7D Chromatic Aberration

[esteah]

I have a Canon 7D with a EFS 15-85mm lens and I am getting what I think is excessive chromatic aberration with this lens. I ran some tests and I see this when comparing my own test results to similar tests that I found on-line.
I need some advise - I am expecting too much from this camera/lens? I can understand that there should be some chromatic aberration, but it seems as if that portion of the image is blurred as well.
Are there any other tests that I can try?
Please advise - I have attached 3 snags from my experiments....I would appreciate some feedback - It's driving me nuts!

Test 1 - Top left corner of test image


Test 2 - Top left corner of test image


Test 3 - Center of test image

[nickjohnson]

Hello Andre,
Most lenses have some Chroma - some more than others. Zoom lenses of the sort that you ask about – wide to telephoto perspective – are particularly prone, it's part of the engineering compromises that have to be made.

I find that most Chroma I see in my raw images can be corrected by raw image processing software. I use DXO to automatically remove Chroma. There are other software applications that also do as good – or better – job.

That all said, I'm not sure that what you show in your post is Chroma. I notice that the shutter speeds are quite slow. Was the camera on a tripod? I can not remember seeing Chroma from a dark edge against a mid tone background – per your post. Usually I see it in dark tree branches against a bright sky – or similar. I wonder if what we are seeing in your post is a combination of printing ink bleed into a rather porous base material, combined with a little bit of focus not being quite right and / or camera shake?

Regards,

Nick.

[Wirefox]

I think Nick is right. This does not look like CA but shake. You probably also need to ask yourself if you will be pixel peeping to this extent on all your images CA will be most noticeable on edges of darker subjects shot against bright backgrounds. Tree branches against a bright sky will bring out CA to its full. That may be a better experiment in my opinion. i would certainly advise that you forget the technical minutiae of lens performance a get shooting what pleases you. If, when viewed at normal resolutions you are still seeing irregularities then perhaps that lens is not best for you. It is not the camera that is at fault by the way.

[ktuli]

All,

I was also curious... is CA restricted to lenses? Or can it actually be a product of the camera itself? What I mean, is would the amount of CA vary with the same lens on different bodies?

Thanks!

- Bill

[Nass]

We probably need a bit more info, are the top two 1:1? Now you did have the lens wide open which will never give its best performance, what light was it shot under? Can you do a wide open tree branch test in sunlight and post it? Yes I can see some colour bleed of sorts, red on the left, green on the right (have a look at the big Ps). Intriguing, as my CA is usually purple =)

[Black Pearl]

Firstly - have your printed any of these shots and if so can you see the CA's? - If you haven't then please do so and report back because I'll bet they don't show and simply aren't an issue.
Secondly - your first shot is wide open and you've not a cat in hells chance of the corners being be pretty.
Thirdly - the third shot is not too bad if you looked at it in a realistic way....in a print....but yes if you zoom in to a magnification that equates to a ten foot print then yes you will see faults. Are you going to print it ten foot though?

Yes I know we are able to view a digital image at ridiculous magnifications these days and yes I know we look for faults in the resulting images but seriously it just doesn't matter in practical terms. Your 15-85mm is a budget zoom - not a 'kit' one but not a Pro one either - so is likely to show some budget imposed limitations. That said it has a very usable range, has a good IS unit, is sharp for normal use and didn't cost you a fortune.

Go out and shoot some real world pictures, process them the way you like and print them out..............please don't pixel peep them, its unhealthy.

[Wirefox]

I was also curious... is CA restricted to lenses? Or can it actually be a product of the camera itself? What I mean, is would the amount of CA vary with the same lens on different bodies?

It s a good question Bill. My understanding that CA is a symptom or product of the configuration/quantity/quality of the glass within the lens tube itself. I too would appreciate input from some of our technical sages

[Colin Southern]

Hi Andre,

Out of curiocity, what sharpening workflow are you applying?

[Black Pearl]

I was also curious... is CA restricted to lenses? Or can it actually be a product of the camera itself? What I mean, is would the amount of CA vary with the same lens on different bodies?

Technically its a lens error where the separate red, green, blue rays of light are not focused at the same exact point so form a blurring/fringing of colours.


The problem is that different manufactures and even different bodies within the range deal with CA's in differing (lots of diffs in there) ways so the results will end up......different. Most modern DSLR's use a sensor level algorithm to reduce or even remove CA's from an image. If you shoot jpegs you will rarely if ever see these manifest themselves in your images, if on the other hand you shoot RAW then you need to deal with them in the RAW converting software.

