Hallöchen!

Helmut Jarausch writes:

> [...]
>
> I've modified the lensfun data in my installation according to
> your suggestion.  [...]  The two images are very different.

Well, I could only use gnuplot, so it was kind of "blind flying" for
me.  Would you mind uploading the 25mm-close-focus image pair
somewhere, e.g. at http://wilson.bronger.org/calibration?  Thanks!

> Even if there would be some additional transformation it would too much  
> work:
>
> darktable / export / Gimp rescaling + cropping / import into
> darktable.

Yes, this is not senseful.

> [...]
>
> Perhaps I can find a way to patch Rawtherapee which doesn't use
> lensfun.

If you prefer DT over RT, patching Lensfun allows for a simpler
workflow.


Possibly the following – hopefully not too tedious – background info
is helpful for you to understand the implications of that
polynomial.

Mathematically, if scaling doesn't matter to you, the a,b,c,d set
becomes under-determined, and you can set any of the parameters to a
non-degenerated value without losing accuracy, which I did in the
post with the transformation method.

An overview of different models gives
<http://wilson.bronger.org/lensfun/group__Lens.html#gaa505e04666a189274ba66316697e308e>.
As you can see, only the Hugin-inherited models have d not equal to
1.  Both Imatest and Adobe (Lightroom) set d=1.  Note that Adobe has
five parameters, but no d nevertheless.

On <http://www.panotools.org/dersch/barrel/barrel.html>, Helmut
Dersch, creator of PanoTools and an active researcher in this field,
discusses the polynomial and explains that "d" is the scaling
parameter.

I scales in the sense that it changes the scale in the centre of the
image, i.e. in the
<https://en.wikipedia.org/wiki/Paraxial_approximation> (where the
focal length is defined).

Of course, one cannot just change d and the image gets bigger or
smaller, the other parameters must be changed, too, which is why I
removed a misleading part from the PanoTools docs:
http://wiki.panotools.org/index.php?title=Lens_correction_model&diff=14463&oldid=14358

By the way, for Lensfun, d != 1 is really unfortunate, because we
need to preserve the effective focal length for certain algorithms.
Thus, Lensfun jumps through some hoops for un-doing the scaling.

Hope this clears up some things ...

Tschö,
Torsten.

-- 
Torsten Bronger    Jabber ID: [email protected]

------------------------------------------------------------------------------
What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic
patterns at an interface-level. Reveals which users, apps, and protocols are 
consuming the most bandwidth. Provides multi-vendor support for NetFlow, 
J-Flow, sFlow and other flows. Make informed decisions using capacity planning
reports. http://sdm.link/zohomanageengine
_______________________________________________
Lensfun-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/lensfun-users

Reply via email to