Hi GC, walking through the various g'mic functions i made some tests on your test-function "compression blur" and i'm asking whether it would be possible and reasonable in future updates extending the upper value of the parameters (now 3 for both) to a bigger value (10?). The reason is that i did a simple test with a picture with big jpeg "squares" so that the value of 3 for the grid was insufficient to remove them appropriately; so i decided to downsize the image from 2000px to 500px width, then rerun the function, then up-size back to 2000 and jpeg squares were removed. I'd like to hear your opinion on that, as well as other experiences from other users, thanks

Hi there, I'm surprised somebody is using it - I was actually considering removing it (it was the first gmic filter I wrote when I knew very little about the subject) but perhaps not now! The max values could be increased to perhaps 4 or 5, much more than that would only make sense on an image of immense size (because at a certain point the grid covers so much of the image that you are just smoothing the whole image). I will have a look back at the code when I redo my filters for the blend update and see what is worth doing, which will be very soon.

In the mean time, I suggest you try playing around with Enhancement > Smooth [anisotropic] since that's the function it's based on and will smooth the whole image (but with edge preservation). Also resizing an image performs a certain amount of smoothing as well due to bicubic filtering but in general results in loss of detail.

Hi GC,
nice to hear from you and thanks for your quick reply.
Yes, you're probably right, 5 as maximum could be good.
Your g'mic filters on jpeg enhancements are very good, so in my opinion you have to go on!
See you soon back on the subject...

OK that's that it updated, there's a new option for 100% smoothing which bypasses the grid entirely. If you need more smoothing than that try either anisotropic or one of Iain Fergusson's noise reduction filters which are very good.