[Wirefox]

Thanks Robin

Good explanation, so essentially the reduction of CA is the lens and the camera working together. In that case it is quite possible that sticking that lens on another body would give different results. I do not think we are seeing CA here though. Young Colin may be onto something though

[tclune]

There is an excellent article on CA here: http://toothwalker.org/optics/chromatic.html The most important thing to understand about CA is that it comes in two flavors -- lateral CA (called transverse CA in the referenced article) and longitudinal CA (LoCA, called axial CA in the referenced article.) In lateral CA, the lens focuses all light frequencies on the same plane, but the different colors may be out of alignment on the focal plane. As a result, you may see one color on one side of a high-contrast dark bar and a different color on the other side. Lateral CA is visible on the edges of the image, but should not be present in the middle of the frame Modern camera bodies will often remove lateral CA automatically for you, and this kind of CA does not impact image sharpness once removed. As a result, many modern lenses seem to optimize their design in a way that the design compromises make lateral CA more pronounced. Folks with older camera bodies may find some new lenses truly awful, while folks with newer cameras think they are wonderful. The Nikon 35mm f/1.8 is this sort of a lens.

With LoCA, the different light frequencies are in focus on different planes. This kind of CA looks like a purple halo around a solid object, and can appear anywhere in the image frame. AFAIK, no camera can automatically correct for LoCA. It is a real chore to correct for in PP. Further, it necessarily limits the sharpness of the image. LoCA, unlike lateral CA, goes down when you shut down your aperture, and is typically a problem of bright glass. It is almost always gone when you stop down one or two stops. If you examine the CA diagram for a lens, LoCA shows up as a large amount of CA wide open that drops down to a minimum after one or two stops. The other CA shown in CA diagrams is lateral CA. So those CA diagrams that you've always ignored on SlrGear tell you everything you need to know about your lens' performance in this regard before you ever buy the lens. FWIW

[esteah]

Thanks to everyone for responding.
The test that I did was indeed in tungsten light with the camera just sitting on a pile of books with mirror lock-up enabled. I used the timer to fire the shots. The screen shots are zoomed in 1:1.
Answer to Colin: No sharpening has been applied to the images - only adjusted slightly in the camera settings (if that is what you meant)
Answer to Robin: I really appreciate the advise and I am probably being be too critical - pixel peeping 'n all that.
I understand that the CA will be most prominent in the first sample (above) but I was surprised at the amount of blur in the corners and at the edges laterally from the center of the image at this aperture.
I have included another example of this. These 2 images were taken with the camera on a tripod - pointing at a PC board laying on the floor. The screen shots are from the same image just displaying the difference between the edge and the center of the of the image.

This is a 1:1 zoom at the center of the image.


This is a 1:1 zoom to the extreme left of the same image.


This is the image that these snags were taken from (Slightly reduced).


Surely this amount of blur is not normal - even for my budget equipment? This blur is especially irritating when I capture macro images or flowers etc. Do you think I have a case to take back to my camera supplier?

Please advise...(Thanks again for all the help!)

[nickjohnson]

Andre,
One of the difficulties I find in attempting to help folks on a forum like CiC is “pitching” an answer in a way that is both helpful, non patronising and non judgemental.

Your second set of example images shows exactly what I would expect given the information in your post. I very much doubt that what we are seeing here is Chroma. Yes Chroma may be an eliment in the mix. IMHO what we are seeing is general lens aberration of a type and magnitude that is entirely to be expected under the circumstances.

Almost no retail photographic lenses are optimised for photographing esentialy flat objects at close range. The vast majority – including yours – are optomised for three dimensional objects at moderate to far distances. Almost no retail photographic lenses will produce anywhere near as good an edge performance as they do for the image centre. I know of none at all that will do so at maximum aperture. Given the level of magnification you are using – and how that might impact your expected circle of confusion – you may like to calculate the necessary depth of field you need to render both the centre and the edge of your image “sharp”. Keeping in mind that the edge of your subject is further from the image plane than the centre. Having done that you may wish to ponder if that definition of “sharp” is physically possible at f5.6. In any event I suggest that ensuring that the camera being perpendicular to the subject is of paramount importance with the type of image you show in your examples.

Your particular lens example may yet prove to be faulty. That said, it's a good, popular lens (deservingly so IMHO) and you have it mounted on a super camera. Please can we see some pictures of your part of the world?

HTH

Regards,

Nick.​

[esteah]

Thanks very much to Nick, Tom & Robin and all others that contributed to my query.
I suppose I am being too much of a perfectionist. I am very with what I have in general and I should really stop the "pixel peeping" and just get on with it!
Having said that, it boils down to one thing, really, and that is not only knowing your instrument, but understanding the limitations of your equipment and then operating to the best of your ability within those parameters.
Thanks again and thanks for a great forum...