More useful: Paul Signac, Claude Monet
Claude Monet bien sûr !
And maybe some Jackson Pollock?
Added some new styles this morning.
Also added new options to make the computation possibly a bit faster (but don’t expect miracles ).
since version 2.4.2 i can not update the filters list from Gmic itself, so i downloaded directly the “update242.gmic” file.
Tried this new appealing filter and it ends up with this error message when i try to use one custom picture for style or any of the ones included as presets in the filter first option :
*** Error in ./fx_stylize/*repeat/*local/ *** Unknown command or filename 'stylize' (did you mean 'at_line' ?).
Not sure if it comes from the manual update of the filters file and i am missing something or if it’s another problem.
Using Gmic 2.4.2 QT x64 for Gimp 2.10 under windows if it helps.
Thank you and very nice work on this filter
Sorry my fault, should be fixed now.
Updated the file manually again, seems to work as i don’t get the error anymore on my first attempt, it’s generating the file and it may take a bit of time to see the result now
I’ll post later if i get some decent results to share or if i encounter more problems.
Thank you very much for the hard work and quick reply & fix
Perhaps there’s a better alternative, but I found an aur package to get the latest build on arch/manjaro. Too bad it’s marked “orphaned” by the packager, but it does actually work for me on manjaro now.
@David_Tschumperle the real time preview is cool!
I see it quite fast compared to other (with worse results) and obtain better results in > 2K
It would like me preview that “patch” goes to associate with another “patch”…
It is much faster when you put the smaller matching precision. I must have to experiment to get the most optimal balance. Maybe we should have a vote on what should be the standard matching precision.
Now, I’m into 1.3, but going to test 1.7 and 2.1.
I find that 1.5 is the perfect number.
Here’s my result and it taken way less time
U started toying myself and 1 was very fast but not that good of a result. Thanks for your input, Reptorian. Saves me a lot of time. lol
OK; this one took around 19 or so minutes (trust me; faster than I expected based on previous attempts). It’s 1600X1600 so rather big too. I used Reptorian’s suggestion of 1.5 and used Grey Tree and no color matching (figure the target was already grey; lol).
original images? is 3d?
This is the original image used.
Note: I did used unsharp mask though to stimulate local contrast before applying the g’mic filter.
@David_Tschumperle Could we get sample of what fidelity to x means? Those things confuse me, but I can work with them and can never figure out what they mean.
I think that:
First parameters are what goes to take of style.
Segmentation by levels does not take in account the colour.
The rest is the precision with which goes to make the patch to patch.
Lamentably it can not associate areas between himself
No; didn’t take any drugs. lol