Chromatic Aberration

[proseak]

But why is there so much more CA from digital cameras than analogue?

The reason I ask is because recently I scanned some negs including some high-contrast images, went to correct the CA...and didn't find any! In a sense, it's not that surprising - they were taken on a Contax, with the Zeiss Planar 50mm f1.7, which was widely regarded as one of the all-time great lenses...30 years ago.

There was just the dust to contend with

Scan settings? 6400 lpi, 16-bit Tiff

So, what am I missing?

Peter

[Dave Humphries]

So, what am I missing?

Hi Peter,

I think the answer is - a zoom lens.

Remember how rare and short zoom range they were, and how soft?

As you say, your example is a high quality prime, so any CA will have been designed out, it only becomes a bigger issue with zooms.

Cheers,

[arith]

I think I read somewhere you only get CA with digital but I can't remember the reason given unless it had something to do with diffraction which I doubt. It is certainly my biggest problem at the moment. cheers

[proseak]

"I think I read somewhere you only get CA with digital but I can't remember the reason given unless it had something to do with diffraction which I doubt."

Yes, that's my query...I do remember in the 60s(!) that there was a marked difference between "Colour" lenses, and those, inferior lenses only suited to B&W. At the time, colour was much more expensive than B&W and thus much rarer.

So, why has this old problem turned up again?

[Dave Humphries]

"I think I read somewhere you only get CA with digital but I can't remember the reason given unless it had something to do with diffraction which I doubt."

Yes, that's my query...I do remember in the 60s(!) that there was a marked difference between "Colour" lenses, and those, inferior lenses only suited to B&W. At the time, colour was much more expensive than B&W and thus much rarer.

So, why has this old problem turned up again?

It has always been there, we just haven't noticed!

Every definition confirms it is a lens problem, so it cannot be confined to digital sensor photography.

A mixture of other changes over the years contribute to the myth, for example (non-exhaustive list):
- In Black and White, CA just softened the image
- Black and White was more popular than colour
- In olden days, how often did you look at your colour work at the equivalent of "100%" so easily possible in digital?
- Most older lenses were softer and less contrasty in general, so less likely to show it
- Zoom lenses were far less common and I maintain that it can be minimised far easier on a prime lens
- Popular demand for internal focusing and the internal contra elements movements to achieve that AND extended zoom ranges make it difficult to control, without recourse to expensive aspheric grinding and various high refractive index glasses - even when these are employed, it just can't be perfect at all focal lengths

Well that's my 2 cents worth.


However, to perpetuate the myth - somebody please tell me I'm talking rubbish, 'cos I really don't believe this is relevant.
Most digital sensors have a bayer array layout for RGB pixels, so the pixels are not co-sited, as (arguably?) colour emulsion is. But then so is a Foveon digital sensor, the BRG pixels are vertically stacked - so someone show me some shots using same lens (with CA) on;
1) a bayer digital camera,
2) a foveon digital camera and
3) a film camera.

Cheers,

[peterb533]

Dave,

Good answer. It has always been there but as you say in the good old days black and white shots displayed the aberration as a softness.

Can't remember the exact physics but it has to do with different wavelengths of light resolving to slightly different point on the film/sensor plane. Shows up when there is large difference in values and especially at the edges of the frame where the lens has greater difficulty resolving the image.

Recently read 'The Camera' by Ansel Adams and he goes into this in some depth. Good book by the way but a little old fashioned now.

Peter

[proseak]

I've been browsing some negs, looking for something I could easily duplicate and came up with a picture of Gonville & Caius College here in Cambridge. These pics, the one whole-frame, and the other a pull-up, were shot on a Minolta XD7, with a Tamron 80-210 lens of similar vintage. The images aren't sharpened, merely colour-corrected.





Peter

[Dave Humphries]

I agree there's not much, if anything, to be seen there Peter.

It should be visible on the horizontal edges against the sky (not the spires) because this is a portrait orientation shot and the crop/pull up is from the centre of frame left/right.

Is it my eyes, or is there a very slight yellow edge above the ridge stone?

EDIT: Upon expanding the whole frame, there is definitely some red/green edges to the cylinderical street lamp on left hand side. I'm happy now (phew)

Cheers,

[proseak]

"Upon expanding the whole frame, there is definitely some red/green edges to the cylinderical street lamp on left hand side. I'm happy now (phew) "

Well, here's a screengrab, at 100%; the one on the left Is "colour-corrected", well, colour-altered is nearer the mark. So, looking to the original, no, I really don't think there is.

The reason I chose this shot was because I can go out and reshoot it on my Fuji 9600 tomorrow afternoon, in similar conditions. That'll be fun! The other reason is that it shows two cyclists, both of who are going in the right direction, and signalling. A rare event!



Peter

ps - there is a tiny amount onthe one that I "corrected". But it is tiny compared to what I've become accustomed to, and probably caused be my efforts!

[Dave Humphries]

Veeeery interesting,

I still think I can see it today

Here's a Vista screen snip at 100%:



Do you not see 'green' on the left and 'red' on the right of the lamp?
I think blowing it up too far dilutes it.

However, I do agree, it is far less than we're accustomed to with digital - I can't explain that, we need to see same lens on a digi-cam.

For completeness, here's the roof details I was looking at;

To be honest I am struggling to see any CA in these this morning, but they might be useful for comparison later.

Look forward to seeing the next shot, although if using your Fuji 9600, that's fixed lens isn't it? - and I know my Fuji S6500 has quite bad CA. We really need input from someone that can use the same (zoom?) lens on both digital and film camera bodies.

Cheers,

[arith]

The lens is the primary culprit for CA but you might not see CA if using the same lens on a film camera, but that is difficult to do unless you use some sort of converter which might corrupt the results.





Neither are cheap but one costs well over a thousand, the cheap one is actually better

[proseak]

"I think blowing it up too far dilutes it." - That wasn't a blowup, Dave - just a screen grab @100%

"Fuji 9600, that's fixed lens isn't it? - and I know my Fuji S6500" - Yes, I'm pretty sure that it's the same lens...some CA, some barrel @ W/A, but very good acutance. Yes, not a fair test in some ways; it's not truly like-to-like. I also plan to reshoot using my Tamron 29-80 on Kodachrome, but the results could take a while...

One question springs to mind - what is the physical depth of a modern sensor? With film, it's just that - a film, so there's no challenge to the depth-of-focus; it's either in focus or not....assuming that the various wavelengths "come in" at the same plane.

Hmm!

Peter

[peterb533]

One question springs to mind - what is the physical depth of a modern sensor? With film, it's just that - a film, so there's no challenge to the depth-of-focus; it's either in focus or not....assuming that the various wavelengths "come in" at the same plane.

Surely that's the point, they dont 'come in at the same place' unless the lens is constructed to correct the difference.

Also once you start scanning film all sorts of other variables may affect the digital representation of the image.

Peter

[proseak]

Well, my test reshoot was totally inconclusive. I had expected an overcast sky, but instead there was bright sunshine. There must be something odd about that spot - I was expecting some fairly pronounced fringing from the 9600...this what i got;



Hmm! I'm perplexed...

Regards

Peter

[AlanP]

Digital sensors are designed for light hitting them at 90 degrees. This is true at the center of the image but not the edges. This creates chromatic aberation in addition to the aberation created by the lense. Four third lenses are telecentric, they have more light hitting the sensor at 90 degrees to prevent the chromatic aberation.

[arith]

So maybe sensors should not be flat, I really don't know what is going on with telecentric lenses, yet. Never heard of it.