proj-imim:
Re: Corrections
Helmut Dersch 2000-Jun-13 17:03:29
Marcus Bointon wrote:
>
> I've kind of given up on the optimizer, and I'm now trying to get radial
> shift correction values by hand, and I'm getting much better results. For my
> Kenko lens, I'm now using:
> a = 0.022
> b = -0.09
> c = -0.025
> d = 1
>
> and things are looking fairly straight.
>
> I am however, running into some strange practical problems. I've been
> experimenting in Photoshop 5.5. The processed image is larger than the
> original, so as is well known, PT needs to create a separate image file
> which it can automatically reopen. Even though all the boxes are ticked
> (except for crop), and I've specified a file location, this is not
> happening.
>
Hi Marcus,
Sorry for the late answer.
The 'correct->radial correction' filter does not resize the image.
Output will always be the same size as input.
> [GraphicConverter 3.8.2 problems]
I haven't upgraded my GC-installation lately. Are there more
problems with newer versions of GC?
>
> I also noticed that PT is in some way image size dependent. I thought the
> radial shift filter was simply pixel based, but it seems to be angle based,
> so if I enlarge the canvas of my source image so the converted one fits on,
> then the result looks completely different with the settings I worked out.
>
The radial shift filter is not size dependent. If you apply it to
an image 1000x1500pixels, the results will look exactly the same
as if applied to the same image sized 2000x3000pixels. Therefor,
you can correct your scans independent of scanner resolution.
It also doesn't matter whether you use landscape or portrait
orientation. However, changing aspect ratio, or cropping the image
matters.
Regards
Helmut Dersch