Correction of spherical aberration in fast lenses

RichA wrote:
There is no doubt molded aspherics have helped, the original Nikon
Noct Nikkors and Leica Noctilux lenses had expensive hand-ground
aspherical elements to control SA, but even the best lenses wide open
exhibit spherical aberration. You can buy SA removal
"plates" (basically, single-element lenses that add spherical
aberration of an opposite character in order to cancel out residual SA
of the lens's wavefront. However, those plates (from outfits like
Edmund Optics) only supply up to about 1 wave of correction and the SA
exhibited by modern normal focal length lenses is far worse than
that. In addition, single-element lenses if they do cancel-out SA
would probably add chromatic aberration but that (in moderation) can
be dealt with in software. It should be possible to find suitable off-
the-shelf achromats that could be used between lens and sensor) to
cancel-out some of the residual SA on these fast lenses. It could
mean "new life" for otherwise good and fast legacy lenses.

If you're happy to correct chromatic aberration in software,
why not do the SA at the same time?

As demonstrated by Hugin, multiple corrections can be folded
together, so you only incur one interpolation problem.

BugBear

RichA wrote:
On Apr 20, 4:31 am, wrote:
RichA wrote:
There is no doubt molded aspherics have helped, the original Nikon
Noct Nikkors and Leica Noctilux lenses had expensive hand-ground
aspherical elements to control SA, but even the best lenses wide open
exhibit spherical aberration. You can buy SA removal
"plates" (basically, single-element lenses that add spherical
aberration of an opposite character in order to cancel out residual SA
of the lens's wavefront. However, those plates (from outfits like
Edmund Optics) only supply up to about 1 wave of correction and the SA
exhibited by modern normal focal length lenses is far worse than
that. In addition, single-element lenses if they do cancel-out SA
would probably add chromatic aberration but that (in moderation) can
be dealt with in software. It should be possible to find suitable off-
the-shelf achromats that could be used between lens and sensor) to
cancel-out some of the residual SA on these fast lenses. It could
mean "new life" for otherwise good and fast legacy lenses.

If you're happy to correct chromatic aberration in software,
why not do the SA at the same time?

As demonstrated by Hugin, multiple corrections can be folded
together, so you only incur one interpolation problem.

BugBear

Doing both requires actually designing a lens specific to the task and
thats a job for the lens manufacturers.

Excuse me? I correct both (SA and CA), routinely, in Hugin, with a standard lens
on my camera.

BugBear

On 4/20/2011 3:31 AM, bugbear wrote:

If you're happy to correct chromatic aberration in software,
why not do the SA at the same time?


Because it is an entirely different problem.

What you are calling "chromatic aberrateion" is actually "LATERAL
chromatic aberration". Thus, to first order the three channels
simply have different magnification, and the blurring
is second order. That's why software correction works so well.

But spherical is a different matter. There is no way to
separate the different focal zones fore software correction.
All you can do is use a "sharpening" algorithm. This increases
noise a lot.

However, with modern molded aspheric lenses any lens
should have zero spherical aberration to all orders
at some one wavelength (unless it is intentionally used to
partially balance out some other worse problem off axis).
What is left of course is that the spherical aberration
is left at other wavelengths.

Doug McDonald

Doug McDonald wrote:
On 4/20/2011 3:31 AM, bugbear wrote:

If you're happy to correct chromatic aberration in software,
why not do the SA at the same time?


Because it is an entirely different problem.

What you are calling "chromatic aberrateion" is actually "LATERAL
chromatic aberration". Thus, to first order the three channels
simply have different magnification, and the blurring
is second order. That's why software correction works so well.

But spherical is a different matter. There is no way to
separate the different focal zones fore software correction.
All you can do is use a "sharpening" algorithm. This increases
noise a lot.

However, with modern molded aspheric lenses any lens
should have zero spherical aberration to all orders
at some one wavelength (unless it is intentionally used to
partially balance out some other worse problem off axis).
What is left of course is that the spherical aberration
is left at other wavelengths.

