet - and IIRC they only work when using Liveview rather than through the viewfinder?

Google it, huge issue on the early Cameras apparently and a lot of Canon denying an issue. Rumours of an unofficial firmware patch.

LongQ said:

Have you seen this:

http://www.learn.usa.canon.com/app/pdfs/articles/A...

Thanks for this. I've performed the microadjustment as per the manual but this is more comprehensive. I've pretty much followed the guide to a T anyway. Currently on a +5 adjustment. My issue is, when performing the tests i'm seeing a -2 to +1 difference in my sample shots, so I have no idea what to put the microadjustment on.

LongQ said:

I note the sample shots were taken with a 50mm f1.4 wide open.

An interesting lens but historically such fast lenses have never really been sharp wide open. They can be very soft wide open.

All of the samples seem much the same.

Nowhere do I see sharpness although the eye tends to discern more possible sharpness on the left of the target.

I'm not sure if this is a faux effect - is the target more BOLD on that side?

It could be that the lens or mount is decentred touch. With a very shallow DoF and a soft lens setting any margin alignment problem could simply result in nothing specific being selected as the point of focus.

Can you get a successful sharpness manually focusing?

I have experimented with micro-adjustments with a few lenses on my 1d3 - mainly telephoto zooms.

Extremely inconclusive and not always consistent results shot to shot. After a short time ones judgement becomes suspect, even when shooting tethered and immediately processing the RAW file.

Stay sane!

Thanks for this. My set up for these tests are likely not perpendicular so focus could be at an angle meaning sharper to the left. Either way, I would expect the majority of the shots to be the same. I'm seeing focus jump between -2 and +1. That's a 3cm change when using a tripod!

The lens is a little soft wide open. But this is generally my sharpest lens (especially at f3.2 which I like to shoot at). But highlights my issues: softness, focus discrepancies and chromatic aberration. All my lenses do this and this actual lens used was replaced under warranty and is brand new (I though this lens was at fault at first and John Lewis just swapped for me).

I'll try these tests with manual focus to see the change and also with my kit lens (18-135mm). But before I sent this off to Canon I noted that Manual focus and live view were spot on. Single point focus was ok and multi-focus terrible.

[EnthusiastOwned]

Ok, had a play this morning. I followed the link above perfectly.

At the recommended 2.5m - My tool is too small so accuracy is a no no. at 1m (same as the tests above) I can pin point more.

The only thing I wasn't doing on my old tests was re-setting the focus so the camera had to re-focus each time. That's made a difference with my tests. This time I've come to set the microadjustment at -3 for this lens. When taking a set of six I've moved to focus ring backwards and forwards between each shot (so the focus either has to pull back or forward).

I've noticed if the focus starts behind my target, the focus settles ever so slightly behind after autofocus. If the focus starts in front of my target, the focus point is ever so slightly in front after auto focus. Basically the lens isn't perfect so can't reach the desired focus point the camera is telling it to. There is still variation but I've come to the conclusion his is the best i'll get but variation is so small it shouldn't be an issue. I've settled on +/- 1 now. I'll take that.

However, something still isn't right. Live view is taking great shots. Same shot, tripod and same setting with the view finder and focus is still out. I took x4 shots, all re-focusing each time. Live view perfect 4/4. Viewfinder 0/4.

E.g:
IMG_9793 by Rich Perkins, on Flickr
IMG_9789 by Rich Perkins, on Flickr

Look at the little metal pins and ring at the top.

It's so minor, but something isn't right. It's the difference between that perfect shot and an ok one. Thoughts?

Edited by EnthusiastOwned on Sunday 18th September 11:21

[LongQ]

I had a look at the EXIF info.

Very comprehensive - but that may not be a great thing for understanding unless one is immersed in it. I think it mainly reports camera settings even if not all settings are pertinent to the shot taken. Not a great help when wondering what to look for.

2 thoughts.

Firstly the AF system benefits from really good light and a high contrast area in the focus zone, ideally horizontal or, second preferent, vertical. Ideally that target would fill the zone in one or both dimensions. A flat subject, as per the test cards, would be better. Looking at your image most of those points are compromised slightly although one might expect a system to be able to deal with that to at greater extent than it appears to.

Secondly the EXIF suggests you are using a focusing zone rather than a spot point and that 2 of the points are "in focus".

Which point those are is difficult to say without working it out with the camera but there seems to be information to do it. As I recall the Canon software allows that Exif data to be deployed to overlay the focus point info an an image so it might be worth checking that. If it does in this case you will, hopefully, be able to see where the camera thinks it has focused.

If the settings are looking for an optimum possible result over a wide target area it is quite possible that no single part of the image will appeer to be in good focus when pixel peeping - as is required for this type of focus analysis.

Bear in mind that this is not an analogue process - it's an interrelation of digital data and dependent on what data the camera designers choose to record under the circumstances of the shot. Thereafter the ultimate result will depend on how the processing software understands and deals with those data.

Which is why, from time to time, we seem to get unexplainable results.

