More on pixel shifting

[Knecht, David]

I am still trying to understand pixel shifting.  I came across this  
old post by Jason Kirk when this came up before:

"The Leica DM-IR/BE is an infinity corrected system, therefore an  
emission filter being slightly off-camber in the filter wheel would  
not effect the optical path since light passing the filters is  
parallel.  The most probable reason for seeing a shift AFTER the  
dichroic in the emission path is not the barrier filter, but is if  
the tube lens (recombining element) is somehow off axis. This is the  
only lens element remaining. "
The implication of Jason's post (and no one seems to have  
contradicted him) is that one should not see pixel shifting in a  
properly aligned infinity corrected microscope due to filters.  I  
then queried Semrock about why they would sell Zero Pixel Shift  
filters if this were indeed the case.  They essentially disagreed as  
to the effect of infinity systems on pixel shifting.  Can someone  
clarify this issue?  THanks- Dave


Dr. David Knecht
Department of Molecular and Cell Biology
U-3125
91 N. Eagleville Rd.
University of Connecticut
Storrs, CT 06269
860-486-2200
860-486-4331 (fax)

[Breukers, C (TNW)]

Dear all,

When the filter is off-camber it does not matter whether the rays are
parallel or a different shape, it will always results in a shift.

It is all about the optical axis and the angle of each component of the
optical system with respect to this axis.

I haven't really looked into the old mail, but I agree that the original
assumption is not correct.

Simply get a pen and paper and you will see what happens when you place
the filter under an angle with respect to the rest of the optical axis
(system).

By the way, if the filter is off-camber you will also face the problem
of chromatic shift, meaning that the light transmitted through the
filter will shift to other wavelengths.

Regards,
Christian Breukers
 
---------------------------------------------------
University of Twente
Faculty of Science and Technology
Biophysical Engineering Group
Building 28 Horst
Room ZH273
P.O. Box 217,
7522 LW  Enschede
P: [phone]
F: [phone]
E: [hidden email]
I: http://bpe.tnw.utwente.nl
---------------------------------------------------

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of
David Knecht
Sent: maandag 2 oktober 2006 16:02
To: [hidden email]
Subject: More on pixel shifting

I am still trying to understand pixel shifting.  I came across this  
old post by Jason Kirk when this came up before:

"The Leica DM-IR/BE is an infinity corrected system, therefore an  
emission filter being slightly off-camber in the filter wheel would  
not effect the optical path since light passing the filters is  
parallel.  The most probable reason for seeing a shift AFTER the  
dichroic in the emission path is not the barrier filter, but is if  
the tube lens (recombining element) is somehow off axis. This is the  
only lens element remaining. "
The implication of Jason's post (and no one seems to have  
contradicted him) is that one should not see pixel shifting in a  
properly aligned infinity corrected microscope due to filters.  I  
then queried Semrock about why they would sell Zero Pixel Shift  
filters if this were indeed the case.  They essentially disagreed as  
to the effect of infinity systems on pixel shifting.  Can someone  
clarify this issue?  THanks- Dave


Dr. David Knecht
Department of Molecular and Cell Biology
U-3125
91 N. Eagleville Rd.
University of Connecticut
Storrs, CT 06269
860-486-2200
860-486-4331 (fax)

[Joachim Walter]

Dave,

if the front- and the back-surface of the emission filter are not
parallel ("wedge"), the filter will act as a prism and slightly tilt the
beam off its original direction. This tilt is translated by the tube
lens into a lateral offset on the detector. A different filter in the
filter-wheel will have a different wedge (or none) and so will lead to a
different offset on the detector. This is one of the reasons, why you
can get a pixel-shift between different color channels (chromatic shift)
seen through different filter sets.
Tilting the filter as a whole will not change the position of the image
in an infinity-optical system. In contrast, in a finite optical system
you will change the position of the image by tilting the filter.
So it's not a tilt _of_ the filter, but it's a tilt _in _the filter
(wedge) that causes the pixel shift. That also means you could not
correct the shift by tilting the filter, but you have to buy zero pixel
shift filters to get rid of it.
Sadly, if you use "zero pixel shift filters" does not mean that you will
have zero pixel shift. This is because the magnification of the
objective+tube-lens varies slightly over the spectrum (I think this has
also been posted, recently). To make things worse, the shift produced by
this effect varies over the field of view. E.g., if the magnification of
two channels differs by only 0.1%, and you take a 1000 pixels wide
image, the shift between the two channels on the left side and on the
right side of the image differs by 1 pixel.
For most objectives this effect is particularly prominent when you use
blue-violet dye (e.g.DAPI) together with a green/red/IR dye.

