[BUG?] latest G'MIC with GIMP 2.10.24

It seems that the latest G’MIC plugin (auto update 2.9.7 and .8) has some bugs and is “bugging some GIMP filters” as well

http://gimpchat.com/viewtopic.php?f=8&t=19358&start=20#p265445
http://gimpchat.com/viewtopic.php?f=8&t=19358&start=30

also now when trying to do Filters/Edge-Detect/Difference of gaussians (Legacy) on GIMP, we got a blank picture…

I’m not able to reproduce the bug right now.
Could you tell me a simple procedure to reproduce it from a single image?

I don’t think a bug suddenly have appeared with latest stable version 2.9.7, because this version does not receive new filter updates anymore (because 2.9.8 release is planed soon).
Also, with 2.9.8, I often build and release new 'development version" binaries as new stuffs are implemented, so you may indeed experience strange behaviors when using “old” 2.9.8_pre (i.e. a few weeks old) , that’s why it’s called “development version” :slight_smile:

Or may be is it python after updating Ubuntu? (I’m no programmer, so I don’t have the knowledge for that), but now it seems that’s not G’MIC the faulty one…

On this forum (the last post) it seems that even when removing G’MIC the problem is still there (that’s why it might not be G’MIC fault after all)
If you want to take a look
https://www.gimp-forum.net/Thread-Some-plugins-not-working?pid=24210#pid24210

To reproduce the step in GIMP > open any image then > Filters/Edge-Detect/Difference of gaussians (Legacy) , few weeks or so ago I was able to do it, I did not downloaded/updated GIMP, but my G’MIC filter is on auto update as well like the guy on the forum.

But 3 days ago I updated Ubuntu-MATE via the auto-updater, and I’m not able to use that filter anymore (result = blank layer)

Sorry to have disturbed you with that

This filter has nothing to do with G’MIC indeed, this is a built-in filter from GIMP.

Indeed, and I’m very sorry about that, it’s because of using this GIMP’s plugins GitHub - cl4cnam/gimp_cartoon_plugin: Pour transformer une photo en style BD
which is using G’MIC in its script as well, we mixed Difference of Gaussians (legacy) with G’MIC as well

But, now I can see that’s not G’MIC fault
again sorry about that

This is the part of the problem which involves gmic. (not just Gimp edge detect)

There is a Gimp plugin bede210.py which creates a comic book effect. This calls gmic_gimp_qt for various filters.

Using gmic_gimp_qt 2.9.8 and updated (today) the plugin fails.

It does work gmic_gimp_qt 2.9.8 with an ealier updates298.gmic file (2411 kb vs. 2451 kb)

Strange.

That’s because with 2.9.8, the correct filter command is :

  fx_curves_interactive 7,0,1,7,"0,0,40,60,100,100,-1,0,0,30,70,100,100,-1,0,0,100,100,-1,0,0,100,100,-1,-1,0,0,100,100"

(note the double quotes).

Couldn’t you make it accept parameters without the quotes? I did it for one of my commands but don’t know how stable that is.

No, I can’t.
The way parameters like value() work has changed in 2.9.8, and value() is now always equivalent to a single parameter.

I mean you could reformat the user input so that value() could accept the input. But you know what’s best. :slight_smile: It will be slightly confusing for users if you leave it be.

Yes, I do agree with reformatting thing. That being said, I never knew value() actually exist. I would have to remember this at some point as I don’t see how it could be useful for my filters.

No programmer here but does this line

pdb.plug_in_gmic_qt(img, layer, 1, 0, '-fx_curves_interactive 7,0,1,7,0,0,'+pointLum+',100,100,-1,0,0,'+pointSat+',100,100,-1,0,0,100,100,-1,0,0,100,100,-1,')

should be written like this?

pdb.plug_in_gmic_qt(img, layer, 1, 0, '-fx_curves_interactive 7,0,1,7,"0,0,"+pointLum+",100,100,-1,0,0,"+pointSat+",100,100,-1,0,0,100,100,-1,0,0,100,100,-1",')

line 35 > gimp_cartoon_plugin/bede2.10.py at main · cl4cnam/gimp_cartoon_plugin · GitHub

It seems with commit 47c51eba3b13, Mar 1 10:33:51 2021 +0100, the interface to -fx_curves_interactive changed such that the two numeric parameters before the final double-quoted parameter should be removed.

Haven’t tested, but - shooting from the hip - I would say ‘yes’. The python code needs a patch.

Thank you, I’ll give it a try.