I am not saying the points mentioned are the reasons for your discrepancies but they are things to be considered and, clearly, the camera and lens combination is capable of providing good results.

BTW the Micro focus adjustments that seem to be registered in the EXIF looks a bit suspicious given what you have written up in the thread. The specific frame seem to have a used a value of -3 Might it be worth clearing the settings and starting again?

Sorry, rather generic and garbled thoughts but maybe they will prompt you with ideas of your own based on what you know you have already tried?

[EnthusiastOwned]

LongQ said:

I had a look at the EXIF info.

Very comprehensive - but that may not be a great thing for understanding unless one is immersed in it. I think it mainly reports camera settings even if not all settings are pertinent to the shot taken. Not a great help when wondering what to look for.

2 thoughts.

Firstly the AF system benefits from really good light and a high contrast area in the focus zone, ideally horizontal or, second preferent, vertical. Ideally that target would fill the zone in one or both dimensions. A flat subject, as per the test cards, would be better. Looking at your image most of those points are compromised slightly although one might expect a system to be able to deal with that to at greater extent than it appears to.

Secondly the EXIF suggests you are using a focusing zone rather than a spot point and that 2 of the points are "in focus".

Which point those are is difficult to say without working it out with the camera but there seems to be information to do it. As I recall the Canon software allows that Exif data to be deployed to overlay the focus point info an an image so it might be worth checking that. If it does in this case you will, hopefully, be able to see where the camera thinks it has focused.

If the settings are looking for an optimum possible result over a wide target area it is quite possible that no single part of the image will appeer to be in good focus when pixel peeping - as is required for this type of focus analysis.

Bear in mind that this is not an analogue process - it's an interrelation of digital data and dependent on what data the camera designers choose to record under the circumstances of the shot. Thereafter the ultimate result will depend on how the processing software understands and deals with those data.

Which is why, from time to time, we seem to get unexplainable results.

I am not saying the points mentioned are the reasons for your discrepancies but they are things to be considered and, clearly, the camera and lens combination is capable of providing good results.

BTW the Micro focus adjustments that seem to be registered in the EXIF looks a bit suspicious given what you have written up in the thread. The specific frame seem to have a used a value of -3 Might it be worth clearing the settings and starting again?

Sorry, rather generic and garbled thoughts but maybe they will prompt you with ideas of your own based on what you know you have already tried?

Thanks for this, it makes sense and is a more eloquent version of what i'm thinking. Basically, this camera has 19 AF Points but I always use the centre weighted 6 point multi-focus (the six centre points, imagine 3x3 squares). The camera will then choose the optimum focus point(s) to use from these 6 - It can use all 6 or just 1, depending on the subject. I found using the whole 19 AF not accurate enough and if I reduce the points any less, far too accurate (such as focusing on a nose rather than face as an example). I have used these 6 focus points since very early on in the camera ownership whereby the first 12 months use were perfect - Up until recently I'd have expected no less of a result than what i'm only seeing in LiveView now - Something has deteriorated.

When selecting x1 specific AF Point things seem much better (although too accurate for my liking). Live view uses it's own AF system. So to me, this would suggest the camera appears to be having trouble selecting the correct points when faced with the option of more than one AF point, when using the view finder.

I have noticed (unsure if it's me being over sensitive) when using these 6 points, the camera is reluctant to use the top three. Possibly the viewfinder AF system is screwed so the camera simply can't detect what it should be focusing on properly.

Micro Adjustments is correct for the lens at -3 (I played with it this morning and managed a +/- 1 accuracy).

I also took the same shot as above with the kit lens (18-135mm) earlier with very, very similar results. To confirm, this lens and all focus settings are as Cannon sent back to me. The issue is not as bad, but it's still there (look at the metal ring at the top of the fitting). I feel Microadjustment is just masking an underlying issue with the AF system.

IMG_9807 by Rich Perkins, on Flickr
IMG_9803 by Rich Perkins, on Flickr

Tell me i'm not seeing the focus issue.

Edited by EnthusiastOwned on Sunday 18th September 14:23

[LongQ]

I think you really need to work out where focus points 5 and 10 are in relation to that subject shot.

Even so, if you are looking at outright lens and camera focusing accuracy, you need to start with a well lit subject preferably in a vertical plane and using good focus point - I'm not familiar with the 70D but on most cameras not all focus points are of equal competence.

If you are looking at a subject that is small compared to the area of focus points you are using the camera is always likely to be in "compromise" mode.

Also if you don't have much good light. And for really critical stuff the AFs seem, in my experience, to me more senssive to colours than one might expect (until one considers the way they work).

If you consider your hanging bulb, and assuming you were focusing on the fitting .... it's not flat and it's not that well defined for contrast. In that resulting image it's not really well lit either, for a test case.

Unless your group of focus points covered ONLY the fitting (and taking distance of potential "DoF" into account given the wide open lens) the range of distance to subject matter (camera to light fitting vs. camera to ceiling) is very significant.

Are there any other setting for the focus decision system that might be in play?

