No, it doesn’t. Also, the downside of the new method is that I can’t replicate what I did with balance option on the version that’s available in gmic-community. Balance is the minimum value or maximum from the combined convolution of different stroke angles. If that can be solved, then potentially, this filter and rodilus filter can be 3 times faster i.e from 15 s to 5 s flat.
EDIT: I’m looking at this though I don’t know how to code for g’mic c++ side of thing.
Your filter attempt made me experiment some funny idea, based on the computation of 1D-blurred versions of rotated versions of the input image, followed by a median blending at the end.
The results looks a bit like what you obtain :
Playing with the final blending modes and different options of the filter, I think it deserves to be published as a new filter.
I’ll doing that this afternoon.
Okay, now I got new progress. However, I would like to be able to run a command to verify if window is on since I am running run(w[-1], and it is used to check on the progress of DLA.
Here’s a example involving new code.
You see that it’s growing around the text image? Let’s say I close the window. I should expect to end with that. I can always add a option to add boundary option to limit numbers of vector though.
EDIT:
Solved the mask issue. Also, I added a new option to erase mask after processing.
There is a bug on the filter with double pixel axis mode, but single pixel axis looks better anyway, so I wouldn’t be so pressed about it. I will fix that bug when the time comes. It rarely shows up anyway.
Non-Isometric Tiling Tool now has fill tool built-in to it. Read the instruction to use it. I thought it was going to be difficult, but highlighting lines with KDE Kate really makes it easy.
You linked to this thread. The smoothening and transition part exists. Just not the structure. From a theoretical perspective, I can take a look at TR’s source code to one of his filter and do something like that.