RSS/Atom feed Twitter
Site is read-only, email is disabled
  1. Chat
  2. Useful links
  3. Polls

Featured content

Lanczos algorithm funnyness?

This discussion is connected to the gimp-developer-list.gnome.org mailing list which is provided by the GIMP developers and not related to gimpusers.com.

This is a read-only list on gimpusers.com so this discussion thread is read-only, too.

3 of 4 messages available
Toggle history

Please log in to manage your subscriptions.
Lanczos algorithm funnyness? William Skaggs 17 Sep 17:48
  Lanczos algorithm funnyness? Geert Jordaens 20 Sep 22:22
op.tfzrw2xxtxpshh@linbox.lo... 07 Oct 20:24
  Lanczos algorithm funnyness? gg@catking.net 17 Sep 09:09
gg@catking.net
2006-09-17 09:09:48 UTC (over 17 years ago)

Lanczos algorithm funnyness?

Hi,

I know this is rather old but it still seems pertinent.

Now that lanczos is firmly in place and seems to work pretty well this probably should be addressed.

It seems to me the there is a fundemental analytical difference between expand and reduce even tho' the two can be presented on the same %age input, does this make sense?

Either the list of "interpolation" methods needs to swap about depending on the % value entered or two different dlgs are needed. Image|Expand and Image|Reduce.

I suggest the latter solution is preferable. This would present the user with clearer choises and technically correct, better performing filters could be applied to reduction.

It seems a user would choose to either increase or decrease an image. One or the other.

How much bigger/smaller is a continuous scale but the choice of direction is binary.

The coding is presumably fairly straightforward bit it seems a different set of choices are needed for reduction.

best regards, GG

William Skaggs
2006-09-17 17:48:43 UTC (over 17 years ago)

Lanczos algorithm funnyness?

From: gg@catking.net

Thanks for your ideas. If you want to play around with Lanczos, you might benefit by looking over the enhancement requests and bug reports relating to it:

Bug 162250 ? Feature Request: Better image scaling algorithms. http://bugzilla.gnome.org/show_bug.cgi?id=162250

Bug 343576 ? Lanczos resize not "the best" http://bugzilla.gnome.org/show_bug.cgi?id=343576

Bug 167956 ? scaling image causes an offset, particularly with Lanczos http://bugzilla.gnome.org/show_bug.cgi?id=167956

Bug 355178 ? transformation tools with Lanczos interpolation brighten the result http://bugzilla.gnome.org/show_bug.cgi?id=355178

The Lanczos algorithm does not use a 3x3 matrix. The code for Lanczos lives in two places:

app/paint-funcs/scale-funcs.c -- for the "scale image" and "scale layer" commands

app/core/gimpdrawable-transform.c -- for the transform tools

As Sven pointed out, this code has been changing pretty rapidly in an effort to make it robust enough for inclusion in the upcoming 2.4 release, so it would be best to look at cvs HEAD if you want to avoid seeing things that are already obsolete.

-- Bill


______________ ______________ ______________ ______________ Sent via the CNPRC Email system at primate.ucdavis.edu

Geert Jordaens
2006-09-20 22:22:50 UTC (over 17 years ago)

Lanczos algorithm funnyness?

Hello,

just to say I'm still working on the lanczos scaling however I can only work limmited time on it.

Geert