Hope this helps,
Joachim

David Knecht schrieb:

> I am still trying to understand pixel shifting.  I came across this
> old post by Jason Kirk when this came up before:
>
> "The Leica DM-IR/BE is an infinity corrected system, therefore an
> emission filter being slightly off-camber in the filter wheel would
> not effect the optical path since light passing the filters is
> parallel.  The most probable reason for seeing a shift AFTER the
> dichroic in the emission path is not the barrier filter, but is if the
> tube lens (recombining element) is somehow off axis. This is the only
> lens element remaining. "
> The implication of Jason's post (and no one seems to have contradicted
> him) is that one should not see pixel shifting in a properly aligned
> infinity corrected microscope due to filters.  I then queried Semrock
> about why they would sell Zero Pixel Shift filters if this were indeed
> the case.  They essentially disagreed as to the effect of infinity
> systems on pixel shifting.  Can someone clarify this issue?  THanks- Dave
>
>
> Dr. David Knecht
> Department of Molecular and Cell Biology
> U-3125
> 91 N. Eagleville Rd.
> University of Connecticut
> Storrs, CT 06269
> 860-486-2200
> 860-486-4331 (fax)

--
-----------------------------------------------
Dr. Joachim Walter

TILL I.D. GmbH
c\o BioImaging Zentrum
Großhaderner Str. 2
D-82152 Martinsried

Tel.: [phone]
Fax:  [phone]

[hidden email]

[Joachim Wesner]

Hi there,

>Sadly, if you use "zero pixel shift filters" does not mean that you will
>have zero pixel shift. This is because the magnification of the
>objective+tube-lens varies slightly over the spectrum (I think this has
>also been posted, recently). To make things worse, the shift produced by
>this effect varies over the field of view. E.g., if the magnification of
>two channels differs by only 0.1%, and you take a 1000 pixels wide
>image, the shift between the two channels on the left side and on the
>right side of the image differs by 1 pixel.
>For most objectives this effect is particularly prominent when you use
>blue-violet dye (e.g.DAPI) together with a green/red/IR dye.

Very true, in a traditional "finite corrected" system, even APO objectives
had a very
large amount of "lateral color", which would only be corrected in the
eyelens ("ocular")
resp. there were special "photo oculars" to correct for this in case you
used any kind
of camera.

Because making infinity corrected objectives that are "perfect" on their
own in any regard
would require a lens design of prohibitive complexity, usually any large
manufacturer
has it´s own design philosophy regarding what is already corrected in the
objective
(i.e. astigmatism, lateral and logitudinal color, curvature of field, coma,
spherical
aberration and so on) and what still needs to be compensated by the tube
lense (and probably
even by the ocular, but the latter compensation clearly is not used so much
any more today
because of problems with attached cameras - sometimes the approach is that
the central area of
field that is seen by todays small sensor is nearly perfect without the
ocular, while at the
outer edges of the field up to a diameter of 20-25 mm some aberration are
still handled by
the eyelens)

Therefore, you should clearly not mix infinity objectives from DIFFERENT
manufacturers, YET,
it also forces ONE manufacturer to try to induce the very same amount (in a
simplified way)
of aberrations in all objectives of it´s line for a perfect cancellation of
those after the tube
lens. This is not easy, as the "natural" aberrations of a certain design
will clearly depend
on magnification, aperture, free distance from from lens etc.

Sincerely

Joachim Wesner



______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________

[Joachim Wesner]

Hi,

>When the filter is off-camber it does not matter whether the rays are
>parallel or a different shape, it will always results in a shift.
>
>It is all about the optical axis and the angle of each component of the
>optical system with respect to this axis.
>
>Simply get a pen and paper and you will see what happens when you place
>the filter under an angle with respect to the rest of the optical axis
>(system).

Nope, a tilted (otherwise ideal, i.e. both faces parallel) filter will
cause
a slight lateral offset of any ray that passes, but, after the tubelens,
those
rays will hit the very same image spot as whithout the filter - only from a
slightly different direction. That "immunity" (within limits as vignetting)
to  extra optical components in the ray path is just what "infinity optics"
is all about.

>By the way, if the filter is off-camber you will also face the problem
>of chromatic shift, meaning that the light transmitted through the
>filter will shift to other wavelengths.

Chromatic shift does not imply that the wavelength of the light changes!!!

JW


______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________