Velocity Reviews - Computer Hardware Reviews

Velocity Reviews > Newsgroups > Computing > Digital Photography > In-camera lens correction idea; myth or reality?

Reply
Thread Tools

In-camera lens correction idea; myth or reality?

 
 
RichA
Guest
Posts: n/a
 
      01-27-2012
The contention: That lenses that are meant to be used in conjunction
with software corrections are actually superior to those that aren't.
Here is (supposedly) why. The lens designer leaves out corrections to
the lens for things like geometric distortion and this allows
distortion to be corrected in software while freeing up the lens
design to achieve higher corrections in things like chromatic
aberration and spherical aberration than would be possible IF
distortion where also corrected optically.
IMO, this idea, floated on Dpreview, is nonsense because we know that
one of the in-camera software corrections IS for chromatic aberration,
at least on Panasonic cameras. Therefore there is no proof (and
absolutely no way to prove it because you'd need identical lenses, one
corrected in conjunction with software and one corrected
conventionally) that a lens that uses software correction for some of
its distortions shows any evidence of having higher correction in
other aberrations.
 
Reply With Quote
 
 
 
 
John A.
Guest
Posts: n/a
 
      01-27-2012
On Fri, 27 Jan 2012 14:25:37 -0800 (PST), RichA <(E-Mail Removed)>
wrote:

>The contention: That lenses that are meant to be used in conjunction
>with software corrections are actually superior to those that aren't.
>Here is (supposedly) why. The lens designer leaves out corrections to
>the lens for things like geometric distortion and this allows
>distortion to be corrected in software while freeing up the lens
>design to achieve higher corrections in things like chromatic
>aberration and spherical aberration than would be possible IF
>distortion where also corrected optically.


Assuming a lens design can either correct entirely (practically) for
one distortion or aberration at the expense of others, or make a
balanced compromise, I can see where the former might save you a bit
of time in post-processing since you'd have one less thing to correct
for. What the greater corrections needed for the other
distortions/abberations would do to your final image quality is
another question.

So, assuming the above is true, that you can eliminate the need for
one type of correction and thus eliminate a step in post, would there
be a price to pay in ultimate image quality? Or is this simply a great
work-flow optimization available to digital photographers who always
post-process anyway? Assuming it's true.

>IMO, this idea, floated on Dpreview, is nonsense because we know that
>one of the in-camera software corrections IS for chromatic aberration,
>at least on Panasonic cameras. Therefore there is no proof (and
>absolutely no way to prove it because you'd need identical lenses, one
>corrected in conjunction with software and one corrected
>conventionally) that a lens that uses software correction for some of
>its distortions shows any evidence of having higher correction in
>other aberrations.


If we're talking about interchangeable lenses, there's your answer:
the software has to be able to correct for whatever
(correction-compatible) lens you throw at it. If the particular lens
does not need a particular correction, that parameter will presumably
be zero. And if the correction code is well-enough-optimized it will
simply skip that correction in its calculations and maybe save a bit
of time & battery power.
 
Reply With Quote
 
 
 
 
David J Taylor
Guest
Posts: n/a
 
      01-28-2012
"RichA" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> The contention: That lenses that are meant to be used in conjunction
> with software corrections are actually superior to those that aren't.
> Here is (supposedly) why. The lens designer leaves out corrections to
> the lens for things like geometric distortion and this allows
> distortion to be corrected in software while freeing up the lens
> design to achieve higher corrections in things like chromatic
> aberration and spherical aberration than would be possible IF
> distortion where also corrected optically.
> IMO, this idea, floated on Dpreview, is nonsense because we know that
> one of the in-camera software corrections IS for chromatic aberration,
> at least on Panasonic cameras. Therefore there is no proof (and
> absolutely no way to prove it because you'd need identical lenses, one
> corrected in conjunction with software and one corrected
> conventionally) that a lens that uses software correction for some of
> its distortions shows any evidence of having higher correction in
> other aberrations.


