PDA

View Full Version : Aperture and CA



Kirk Gittings
27-Apr-2007, 16:00
How does aperture size effect issues like barrel distortion and chromatic aberration?

Donald Qualls
27-Apr-2007, 19:19
Most lenses will have more aberrations (chromatic and spheric, even field curvature) at larger apertures, with the exception of modern lenses for 35 mm and medium format that are designed to be at their best wide open. For distortion, it depends very much on the lens, but if you have distortion, it'll usually be worse wide open than stopped well down.

David A. Goldfarb
27-Apr-2007, 19:30
Spheric aberration and field curvature are significantly improved by stopping down, but I don't know that chromatic aberration or barrel/pincushion distortion are. On the Verito, for instance, spheric aberration is substantially diminished by around f:8, but it's fairly easy to see chromatic aberration at much smaller apertures.

Barrel and pincushion distortion seem to be more a function of using the lens outside its optimum magnification range (for single focal length lenses that are well corrected for distortion at normal subject distances). I think I first noticed this with one of my classic Nikkor lenses for the Bronica S2a, where I was trying a macro shot with what was otherwise a very good lens, but it started to show a lot of pincushion distortion at magnifications around 1:2.

JW Dewdney
27-Apr-2007, 19:35
The way I see it... the more you stop down and approach 'pinhole', you're trading off one set of aberrations for another. While a pinhole has extreme softness due to diffraction - it should, technically, be geometrically perfect - as long as the film is flat.

Helen Bach
28-Apr-2007, 05:33
Well there's axial chromatic aberration (often referred to as simply 'chromatic aberration') and lateral chromatic aberration (also referred to as 'lateral colour'). Axial chromatic aberration (different wavelengths coming to focus at different distances from the lens) is reduced by stopping down, in common with the other aberrations that result in a spread patch of light (rather than a point) roughly centred at the right place in the plane of focus, so to speak.

In general, lateral chromatic aberration and distortion aren't affected by stopping down: they are aberrations that result in the image, or part of it, being in the 'wrong' place in the plane of focus. In the case of lateral chromatic aberration the lens has different effective focal lengths for different wavelengths, thus the image magnifications are different.

Best,
Helen

Robert A. Zeichner
28-Apr-2007, 06:53
I would have to believe that in reducing the number of marginal rays that reach the film which would be the result of using a smaller aperture, one would witness less circum-marginal chromatic aberration. Of course as another poster pointed out, as you reduce the size of the aperture, you would see an increase in diffraction artifacts that would soften the picture overall.

Kirk Gittings
28-Apr-2007, 07:17
Comparing good very wide lenses for 4x5 (75mm), 6x9 (47mm) and a Canon 24 TS (all of which I use for architecture at mid apertures). I know in terms of lens design that I am comparing apples and oranges a little bit, particularly with the Canon.

Why is there significantly more barrel distortion and CA as you move to smaller formats and equivalent focal lengths (more or less)?

David A. Goldfarb
28-Apr-2007, 07:31
Well, the Canon 24mm is a retrofocus lens, and if the 47mm is a lens of the most recent generation it probably is as well, but not so much as the 24mm in terms of percentage of focal length. That would probably have something to do with the distortion you're seeing (though I would expect a modern 47mm LF lens to have very low distortion).

Dan Fromm
28-Apr-2007, 08:38
Guys, I just went to my gurus. S. F. Ray, Applied Photographic Optics, 3d ed; and H. M. Brandt, The Photographic Lens.

Kirk asked about the dependence of distortion and chromatic aberration on relative aperture (the f/ number). Subject to my poor ability to read plain english, there's no mention of the effect of f/ number on distortion or chromatic aberration in either book.

Henry Ambrose
28-Apr-2007, 09:21
Why is there significantly more barrel distortion and CA as you move to smaller formats and equivalent focal lengths (more or less)?

