On Thu, Apr 21, 2011 at 02:14:01AM -0700, Jeff Harmon wrote:
> > Exchanging "settings" between advanced imaging processing packages is
> > often extremely difficult or impossible, simply because they all work
> > differently (internally).
>
> Thanks for your sensible reply. This is a solid general point to
> raise; however, I am curious how true it is in this exact case?
> Perhaps it is a tall hill, but one that could be climbed?  We are open
> to funding such an effort. 

Some of the basic settings could be mapped in a quite straightforward
manner, e.g. exposure and colour temperature where probably most
software uses the same units (EV and K), or some other scale with a
clear linear or otherwise simple relation.

Others would be a little harder, e.g. many packages give a contrast
control, whereas UFRaw lets you input an arbitrary curve as a way to
adjust contrast more flexibly. Such a curve could be generated from a
contrast value (and indeed there was a patch a while back that did this
in UFRaw), but it would be hard to define the right way to generate that
curve such that it acheived the same results as the other software,
especially when the target is a closed source package meaning everything
would have to be done by trial and error.

That trial and error process could be automated to some degree -
optimizing to find the input on a given control in UFRaw which gives the
closest matching output image to the result of a corresponding control
input in the other package. But that assumes you can even get the two
packages to produce closely matching output images in the first place at
"neutral" settings. And that each control in UFRaw maps clearly to one
control in the target package. These assumptions probably aren't true,
which means it becomes a multi-parameter optimisation problem, probably
quite non-linear, and expensive to evaluate. And it's not clear what the
cost function would be, i.e. how do you define how "wrong" a given
output image is. That comes down to being a subjective question.

If the intent is just to get some roughly reasonable  settings for UFRaw
from an existing collection of .cos files, then that might be practical
based on some rules of thumb for mapping parameters, depending what
combination of controls have been used and whether UFRaw can duplicate
their functionality. But it would be highly dependent on subjective
judgement, so it may not be something you can usefully fund somebody
else to do.


Martin

------------------------------------------------------------------------------
Benefiting from Server Virtualization: Beyond Initial Workload 
Consolidation -- Increasing the use of server virtualization is a top
priority.Virtualization can reduce costs, simplify management, and improve 
application availability and disaster protection. Learn more about boosting 
the value of server virtualization. http://p.sf.net/sfu/vmware-sfdev2dev
_______________________________________________
ufraw-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ufraw-devel

Reply via email to