He's probably thinking of pincushion/barrel distortion, and possibly
Rich is too. 'Spherical aberration' is where different colors mis-focus
due to fast spherical lens designs, resulting in soft (corners?);
different colors will be out of focus, not just mis-aligned like lateral
chromatic aberration. Positive spherical aberration will actually
improve the softness of background OOF/bokeh areas but mess up the
foreground OOF. Negative SA has the opposite effect.

Rich, what exact correction lenses are you thinking of?

Paul Furman wrote:
Doug McDonald wrote:
On 4/20/2011 3:31 AM, bugbear wrote:

If you're happy to correct chromatic aberration in software,
why not do the SA at the same time?


Because it is an entirely different problem.

What you are calling "chromatic aberrateion" is actually "LATERAL
chromatic aberration". Thus, to first order the three channels
simply have different magnification, and the blurring
is second order. That's why software correction works so well.

But spherical is a different matter. There is no way to
separate the different focal zones fore software correction.
All you can do is use a "sharpening" algorithm. This increases
noise a lot.

However, with modern molded aspheric lenses any lens
should have zero spherical aberration to all orders
at some one wavelength (unless it is intentionally used to
partially balance out some other worse problem off axis).
What is left of course is that the spherical aberration
is left at other wavelengths.

He's probably thinking of pincushion/barrel distortion, and possibly
Rich is too. 'Spherical aberration' is where different colors mis-focus
due to fast spherical lens designs, resulting in soft (corners?);
different colors will be out of focus, not just mis-aligned like lateral
chromatic aberration. Positive spherical aberration will actually
improve the softness of background OOF/bokeh areas but mess up the
foreground OOF. Negative SA has the opposite effect.

Rich, what exact correction lenses are you thinking of?


Dunno about Rich, but you got me bang to rights. I was indeed thinking
of pincushion/barrel distortion.

Brief googling shows a lovely example of the corrections I was thinking
of.

http://www.camerasunderwater.info/op...nscorrect.html

BugBear

On 4/20/2011 10:22 AM, Paul Furman wrote:


He's probably thinking of pincushion/barrel distortion, and possibly Rich is too. 'Spherical
aberration' is where different colors mis-focus due to fast spherical lens designs, resulting in
soft (corners?);

No. Spherical aberration is when light hitting the lens EXACTLY ON AXIS
at different distances from the center focus at different focal lengths.


different colors will be out of focus,


No, that's CHROMATIC ABERRATION (regular, on-axis chromatic aberration.)

Doug

On Wed, 20 Apr 2011 09:21:35 -0500, Doug McDonald
wrote:

On 4/20/2011 3:31 AM, bugbear wrote:

If you're happy to correct chromatic aberration in software,
why not do the SA at the same time?


Because it is an entirely different problem.

What you are calling "chromatic aberrateion" is actually "LATERAL
chromatic aberration". Thus, to first order the three channels
simply have different magnification, and the blurring
is second order. That's why software correction works so well.

But spherical is a different matter. There is no way to
separate the different focal zones fore software correction.
All you can do is use a "sharpening" algorithm. This increases
noise a lot.

However, with modern molded aspheric lenses any lens
should have zero spherical aberration to all orders
at some one wavelength (unless it is intentionally used to
partially balance out some other worse problem off axis).
What is left of course is that the spherical aberration
is left at other wavelengths.

Doug McDonald


Reading the thread got me to wondering about longitudinal CA (I hope
that's a correct term.)

Would focus stacking, handled with each color separately work to
minimize this? By each color, I mean the RGB, not all possible
colors, which may be why the idea won't work.

Just wondering.

In rec.photo.digital Doug McDonald wrote:
On 4/20/2011 10:22 AM, Paul Furman wrote:

He's probably thinking of pincushion/barrel distortion, and possibly Rich is too. 'Spherical
aberration' is where different colors mis-focus due to fast spherical lens designs, resulting in
soft (corners?);

No. Spherical aberration is when light hitting the lens EXACTLY ON AXIS
at different distances from the center focus at different focal lengths.