Yes, it makes complete sense, as it gives the optical designer more
freedom. In similar vein, consider how when you design the lens and
sensor together you can get better performance, or a smaller, lighter,
less complex arrangement for the same quality. "Superior" may also mean
lower system cost for the same performance, as well as higher performance
for the same cost.

David

 
Reply With Quote
 
Bruce
Guest
Posts: n/a
 
      01-28-2012
"David J Taylor" <(E-Mail Removed)> wrote:
>"RichA" <(E-Mail Removed)> wrote in message
>news:(E-Mail Removed)...
>> The contention: That lenses that are meant to be used in conjunction
>> with software corrections are actually superior to those that aren't.
>> Here is (supposedly) why. The lens designer leaves out corrections to
>> the lens for things like geometric distortion and this allows
>> distortion to be corrected in software while freeing up the lens
>> design to achieve higher corrections in things like chromatic
>> aberration and spherical aberration than would be possible IF
>> distortion where also corrected optically.
>> IMO, this idea, floated on Dpreview, is nonsense because we know that
>> one of the in-camera software corrections IS for chromatic aberration,
>> at least on Panasonic cameras. Therefore there is no proof (and
>> absolutely no way to prove it because you'd need identical lenses, one
>> corrected in conjunction with software and one corrected
>> conventionally) that a lens that uses software correction for some of
>> its distortions shows any evidence of having higher correction in
>> other aberrations.

>
>Yes, it makes complete sense, as it gives the optical designer more
>freedom. In similar vein, consider how when you design the lens and
>sensor together you can get better performance, or a smaller, lighter,
>less complex arrangement for the same quality. "Superior" may also mean
>lower system cost for the same performance, as well as higher performance
>for the same cost.



You're right, but it took me some time to come to terms with that.
When I first tried m4/3 I was using a Panasonic GF1 with a 20mm f/1.7
lens. The results seemed fine, but I made the error of sitting in on
a bench test of that lens. The quite severe rectilinear distortion
came as something of a shock. Given my keen interest in architectural
photography, that was enough to put me off m4/3.

Now I am trying m4/3 for a second time, I have put my discomfort to
one side and just concentrate on the results. Using the same 20mm
f/1.7 lens on my G3 gives results that I am very happy with.

It pays not to think too much about how the software correction works.
It works well, and that's all that matters.

The correction works pretty well with zoom lenses too. Zooms usually
display a lot of barrel distortion at the wide end and tend to
pincushion distortion at the long end. With m4/3, the appropriate
amount of correction is applied throughout the zoom range giving very
low overall distortion.

The result is that m4/3 lenses tend to be small, light and cheap yet
they perform exceptionally well - all thanks to software correction.

I'm not sure I will stay with m4/3 in the long term. The Fujifilm
X-Pro 1 is very attractive and I look forward to a full range of
lenses including the all-important 35mm equivalent - actually a 23mm
f/2. There will be a Leica mirrorless system announced at Photokina
in September. Either of these top quality systems may be too tempting
to ignore.

But in the meantime, m4/3 is gradually winning me over. The cameras
and lenses are small, light and robust and the results are far better
than I expected.


 
Reply With Quote
 
David J Taylor
Guest
Posts: n/a
 
      01-28-2012
"Bruce" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
[]
> You're right, but it took me some time to come to terms with that.
> When I first tried m4/3 I was using a Panasonic GF1 with a 20mm f/1.7
> lens. The results seemed fine, but I made the error of sitting in on
> a bench test of that lens. The quite severe rectilinear distortion
> came as something of a shock. Given my keen interest in architectural
> photography, that was enough to put me off m4/3.
>
> Now I am trying m4/3 for a second time, I have put my discomfort to
> one side and just concentrate on the results. Using the same 20mm
> f/1.7 lens on my G3 gives results that I am very happy with.
>
> It pays not to think too much about how the software correction works.
> It works well, and that's all that matters.
>
> The correction works pretty well with zoom lenses too. Zooms usually
> display a lot of barrel distortion at the wide end and tend to
> pincushion distortion at the long end. With m4/3, the appropriate
> amount of correction is applied throughout the zoom range giving very
> low overall distortion.
>
> The result is that m4/3 lenses tend to be small, light and cheap yet
> they perform exceptionally well - all thanks to software correction.
>
> I'm not sure I will stay with m4/3 in the long term. The Fujifilm
> X-Pro 1 is very attractive and I look forward to a full range of
> lenses including the all-important 35mm equivalent - actually a 23mm
> f/2. There will be a Leica mirrorless system announced at Photokina
> in September. Either of these top quality systems may be too tempting
> to ignore.
>
> But in the meantime, m4/3 is gradually winning me over. The cameras
> and lenses are small, light and robust and the results are far better
> than I expected.