For example my 1D3 has an option to prefer speed of shutter response over accuracy of focus activity in order to grab action shots where ultimate sharpness (for press work for example) may be secondary to capturing the action. I don't recall seeing anything in the EXIF of your file ... but then there is not reason to suppose that everything is reported.

I would also observe that I have in the past found it very easy to lead myself on a wild goose chase when what has changed is not the camera gear bu trather the way I have been shooting, some unacknowledged change in predominant subject matter or approach to shooting or, perhaps, a setting I have made in camera and then totally forgotten about. Most recently that was using a body that I normally only use with manual lenses but had a few weeks back been using to experiment with "back button focus". Totally forgot that it was set that way, completely forgot (and so ignored) the setting warning wondering what it might be and it was much too long before the penny dropped. It wasn't helped an occasional press of the button during normal camera handling causing things to suddenly work (or not work) and the working distance likely to be set being close to what I needed anyway.

I really could not work out why the AF seemed to playing up - although live view seemed to be OK ...

Switched off the custom setting and life returned to normal - or what passes for normal.

I only ever use a single focus point - or rather I can't recall using area focus points for a long time. Tried them when I hired a 7D once and was not convinced they were right for the subject matter and lens I was using at the time.

[EnthusiastOwned]

LongQ said:

I think you really need to work out where focus points 5 and 10 are in relation to that subject shot.

Even so, if you are looking at outright lens and camera focusing accuracy, you need to start with a well lit subject preferably in a vertical plane and using good focus point - I'm not familiar with the 70D but on most cameras not all focus points are of equal competence.

If you are looking at a subject that is small compared to the area of focus points you are using the camera is always likely to be in "compromise" mode.

Also if you don't have much good light. And for really critical stuff the AFs seem, in my experience, to me more senssive to colours than one might expect (until one considers the way they work).

If you consider your hanging bulb, and assuming you were focusing on the fitting .... it's not flat and it's not that well defined for contrast. In that resulting image it's not really well lit either, for a test case.

Unless your group of focus points covered ONLY the fitting (and taking distance of potential "DoF" into account given the wide open lens) the range of distance to subject matter (camera to light fitting vs. camera to ceiling) is very significant.

Are there any other setting for the focus decision system that might be in play?

For example my 1D3 has an option to prefer speed of shutter response over accuracy of focus activity in order to grab action shots where ultimate sharpness (for press work for example) may be secondary to capturing the action. I don't recall seeing anything in the EXIF of your file ... but then there is not reason to suppose that everything is reported.

I would also observe that I have in the past found it very easy to lead myself on a wild goose chase when what has changed is not the camera gear bu trather the way I have been shooting, some unacknowledged change in predominant subject matter or approach to shooting or, perhaps, a setting I have made in camera and then totally forgotten about. Most recently that was using a body that I normally only use with manual lenses but had a few weeks back been using to experiment with "back button focus". Totally forgot that it was set that way, completely forgot (and so ignored) the setting warning wondering what it might be and it was much too long before the penny dropped. It wasn't helped an occasional press of the button during normal camera handling causing things to suddenly work (or not work) and the working distance likely to be set being close to what I needed anyway.

I really could not work out why the AF seemed to playing up - although live view seemed to be OK ...

Switched off the custom setting and life returned to normal - or what passes for normal.

I only ever use a single focus point - or rather I can't recall using area focus points for a long time. Tried them when I hired a 7D once and was not convinced they were right for the subject matter and lens I was using at the time.

Ok, so maybe the bulb/shade wasn't the best test subject due to the reflections. I believe 5/10 were pointed centre and bottom right of the bulb respectively. Either way, I always re-focus a couple of time to ensure the Multi-AF is pointing where I want it on the subject. My method hasn't changed in over 12 months, Hardware hasn't. But my results have.

I did think about settings, but I hadn't touched a thing. I've reset back to factory to be safe, and I hadn't touched MicroAdjustment once, only when the issue appeared did I start playing. I was at a wedding and literally half way through a set I saw the focus issue appear. I have 4 photo's I took one after another (literally click click click click) where you can see it get worse. I'll try dig them out.

New test subject, this ones a little more life like. Hand held, point and shoot, same settings. Just to be more realistic with the fault - As i'd find it in real world scenarios. Same issue. Single Point good, Live View good, Multi AF bobbar (for you Yorkshire men out their).

Harry Multi Point AF by Rich Perkins, on Flickr
Harry Single Point AF by Rich Perkins, on Flickr
Harry Live View by Rich Perkins, on Flickr

Harry Multi Point AF II by Rich Perkins, on Flickr
Harry Single Point AF II by Rich Perkins, on Flickr
Harry Live View II by Rich Perkins, on Flickr

Best way to describe what i'm seeing, looks like I'm looking at Multi AF shots with sleep in my eyes.

[LongQ]

Rich,

I haven't read all of it yet but this article may be of interest.

http://photo.stackexchange.com/questions/43791/why...

Bear in mind you are really talking Zone Focusing with multiple points for your purpose not a selected Multi Point display area with selected (or automatic follow) single point focusing.