[ptx] Release: waveblend 0.1.0 - focus enhancing image fusion tool

[Glenn Barry]

For close up stuff in the big magnification ranges if you had belllows, you
could maintain the nodal point by using the rear camera movement to focus, I
think...
----- Original Message -----
From: "Rik Littlefield" <rj.littlefield at computer.org>
To: "Sebastian Nowozin" <nowozin at cs.tu-berlin.de>
Cc: <ptx at email-lists.org>
Sent: Monday, July 12, 2004 5:08 PM
Subject: Re: [ptx] Release: waveblend 0.1.0 - focus enhancing image fusion
tool


> Sebastian,
>
> Great job!
>
> I ran the same images through the visibility mask algorithm available at
> http://www.janrik.net/ptools/ExtendedFocusPano12/ .  Results are posted
> there -- pretty much the same as yours.
>
> I am not sure how to compare the two approaches.  Both the beaded deer
> and the microscope images seem to be fairly simple cases.  I suppose
> it's time to haul out the macro lenses and try to work a harder
> problem.  (I play with insects in the 1-10X magnification range.  It's
> difficult maintaining fixed nodal point in that range, which makes
> perfect alignment problematic.  I always end up having to edit the
> masks.  But perhaps there are other ways.)
>
> --Rik
>
> Sebastian Nowozin wrote:
>
> >Hello everybody,
> >
> >
> >I am happy to announce the results of my last week efforts to implement
the
> >focus-enhancing algorithm I was pointed to by Rik Littlefield (here:
> >http://bigwww.epfl.ch/demo/edf/).
> >
> >To summarize the purpose of my tool in one sentence: Take the input
images,
> >each with different focus depth and blend them so the final image is in
focus
> >everywhere.
> >
> >My tool is called waveblend and more or less implements the algorithm
1:1. But
> >instead of using the simple 2d CWT as in the original paper, I use the
more
> >advanced dualtree CWT. A screenshot, technical description and the whole
source
> >under the GPL is available at:
> >
> > http://user.cs.tu-berlin.de/~nowozin/waveblend/
> >
> >For merging real photographs the photographs have to be _exactly_
aligned, even
> >a 1-2 pixel misalignment causes trouble to the algorithm (I may have to
find a
> >solution for this).
> >
> >As always feedback is welcome :-)
> >
> >
> >Regards,
> >Sebastian
> >
> >
> >
>
>