Extension tube causing CA??

[rick55]

I got a 12mm extension tube (from Canon), and just got my good lens back (fixed under warranty).

So I went out to try some flowers with off-camera flash at dusk, see what I could get.

I was shocked to see some really nasty CA at one corner. I don't normally have any problems with this lens (EF-S 17-55). I'm guessing that putting it out on an extension tube means that all the high-tech computer calculations they do for the lens get screwed up?

Here's the full frame, no PP except capture sharpening:



and a crop of the upper-right:



You can really see it on the edge between the last white petal and the dark background, also some between the nearer and farther petals.

So, do I just try to center and crop with the extension tube? PP it out? Or do I have to spring for a macro lens?

Cheers,
Rick

[arith]

Now that is interesting; I've never used them but it seems to have exaggerated differences in focus or dispersion that is CA.

[Dave Humphries]

I was shocked to see some really nasty CA at one corner. I don't normally have any problems with this lens (EF-S 17-55). I'm guessing that putting it out on an extension tube means that all the high-tech computer calculations they do for the lens get screwed up?

You can really see it on the edge between the last white petal and the dark background, also some between the nearer and farther petals.

So, do I just try to center and crop with the extension tube? PP it out? Or do I have to spring for a macro lens?

Hi Rick,

If CA on digital sensors were caused by the use of non-telecentric lenses I was reading about elsewhere here, the use of an extension tube should improve it, not make it worse. I believe this to be so because the tube moves the lens away from the sensor and the reduction in exposure is caused by the wasted image around the edges, this implies we're only using the middle of the lens, so CA should improve (shouldn't it?).

For it to be definitively proven as CA, I'd like to see a white-black and black-white edge pairs (near the edge of frame) and if CA that would have opposite colours either side of the black bar. This might show it.



But the yellow, if present, is very subdued - I suppose it is possible the WB has neutralised it. This is a vertical transition that isn't so near the top edge of frame though. Not sure I'm explaining myself too well.

Try photographing a postcard (6 in x 4 in white card) which has a strong black border drawn near the edge, then we can look at either side of the black line (if contrasty enough).

This might be a weird over-exposure issue on the sensor?
Although I now see there are red/blue edges in bottom left corner

I'm guessing that putting it out on an extension tube means that all the high-tech computer calculations they do for the lens get screwed up?

I think you're right and that may be why some lenses are better than others on tubes.

Not sure that helps

[Andy]

Yes. Extension tubes do not add any new abberations but they do magnify any that are there.

Canon fit 50mm lenses seem to work well on tubes but 12mm isn't going to do that much. You'd need something like 50mm of tubes to get the kind of magnification a macro lens can give. A macro lens (like the EF-S 60mm?) should give you more space between you and the subject which could make lighting the subject easier too.

[Andy]

Also, the extension tube means you get a crop from an enlarged image circle - the edges of the frame won't be as near the edge of the image circle when using tubes. Vignetting shouldn't be a problem!

[Dave Humphries]

A macro lens (like the EF-S 60mm?) should give you more space between you and the subject which could make lighting the subject easier too.

I have to agree with this
Since (recently) using a (manual) macro lens compared to (manual) extension tubes, the ease of use of a macro lens cannot be under-estimated.

Cheers,

[arith]

Hi Dave; what confuses me is the fact my 28mm is sharper wide open and CA which starts at 1 pixel gets progressively worse with higher f numbers. Now the 17-40mm f4L is 0.5 pixel CA but over 4% distortion so you gain on the swings but lose on the roundabout.

A much more expensive more recent L type is even worse and also has the odd behaviour exhibited by my prime.

I have found a better prime, but non of these are made anymore. The 20mm f2.8 is a brilliant but very expensive for its age lens.

So I don't know what is going on here, an optician didn't know either since I have to have my eyes photographed I thought I would ask. Underexposing shouldn't affect dispersion but raising fill light leaves highlights and maybe there is mileage in that but at the expense of a loss of some definition.

I'm still thinking about how to take a pic with the sun in without any CA or flare.

[rick55]

If CA on digital sensors were caused by the use of non-telecentric lenses I was reading about elsewhere here, the use of an extension tube should improve it, not make it worse. I believe this to be so because the tube moves the lens away from the sensor and the reduction in exposure is caused by the wasted image around the edges, this implies we're only using the middle of the lens, so CA should improve (shouldn't it?).

Thanks, Dave, that's good information, a good point.

Try photographing a postcard (6 in x 4 in white card) which has a strong black border drawn near the edge, then we can look at either side of the black line (if contrasty enough).

I'll definitely have to do some testing. And you're right, it may be more at the sensor. It's an edge that's near max white (RGB around 220, 220, 230), and near black behind (35, 35, 35,).

Yes. Extension tubes do not add any new abberations but they do magnify any that are there.

Thanks, Andy. Another reason to do testing, with and without the tube.

Canon fit 50mm lenses seem to work well on tubes but 12mm isn't going to do that much. You'd need something like 50mm of tubes to get the kind of magnification a macro lens can give. A macro lens (like the EF-S 60mm?) should give you more space between you and the subject which could make lighting the subject easier too.

That's a good point, too: it would also help with my current love of photographing water. I've discovered that with the nose of the lens pretty close to the tray, I spend all my time wiping off splashes.

Cheers,
Rick

[rick55]

Hi Dave; what confuses me is the fact my 28mm is sharper wide open and CA which starts at 1 pixel gets progressively worse with higher f numbers. Now the 17-40mm f4L is 0.5 pixel CA but over 4% distortion so you gain on the swings but lose on the roundabout.

A much more expensive more recent L type is even worse and also has the odd behaviour exhibited by my prime.

Hi, Steve;

Given the comments above, that an extension tube normally decreases CA, since it moves the corners out of the sensor's view, and the fact that you're seeing contrary behavior, I wonder if what we're both seeing is sensor behavior. I think this this was discussed in another thread, about why digital has more CA, and I think, in general it was agreed that it doesn't. But I wonder if there are some circumstances of exposure, or some combinations of lens and sensor, that trigger the issue. I'll certainly post the results of testing I do with the tube: maybe they'll provide some insight for us all.

Cheers,
Rick