Thanks for those reports, Bruce. Appreciated. I am now think about what
might replace my DSLR in the future and, of those you mention, I would
rule out the Leica on cost grounds, and I like the idea of multiple
suppliers which the micro-4/3 systems offer. Fuji - I would have to wait
for the 28-300mm (equivalent) lens (which I know won't get the best from
the system).

With the advances in lens design brought about by computers, we are now
seeing the extra improvements brought about by image processing made
possible by fast processors in the camera.

Cheers,
David

 
Reply With Quote
 
Martin Brown
Guest
Posts: n/a
 
      01-28-2012
On 28/01/2012 08:24, David J Taylor wrote:
> "RichA" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
>> The contention: That lenses that are meant to be used in conjunction
>> with software corrections are actually superior to those that aren't.


They can be or they can be manufactured to give the same final image
quality more cheaply. This is fairly obvious if you stop to think about
it - the lens designer is freed to focus (excuse the pun) on only the
issues that cannot be corrected afterwards in software. Namely bringing
all the light from the subject to a focus at a flat perpendicular plane
irrespective of wavelength (or a best approximation to that ideal).

>> Here is (supposedly) why. The lens designer leaves out corrections to
>> the lens for things like geometric distortion and this allows
>> distortion to be corrected in software while freeing up the lens
>> design to achieve higher corrections in things like chromatic
>> aberration and spherical aberration than would be possible IF
>> distortion where also corrected optically.


There are two sorts of chromatic abberation in play.
http://en.wikipedia.org/wiki/Chromatic_aberration#Types

The longitudinal one which *has* to be corrected by the *lens* is
ensuring that all wavelengths come to focus in the same plane. Otherwise
all bright points are surrounded by a usually purple halo.

But lateral chromatic abberation can be easily corrected on digital
sensors (but not on film). It doesn't matter on a digital sensor if the
size of the image is slightly different in each of the red, green and
blue channels the final result can be rescaled in software. I have
previously demonstrated this on images posted here in the past.
(long before it was realistic to do it in-camera)

>> IMO, this idea, floated on Dpreview, is nonsense because we know that
>> one of the in-camera software corrections IS for chromatic aberration,
>> at least on Panasonic cameras. Therefore there is no proof (and
>> absolutely no way to prove it because you'd need identical lenses, one
>> corrected in conjunction with software and one corrected
>> conventionally) that a lens that uses software correction for some of
>> its distortions shows any evidence of having higher correction in
>> other aberrations.


It is obvious that if you have to match several hard constraints
simultaneously that it is much harder to obtain a design that works.
Reducing the number of constraints makes it easier to get a high
performance lens by moving certain of the geometrical corrections (scale
and pincushion/barrel distortion) into software.

The designer is then freed to make sure that the in focus point spread
function is as tight as possible and and in a flat perpendicular plane.
>
> Yes, it makes complete sense, as it gives the optical designer more
> freedom. In similar vein, consider how when you design the lens and
> sensor together you can get better performance, or a smaller, lighter,
> less complex arrangement for the same quality. "Superior" may also mean
> lower system cost for the same performance, as well as higher
> performance for the same cost.
>
> David


And the correction in firmware is a lot cheaper than polishing aspheric
surfaces and exotic low dispersion quirky to work with fluorite glass.

