gimp gmic mixer filters

One of the members of gimp-forum came up with this. The Colors → Mixers filters

With anything other than All-tones in the Tones Range drop down, creates a desaturated result. This the RGB mixer, the others similar.

1 Like

Thanks, I’m checking this…

Filters updated, press the “Update Filters” button to get the fix (requires at least G’MIC 2.9.9).

New day, another go. Unfortunately still getting a desaturated result:

This time kubuntu 20.04 / Gimp 2.10.24 (from PPA) gimp_gmic_qt 3.0 from gmic.eu
The updatexxx.gmic for 2.99 and 3.0 both 4685061 B All the Mixer(XXX) versions the same.

Now fixed :slight_smile:

1 Like

Yes, now it’s fixed. :+1:

@David_Tschumperle,

Please can you also take a look at the “Freaky Details” filter, the amplitude and scale sliders do nothing there. It’s been like this for a while.

This does make a difference for me. What version are you using?


[gmic_gimp2.10_win64.zip](https://gmic.eu/files/prerelease/gmic_gimp2.10_win64.zip) 2021-09-02 18:36

And there is no interference with my user.gmic file (temporarily removed)

But what happens when you set Scale=1 for instance ?

It will take much more time to process, but no difference in result. (preview and end-result)

So that’s not the same as with Scale=100 right?

The result is the same (also scale 100), but the computation time is different.

the only slider that makes difference, is the iterations slider

This is what I get with the gmicky image and different values of scales :

  • Scale=1:
    scale1

  • Scale=10:
    scale10

  • Scale=100:
    scale100

It’s not the same result. I’m using 3.0.0_pre (but 2.9.9 should render the same).

I will do the same and post here:

Original (no freaky details):
original

Everything hereafter with: Iterations=1 and Amplitude=2

Scale=1:
scale1
Scale=10:
scale1
Scale=100:
scale1

The strange thing here is, that the image is exactly the same for the website, it automatic stores it as the same image, but I uploaded it 3 times. (with different file names)

@David_Tschumperle:

The filter works with removing the alpha channel. (I just found out by accident)

1 Like

This tells me that there should be a guideline to factor in the existence of alpha channel when creating filters for the GUI version.

The general rules:
GIMP - Any number of channel from 1-4 ? Scaled to 32F * 255.
Krita - RGBA-32F scaled to 0-255 regardless of picture being in non-RGB space or not.
Paint.NET - RGBA-8I

Not sure for the rest. I’ll ping @PDN_GMIC.

1 Like

Great catch!

https://github.com/dtschump/gmic/commit/bedee63ba18c12c5f37405af1f35a261443995c3

Try to update your filters, it should work now.

Basically, if a filter should not deal with the alpha-channel of an image, then a usual construct would be something as:

repeat $! l[$>] split_opacity l[0]
   ... # Your filter code here
endl a c endl done

(it just preserves the initial alpha-channel).

Maybe @grosgood could be interested to add this construct into its tutorial pages :wink:

2 Likes

@garagecoder 's “LMS Adjustment” (G’MIC>Testing>Garagecoder>LMS Adjustment) can use this “treatment” too and then straight to the “Colors” filters. :smiley: