an actual function in gmic (on line 4468 of the definition)
maybe the problem may be "w" or "h", that in the def are for image width and height but is not specified if that indicate also the respective value in pixel

SOLVED!!

it was just a wrong type of parentheses "()" instead than " {}" , thank to Karo to point it out
http://www.flickr.com/groups/gmic/discu ... 1608183437

The filter is ready ...but now how you type " {}"...till now i solved by copy and paste

The filter is ready viewtopic.php?f=4&t=5303 and may set automatically the best theoretical value for pre blur

BUT i am no so convinced that the theory give the best result...give very good result if the scaling followed by a moderate sharpening (more or less the default in the filter )..but without post sharpening seems to add a soft focus effect .

So i added manual control for blur ...and returning to the best method for downscale i agree for a minimum of about 1 pixel but then i will seldom go over 1.6..
let 's say to to scale 3:1 i get better result with 1.5 then with 3, that will be the theoretical value.

Also with this values a moderate post sharpening improve much but is no more strictly needed to avoid a soft focus effect

The theory tells you why you get artefacts sometimes. Correct pre-blur has very little effect on final output in most cases, but it an in some cases prevent an artificial accentuation caused by frequency folding which isn't so violent as to appear as moiré or aliasing.

My Rough Idea of Gimp Scaling Options
Gimp Default, None, nearest neighbor, (digital value from the pixel in the original image which is nearest to the new pixel location in the scaled image), possible side-effects, aliasing, artifacts.
Good, (Bi)Linear, 4 nearest pixels (2x2), possible side-effect, aliasing.
Better, (Bi)Cubic, 16 nearest pixels (4x4), possible side-effect, blurring.
Better, Sinc (Lanczos3), 36 nearest pixels (8x8), possible side-effect, ringing/halo.

All the technical stuff is at the following link. There's lot's of good reading along with the math. Make sure to check all 3 tabs, article, demo, archive.

Linear Methods for Image Interpolation
http://www.ipol.im/pub/art/2011/g_lmii/

Hello,
i know this thread is old but it exactly matches my problem. Im looking for a way to reduce the resolution without losing too much detail.
Several times (also in this thread) the gluas plugin plus decimation algorithm was named to bring good results. So i wanted to try that out.
I downloaded GIMP ver. 2.8 and a compiled version of gluas 0.1.20 and put it into the plugin folder.
It seems to appear properly under "Filters-Generic-gluas" (im using the german version as you can see in the picture). But i dont know how i can use it, since its not activated (see picture).



Im kinda new to GIMP so i might miss something there.

Someone, having experience using this plugin, can help?

I tried to find a good tutorial for gluas too... a lot of them explaining the syntax and other details. I found a algorithm for the Decimation process but where to add/insert this?
Do I have to generate additional script-files in the plugin folder?
Does the faded button means its not compatible (tried 32bit & 64bit plugin)?
OS is Win7 Ultimate x64 btw.

Im also interested in alternative solutions. I know there is no ultimate algorithm for EVERY picture, but im asking to share your experience

Greetings

Do you have a layer open? It won't activate without an open layer.

Hello, for fast reply.
Yes, i opened a file and selected the layer... but my testfile was a greyscale image
I switched to RGB mode and its working now. Results with decimation script are indeed very good. I think i can work with that.
Are there any other nice plugins for downscaling?

Greetings

I use image scale image when I resize . There is also filters > generic > mathmap > time
and filters > mathmap > geometry