--
Regards,
Martin Brown
 
Reply With Quote
 
PeterN
Guest
Posts: n/a
 
      01-30-2012
On 1/30/2012 10:45 AM, bugbear wrote:
> RichA wrote:
>> The contention: That lenses that are meant to be used in conjunction
>> with software corrections are actually superior to those that aren't.
>> Here is (supposedly) why. The lens designer leaves out corrections to
>> the lens for things like geometric distortion and this allows
>> distortion to be corrected in software while freeing up the lens
>> design to achieve higher corrections in things like chromatic
>> aberration and spherical aberration than would be possible IF
>> distortion where also corrected optically.
>> IMO, this idea, floated on Dpreview, is nonsense because we know that
>> one of the in-camera software corrections IS for chromatic aberration,
>> at least on Panasonic cameras. Therefore there is no proof (and
>> absolutely no way to prove it because you'd need identical lenses, one
>> corrected in conjunction with software and one corrected
>> conventionally) that a lens that uses software correction for some of
>> its distortions shows any evidence of having higher correction in
>> other aberrations.

>
> You appear to be confusing the validity of an idea
> with the quality of a particular implementation of the idea.
>
> BugBear



Sounds more like he is confusing gobbledegook with reality.


--
Peter
 
Reply With Quote
 
Bruce
Guest
Posts: n/a
 
      01-30-2012
Bowser <(E-Mail Removed)> wrote:
>On Sat, 28 Jan 2012 10:00:31 +0000, Bruce <(E-Mail Removed)>
>wrote:
>>It pays not to think too much about how the software correction works.
>>It works well, and that's all that matters.
>>
>>The correction works pretty well with zoom lenses too. Zooms usually
>>display a lot of barrel distortion at the wide end and tend to
>>pincushion distortion at the long end. With m4/3, the appropriate
>>amount of correction is applied throughout the zoom range giving very
>>low overall distortion.
>>
>>The result is that m4/3 lenses tend to be small, light and cheap yet
>>they perform exceptionally well - all thanks to software correction.
>>
>>I'm not sure I will stay with m4/3 in the long term. The Fujifilm
>>X-Pro 1 is very attractive and I look forward to a full range of
>>lenses including the all-important 35mm equivalent - actually a 23mm
>>f/2. There will be a Leica mirrorless system announced at Photokina
>>in September. Either of these top quality systems may be too tempting
>>to ignore.
>>
>>But in the meantime, m4/3 is gradually winning me over. The cameras
>>and lenses are small, light and robust and the results are far better
>>than I expected.
>>
>>

>I came to the same conclusion with the GX1. The usual comparison to
>the NEX 5n showed me that the Panny is all but equal at lower
>resolutions, and about a stop behind further up in the scale. But the
>glass available on the m4/3 side convinced me to go that way. And
>while Sony keeps introducing new NEX bodies, the glass situation for
>native E mount lenses is pretty bad. And no, I refuse to use a clunky
>adapter to use good glass. If I could have a NEX body and sensor with
>m4/3 glass, it'd be perfect.



The situation with E mount lenses for NEX is getting desperate.
Although NEX cameras are selling well, Alpha DSLRs and SLTs are not.
So Sony has decided to cut back on lens development and a total of
nine lenses have been cancelled.

The result is that Alpha and NEX will be starved of good lenses for
the foreseeable future.

Meanwhile, Panasonic and Olympus are both investing heavily in new
lens designs.

 
Reply With Quote
 
 
 
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
One major downside possible with camera-based lens correction RichA Digital Photography 11 06-20-2010 02:29 PM
lens correction Peter Digital Photography 1 06-18-2010 12:35 PM
ACR Chrom. Abberation vs. 'Lens Correction' CA W (winhag) Digital Photography 3 12-05-2005 04:18 PM
pooled connection myth David McDivitt Java 42 03-18-2005 09:26 PM
ICMP myth. NNTP Cisco 11 05-24-2004 11:02 PM



Advertisments