Dear friend Diego, your script works very well but could I ask you that the remove option black line was optional by the user? I like the black definitory line in some items. Thanks a lot.

Issabella, it is optional.

Anyhow, I made an addition which preserves also the filled with borders, when the option "remove grid" is selected.
So you get 2 layers, on top the one without the black borders, the other one below with the black borders.


Could you try?

I also updated the BatchFilling (which did no more work with the new "filling").
In this case, if you ask to remove the gris, two images are stored in the outcome folder, one with the borders, one flattened.


regido and anyone else interested can try.

(remember to use threshold on each source image if you want to be sure the drawing are pure black&white)

Hello Diego.
Let me explain you more in the detail my problem to see if there´s any solution.
This is the original image i´m going to use. Is pure black and white



I apply your filling plugin selecting white as the color to be erased and use random patterns. The result is here



There´s something very strange where part of the patterns are not well represented as you will see in this next example with a completely different result



Instead of using black and white i use this chroma green and black.



Using the same patterns, instead of selectin white i use that green chroma to be removed, the result is much cleaner as you can see.



Here is the image layer without the black border



Here is the zoom of the layer wihout the black border.

Do you have any idea about what could be the problem or the issue with this?

Thanks a lot in advance

I used a white and black image, too. I chose from gradients and keep the black lines. I love the result.

Downloaded your B&W source image.
I will do many tests with my patterns sets to see what happens.
I'll let you know.

First tests:
- I ran 10 times the filter (w/o removing the black grid) and all of them came out cleanly filled with random patterns.
Hence my questions:
- which sets of patterns have you used (they look very strange to me)?
- the all-green result came out just only once or can you repeat the event?
I cannot understand the second of your trials, but I'm interested to solve the first problem.
Awaiting your reply.
Maybe you can post your pattern sets used in step 1 so I can repeat another 10 tests,

Also, you can post the screenshot of your launch.

Hello Diego.
Here you can see a different test.


BW image with "animal" patterns (the ones that came with GIMP)



As you can see i´m still getting a white border.

Concerning my patterns i´m still getting the same results.
Maybe i´m using a set that is too big (i have around 300 patterns there) each pattern is 2000x2000px.

The strange thing is that if i do the test with greens instead of whites the result is much better, like the one i already showed you on my last post.

Thanks a lot again

You didn't do what I suggested.
In fact:
- downloaded your B&W drawing
- launched the filter (random patterns)
- got me too very subtle white margins
Then:
- run threshold on the B&W drawing
- launched again the filter (random patterns)
- NO white margins
What happens when using greens instead of white is very likely the fact that you succeed in filling all what is not black with pure green, hence creating the two-pure-colours situation like threshold does with black and white.

Hello diego.
Im getting better results with treshold.
Even with treshold my patterns look completely different if i use green instead of white. If i choose to remove white they don´t look like those are, but if i use green i get better results (only applying the patterns, the border is worse for the reason you said in your post)

Bellow is a sample of the same image applying the filler with treshold and filling white colors (I´m using exactly the same patterns, only variation is the color to fill).




Here is the version filling the green color.



It is a matter of resolution or quantity of patterns or something like that?
that´s a very strange problem. Check the difference

I see greens in the design lines, while it should be black (you said replaced whites with greens, not blacks with greens...)?!?

I wanted to do what you said to me in your post.
Using original source, black and white:

Using the source with white replaced by green:


Obviously the patterns chosen are different (random), but no significant difference.

maybe its a matter of the file size or numbers of patterns?
If i put the design in black and green and fill green the patterns look good, if i put the design in black and white and fill the white it doesn´t look good, patterns are messed up.

Try defining a set of a few patterns, so that the random function will in any case select the pattern among those few, repeat the double test and see.

Another test you can do is:
- run 10 times the filter with the same source image and same parameters
look:
- are all 10 outcomes using White-Black bad and all 10 outcomes using Green-Black good?
And: what/why you consider good and what bad?
I guess you have a set of patterns with so different structures that some outcome could be seen good or bad depending on the pattern chosen.

Another question: do some of your patterns include whites?
That causes that at the next step of random choice also the white parts of the already filled areas are selected and replaced.

Made a change to the filter, which (hopefully) avoid inserting a colour identical to the one to be replaced, so (in case of drawings black&white) patterns having white areas can be used without being "ruined" by the gray replacing colour or a new area selection+fill.
Issabella, Regido, could you do a check test?

@Krikor
If you are still interested in me adding the additional option of random gradients filling, I can try to implement it (at this point -after all the last amendments- the filter is no more a "divertissement", so no reason to refuse this interesting addition....)