Exactly. It's called spherical aberration because it's the result of
using lenses whose curvature is part of a sphere. The perfect lens
shape for accurate sharp focus is not speherical. It's just a rough
approximation to a sphere which is best at long focal lengths and
small apertures. But until recently it was extraordinarily costly and
difficult to make aspherical lenses.

It's the reason for the notorious aperture-related focus shift in wide
aperture 50mm lenses. As you widen the aperture the best compromise
average best focus position over the whole amount of the spherical
lens surface shifts. So if your DSLR's AF operates at a nominal f5.6
and you're shooting at f1.4...

--
Chris Malcolm

On 2014-06-28 00:00:57 +0000, RichA said:

On Wednesday, April 20, 2011 11:22:05 AM UTC-4, Paul Furman wrote:
Doug McDonald wrote:
On 4/20/2011 3:31 AM, bugbear wrote:

If you're happy to correct chromatic aberration in software,
why not do the SA at the same time?


Because it is an entirely different problem.

What you are calling "chromatic aberrateion" is actually "LATERAL
chromatic aberration". Thus, to first order the three channels
simply have different magnification, and the blurring
is second order. That's why software correction works so well.

But spherical is a different matter. There is no way to
separate the different focal zones fore software correction.
All you can do is use a "sharpening" algorithm. This increases
noise a lot.

However, with modern molded aspheric lenses any lens
should have zero spherical aberration to all orders
at some one wavelength (unless it is intentionally used to
partially balance out some other worse problem off axis).
What is left of course is that the spherical aberration
is left at other wavelengths.

He's probably thinking of pincushion/barrel distortion, and possibly
Rich is too. 'Spherical aberration' is where different colors mis-focus
due to fast spherical lens designs, resulting in soft (corners?);
different colors will be out of focus, not just mis-aligned like lateral
chromatic aberration. Positive spherical aberration will actually
improve the softness of background OOF/bokeh areas but mess up the
foreground OOF. Negative SA has the opposite effect.

Rich, what exact correction lenses are you thinking of?

Aspherics are made to correct spherical aberration. Distortion should
be way down on anyone's list of problems to correct.

How the Hell did you manage to dig up a 2011 post from Paul Furman?
He hasn't been active in the photo NGs for 2+ years.

--
Regards,

Savageduck

On 6/27/2014 8:09 PM, Savageduck wrote:
On 2014-06-28 00:00:57 +0000, RichA said:

On Wednesday, April 20, 2011 11:22:05 AM UTC-4, Paul Furman wrote:
Doug McDonald wrote:
On 4/20/2011 3:31 AM, bugbear wrote:

If you're happy to correct chromatic aberration in software,
why not do the SA at the same time?


Because it is an entirely different problem.

What you are calling "chromatic aberrateion" is actually "LATERAL
chromatic aberration". Thus, to first order the three channels
simply have different magnification, and the blurring
is second order. That's why software correction works so well.

But spherical is a different matter. There is no way to
separate the different focal zones fore software correction.
All you can do is use a "sharpening" algorithm. This increases
noise a lot.

However, with modern molded aspheric lenses any lens
should have zero spherical aberration to all orders
at some one wavelength (unless it is intentionally used to
partially balance out some other worse problem off axis).
What is left of course is that the spherical aberration
is left at other wavelengths.

He's probably thinking of pincushion/barrel distortion, and possibly
Rich is too. 'Spherical aberration' is where different colors mis-focus
due to fast spherical lens designs, resulting in soft (corners?);
different colors will be out of focus, not just mis-aligned like lateral
chromatic aberration. Positive spherical aberration will actually
improve the softness of background OOF/bokeh areas but mess up the
foreground OOF. Negative SA has the opposite effect.

Rich, what exact correction lenses are you thinking of?

Aspherics are made to correct spherical aberration. Distortion should
be way down on anyone's list of problems to correct.

How the Hell did you manage to dig up a 2011 post from Paul Furman?
He hasn't been active in the photo NGs for 2+ years.


He was bored?


--
PeterN