Digital BW, The Print

A week or so ago we were discussing the various published tests and 
how well (or not) they seemed to predict the performance of lenses or 
agree with our own evaluations.

Using my Canon 8 mp Rebel, I tested a number of lenses I had that 
were also tested by some magazines (Popular Photo in particular), 
Photodo, and Photozone.  In comparing the results of my tests, which 
use standard resolution charts as well as printed pages with assorted 
print sizes and shapes, I must say the only tests that seem to 
correlate well are the Photozone tests.  

See http://photozone.de/

In fact, their tests, which use a similar digital camera, give almost 
identical relative results.  This is probably becuase our test 
routines are similar.  On the other hand, the Photodo tests, which I 
used to think were about the best with respect to film cameras, have 
a much lower correlation with what I see with the Rebel.  I suspect 
part of the difference is the distinction between digital and film 
cameras.

At any rate, I'm impressed with Photozone, although I hope they start 
testing on full frame Canons also.

Paul
www.PaulRoark.com

pr_roark wrote:

> At any rate, I'm impressed with Photozone, although I hope they start 
> testing on full frame Canons also.

Part of their problem is that they like to keep correlation 
between all the lens test results so the relative old Canon 
350D is still used, even for the latest lenses.
So FF but then with a pixel density that is the same to the 
pixel density of the Canon 350D = 3.1ML/CM2 and the 5D MII 
is  on 2.4 right now so we have to wait another year. Pixel 
density isn't telling all as the pixel wells grow in size 
and quality per generation even with the same pixel density.

DPreview faces the same problem: Andy Westlake's comment on 
that issue somewhere in the middle of this page:
http://forums.dpreview.com/forums/readflat.asp?forum=1000&thread=28994243&page=2

ColorFoto has less compatibility between their lens test 
results but tested more lenses on different sensors 
including the most recent ones. They also lay more emphasis 
on low contrast resolution in their latest test methods so 
you have to check the test number next to the results: 
Testverfahren 1.3-1.4-1.5.

-- 
Met vriendelijke groeten,  Ernst


|  Dinkla Grafische Techniek  |
|     www.pigment-print.com    |
|             ( unvollendet )            |

Ernst Dinkla wrote:
>
> pr_roark wrote:
> 
> > At any rate, I'm impressed with Photozone, although 
> > I hope they start 
> > testing on full frame Canons also.
> 
> Part of their problem is that they like to keep correlation 
> between all the lens test results so the relative old Canon 
> 350D is still used, even for the latest lenses. ...

I would think that consistency would be most important for lens 
tests.  So, using the same camera body -- even if old -- makes a lot 
of sense.  The question then is whether the tests remain relevant for 
the newer, better cameras.

As the cameras get better, I'd think we'd see the differences among 
lenses magnified, but I wonder if the relative performance changes 
much.  The Photozone tests, as well as my own using the 8 mp Rebel, 
can see some of the results of diffraction limits in the center from 
f5.6 to f8 (and more as we go to smaller f-stops).  So, even with 
older camera designs we're seeing some changes (for example, f5.6 to 
8 central sharpness) that are probably not going to be relevant to 
actual shooting.  I don't hear many photographers complain of f8 
diffraction being a serious limit.

As a practical matter, the lenses I was looking at are all very good 
in the center at f5.6 - 11.    

The edges and larger larger apertures are where the action is, and 
there the image quality can look bad even with the older cameras.  
I'd guess it's just going to be that much more worse compared to the 
central image at f8 with the newer cameras. 

On the other hand, I wonder if some of the color fringing at the 
edges is due to sensor design as well as lens aberrations.  I don't 
recall the color fringing being such a big deal with film.  The Canon 
90 TS is the only lens I have that doesn't show that annoying defect 
on both the Photozone tests and mine.  If fringing -- particularly 
with wide angle lenses -- is, in part, a camera defect, they we might 
have some changes that alter relative lens performance.

Paul
www.PaulRoark.com

pr_roark wrote:

> On the other hand, I wonder if some of the color fringing at the 
> edges is due to sensor design as well as lens aberrations.  I don't 
> recall the color fringing being such a big deal with film.  The Canon 
> 90 TS is the only lens I have that doesn't show that annoying defect 
> on both the Photozone tests and mine.  If fringing -- particularly 
> with wide angle lenses -- is, in part, a camera defect, they we might 
> have some changes that alter relative lens performance.

And I thought for some time that as color fringing can be 
removed with software the lens designers compromise it more 
in order to get other specs in line. Tokina for example in 
its recent designs as tested by Photozone. An interesting 
comment in the conclusion of DPreviews test report of the 
Fuji S100FS:

"There is a cloud on the horizon, though, and its a cloud 
with oddly-colored edges. Chromatic aberration (CA) is a 
major problem at the focal lengths that are likely to be 
most used. The effect is so pronounced at times that it is 
visible even in small prints. The frustrating thing is that 
CA can be removed automatically - we've seen it done in 
other long-zoom cameras, so know it's possible, 
economically. It can also be manually removed from RAW files 
but the supplied FinePix Studio software doesn't provide the 
tools to do, so again you're pushed towards using 
third-party software that don't do the images justice."

If reviewers accept that approach then it is likely to 
become the standard.


-- 
Met vriendelijke groeten,  Ernst


|  Dinkla Grafische Techniek  |
|     www.pigment-print.com    |
|             ( unvollendet )            |

Ernst Dinkla wrote:
>
> pr_roark wrote:
> 
> > On the other hand, I wonder if some of the color fringing at the 
> > edges is due to sensor design as well as lens aberrations. ...


> ... I thought for some time that as color fringing can be 
> removed with software the lens designers compromise it more 
> in order to get other specs in line. ...

My experience with B&W is that the color fringing manifests itself as 
softenss on the affected edges, and the software "fixes" I've tried in 
the Raw converters do not improve B&W sharpness.

I'd prefer better lenses (or sensors if they contribute) to a software 
fix. 

Paul
www.PaulRoark.com

On 7/10/2008 pr_roark wrote:
> My experience with B&W is that the color fringing manifests itself as 
> softenss on the affected edges, and the software "fixes" I've tried 
> in the Raw converters do not improve B&W sharpness.

Paul, you might be interested to try the software corrections in PTLens, 
which include color fringing (called chromatic aberration in PTLens).

See http://epaperpress.com/ptlens/ (free demo available).

Peter Marquis-Kyle

pr_roark wrote:
> Ernst Dinkla wrote:
>> pr_roark wrote:
>>
>>> On the other hand, I wonder if some of the color fringing at the 
>>> edges is due to sensor design as well as lens aberrations. ...
> 
> 
>> ... I thought for some time that as color fringing can be 
>> removed with software the lens designers compromise it more 
>> in order to get other specs in line. ...
> 
> My experience with B&W is that the color fringing manifests itself as 
> softenss on the affected edges, and the software "fixes" I've tried in 
> the Raw converters do not improve B&W sharpness.
> 
> I'd prefer better lenses (or sensors if they contribute) to a software 
> fix. 

Right, it was only to sketch that it may be more difficult 
to get that quality in lenses when this fix is widely accepted.

-- 
Met vriendelijke groeten, Ernst


|  Dinkla Grafische Techniek  |
|     www.pigment-print.com    |
|             ( unvollendet )            |

(Slightly OT)

I'd previously mentioned that the Photozone <http://photozone.de/> lens 
tests seemed to correlate well with what I saw on my 1.6 x digital 
camera.  But Photozone lacked full frame tests.

I've now found a site that has taken the Photozone types of tests to 
full frame.  See http://slrlensreview.com/

This, plus, of course, the old Photodo 
<http://www.photodo.com/products.html> tests may help us find glass 
that is worthy of the latest sensors.

Paul
www.PaulRoark.com