(none) imager-devel
/ help / lists / applications / search /

Re: Using other applications affine transformation matrixes with Imager

From: Tony Cook (
Date: Sat 07 Nov 2009 - 22:53:39 GMT

  • Next message: Bernhard Wei▀huhn: "Re: Using other applications affine transformation matrixes with Imager"

    Hi Bernhard,

    It seems the main difference is that Imager uses the matrix to convert
    a location in the destination image to a location in the source image,
    while ImageMagick uses it to convert a location in the source image to
    a location in the destination image.

    I expect what you need is the inverse of the matrix you're using with


    On Sat, Nov 07, 2009 at 08:03:46PM +0100, Bernhard Wei?huhn wrote:
    > Hi,
    > being seriously impressed with the performance of Imager, I want to
    > convert an application from using Imagemagick's affine transform
    > functions to Imagers own matrix_transform().
    > Unfortunately, their behaviour is quite different.
    > ImageMagick (and Cairo too btw.) use six-coefficient matrices like to:
    > [ sx, rx,
    > ry, sy,
    > tx, ty ]
    > ... which i naively converted to Imagers nine coefficient matrix like
    > this:
    > [ sx, rx, tx,
    > ry, sy, ty,
    > 0, 0, 1 ]
    > Unfortunately, it's not that easy. By trial and error I ended up with
    > some modifications (1/sx etc), which give better results for easy
    > cases, but fail as soon as negative parameters or more complex
    > combinations are involved. I suspect a relatively easy mathematical
    > transformation of the paramters, but I can't figure it out by myself.
    > Can somebody point me in the right direction?
    > ImageMagick's matrix transformations are explained in detail here: http://www.imagemagick.org/Usage/distorts/affine/
    > Btw: I was supprised to see Cairo working with imagemagicks matrices
    > out of the box, but I'd like to avoid my current hack moving image
    > data back and forth between Imager and Cairo ImageSurfaces, and would
    > rather use Imagers builtin matrix_transform, since it not only seems
    > faster, but also prettier.
    > Any help greatly appreciated,
    > bkw