I don't think this has to be the case but I do agree it is the case with most lenses that exist for SLRs. I put it off to how good the lenses have to be (can be) to sell at a given price. I think that if challenged to make a supremely excellent version of the TS lenses Canon could do it but the price might be high, maybe too high for the market Canon perceives. I'd like perfect wides for my Canon as well and I think Canon is mistaken about the number of excellent wides they would sell.

There are a number of wide lenses in 35mm that are practically perfect. Leica's 21, 24, 28 Aspericals and some of the Zeiss ZM lenses are examples of really great lenses in 35mm. All made for rangefinder cameras, they don't have the trade offs that have to be made for retrofocus design. They are compact and simple in mechanical construction without a bunch of moving parts, at least compared to an SLR lens like the Canon TS models and the lens elements can be located with high precision. The Leica lenses are very, very expensive, the Zeiss are moderately high priced, but even the little Voightlander 15mm is very low distortion at only $350. Or look at lense like the Mamiya 7 for examples of fine medium format glass.

I think that for the Canon TS lenses some good part of the cost has been expended in the tilt/shift mechanism so we end up comparing a $1000 SLR lens to a $1400 large format lens that has only to have its elements mounted in a barrel and screwed into a stock shutter. Throw in that the TS lenses are a low volume item coming from a high volume manufacturer and not likely a profitable as a consumer zoom for a point and shoot.

I'm also not convinced that all the CA we see in DSLR lenses is from the lens. I see much more CA like effects in scenes with high contrast edges that I don't think is explained by lens faults alone. I think this effect is aggravated by the sensor characteristics.

Kirk Gittings
28-Apr-2007, 09:21
For instance, I could shoot the same scene with the above mentioned lenses with say an extreme amount of rise movements and vertical building edges near the edge of the film. The 4x5 would require no BD or CA correction, the 6x9 will require some minor BD correction and the DSLR would require some moderate correction of both.

Kirk Gittings
28-Apr-2007, 09:31
Henry,

"I'm also not convinced that all the CA we see in DSLR lenses is from the lens. I see much more CA like effects in scenes with high contrast edges that I don't think is explained by lens faults alone. I think this effect is aggravated by the sensor characteristics."

Interesting, you are absolutely right, because I rarely saw CA when I shot 35mm slides with PC lenses.

On the Canon TS lenses, They certainly have their issues, yet they are entirely adequate, for many jobs, editorial for instance.

My issue with using fixed primes instead, is of course the extreme amount of PC you have to apply in PS at times AND After 30 years at this I can't see and compose with unperspective corrected images nor when shooting tethered with clients present (usually editorial but some ad agency people) do I want to show them uncorrected images or take the time to correct in PS.

Henry Ambrose
28-Apr-2007, 11:12
On the Canon TS lenses, They certainly have their issues, yet they are entirely adequate, for many jobs, editorial for instance.

My issue with using fixed primes instead, is of course the extreme amount of PC you have to apply in PS at times AND After 30 years at this I can't see and compose with unperspective corrected images nor when shooting tethered with clients present (usually editorial but some ad agency people) do I want to show them uncorrected images or take the time to correct in PS.

I agree completely, but the 24 TS is all we've got on a Canon or any other DSLR for shift. (unless maybe you can find an Olympus 24mm shift?) I use the 20 and 28 a good bit but I can't show the immediate results to a client. Usually I get to shoot, pick my selects, correct in PS and present them on the web. But you've got a big jump to convince someone to accept that workflow if they want to see whats happening right there on location. Maybe your two best-clients-forever would go for it, all the others would hire someone else pretty quick I think.

Also, the 20 and 28 have all the bad habits that the 24 has along with the 20 having lots of barrel distortion that's not totally correctible.

Another solution (that won't help with seeing the end result) is to use the 45TS and stitch. But that's another layer thats best avoided anyway.

I'd buy a GREAT 20-24mm shift lens for my Canon so quick it'd make their heads spin. In the meantime for anything that matters it'll get done on 4x5. But I do a lot of work where the wonderfulness of large format is wasted and a DSLR would be a great time and money saver.