PanoTools mailing list archive

Mailinglist:PanoTools
Sender:gerdsy
Date/Time:2005-Oct-28 01:50:40
Subject:Re: pano12.dll ver 2.7.0.11

Thread:


PanoTools: Re: pano12.dll ver 2.7.0.11 gerdsy 2005-Oct-28 01:50:40
--- In #removed#, "Erik Krause" <#removed#> wrote:
>
> On Thursday, October 27, 2005 at 20:13, gerdsy wrote:
> 
> > The next beta version of PTShift will be compatible with both PS CS
> > and PS CS2 on PC/Mac, and will not require any external optimizer.exe
> > (Windows) or Adjust plugin (Mac OSX) to calculate the abcd values to
> > remove Transverse Chromatic Aberration. 
> 
> How do you optimize the parameters? Or do you a simple polynomial 
> fit?
> 
> best regards
> 

The latest PTShift I sent you with all the documention in it has a
routine called PTShiftOptimizer(). If you read the docs
for that routine, it covers what I did.

Given:
    rs = a*(rd^4) + b*(rd^3) + c*(rd^2) + d*(rd)

Ideally, I would have liked to minimize the sum of the squares of the
residual errors in rd, but then solving for abcd gives nonlinear
equations, and hence the Marquardt and Levenberg algorithm
could be used. My eyes kind of glazed over as I did not feel like
figuring Marquardt out.

So, I attacked the problem from another angle. I can minimize the
weighted sum of the residual errors in rs instead, which gives a
linear set of equations for abcd (easy to solve). Then I can adjust
the weighting coefficients for another iteration to see if the
residuals in rd (not rs) become smaller.

I compared the total errors using the PTOptimizer.exe and my new built
in PTShift optimizer, and there seems to be very little, if any
difference. (This is especially true when the baseline abcd = 0001))

And if I ever decide (or have the time) to figure out that Marquardt
and Levenberg stuff, I can just rewrite that one function
PTShiftOptimizer() without messing anything else up in PTShift.
But the way I do it now is very fast - only a couple iterations, and
the results are pretty good.

The built in optimizer was necessary because the latest Kekus plugins
are apparently not compatible with any version of PTShift (the Kekus
plugins were used by PTShift on Mac OSX to calculate abcd).

Eric
 
 






------------------------ Yahoo! Groups Sponsor --------------------~--> 
Get Bzzzy! (real tools to help you find a job). Welcome to the Sweet Life.
http://us.click.yahoo.com/A77XvD/vlQLAA/TtwFAA/.Cr1lB/TM
--------------------------------------------------------------------~-> 

 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/PanoTools/

<*> To unsubscribe from this group, send an email to:
    #removed#

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/
 




Next thread:

Previous thread:

back to search page