How to avoid random nearby repeating colors in a square based pattern?

Since long time ago I wanted to be able to make something similar to what is shown above here where it has random square based patterns with different colors that are NOT repeated nearby, I mean that the color is not repeated in the X or Y sides of the same color, it can be repeated in diagonally, but not more than once as in the example above

I have tried to use bomber, bomber plus and tiles, but I always get repeated colors patterns randomly

Long time ago I made this filter Color squares with photos

And as you can see here the colors are repeated in X or Y sides as seen here

Already 5 years ago in 2013 I have put this thread

How can you control the distribution of the particles inside Bomber?

I already tried to discover a way to do it in this thread, but failed to do it

So now there are new ways to do things and I know that with FF 6 and FF7 there are new ways to use slaves components, and there are new greenified components, but I regrettably really do not know how to use them, so can not experiment or test them

I think that probably the best way would be with bomber or bomber plus, as I want to put also images over the colors, but maybe there could be an alternative way with other components, for example using tiles, or any other way

So, please, can you be so kind to tell me if there could be a way to do this as in the example of the first post in a random way?

It would not only be random colors, also is needed for random images over the squares so they are not repeated in X or Y either

I would like to have 7 or 8 different colors if possible to make it more random, and 7 or 8 images

Thanks very much for any possible help, much appreciated

I hvae been trying to search in the FF filter library for any possible filter that could make something similar in some way, but until now I have not found any

Circumsquare by Ramlyn

That seems to have the same problem as my filter where some of the colors and shapes are repeated on X and Y sides

The thing with randomness is that it's... random. These examples have a very limited selection of colors and tiles, so it's very unlikely that none of the colors will repeat when randomized. Perhaps if you randomized the pattern a million times, one of those patterns would have no repeating colors - or not. You wouldn't expect to never have repeating values when flipping a coin.

So, a brute-force method would be to go one by one in a loop and check the neighbouring-cells to avoid repetitons. I guess you would actually only have to check two neighbours if you fill the cells in order, not sure without testing it. Anyway, i think that would get very slow, checking through all the iterations. Maybe there's a clever way to optimize this.

Another way could be to carefully prepare smaller non-repeating patterns that can then be positioned and rotated to form a larger pattern. I'm sure there's been research done on this. Most likely there are existing algorithms to generate these kind of non-repeating patterns. I'll see what I can find at some point.

Maybe I have explained it wrong in some way, any of the colors CAN be repeated again many times in the randomized pattern, and is not a problem,

what I am telling is that

1 - as seen on the first post, NONE of the repeating colors are either on the top or below the same color, or is not at any of the sides

2 - what is wrong is as on second post example on the top left, that there is two red squares together, one below the other (and more like this)

3 - It can be repeated the same color diagonally, but not many times, because then it would create a link of the colors

4 - it would have a minimum of 7 or 8 different colors so there is less chance that the random colors are repeated nearby

Quote
emme wrote: The thing with randomness is that it's... random

Yes, I totally agree and is true that randomness is ...random, but as I see it from what I know, there is pure absolute random and also controlled random, I mean that you can give some help to the random to be able random with some possible rules or between some values, and I have seen that in some way controlled random is possible in FF on some filters, although I do not have enough FF knowledge and experience to do it myself

Quote
emme wrote: These examples have a very limited selection of colors and tiles, so it's very unlikely that none of the colors will repeat when randomized n

Sorry that I do not understand, I agree that the examples shown have a few colors so is a limited selections, but in that case is very likely that the colors will be repeated and not unlikely as you said, or I am understanding it wrong

Quote
emme wrote: . Perhaps if you randomized the pattern a million times, one of those patterns would have no repeating colors - or not. You wouldn't expect to never have repeating values when flipping a coin

As at start of this post, I agree that the colors can be repeated and I do not expect of course that the colors will not be repeated

Quote
emme wrote: So, a brute-force method would be to go one by one in a loop and check the neighbouring-cells to avoid repetitons. I guess you would actually only have to check two neighbours if you fill the cells in order, not sure without testing it. Anyway, i think that would get very slow, checking through all the iterations. Maybe there's a clever way to optimize this.

Well, It can be a good idea, but I am sorry that I have no idea how to do it

Quote
Skybase wrote: Four color theorem comes to mind. https://www.wikiwand.com/en/Four_color...#/overview

Cool and interesting, now I remember I had already seen this long time ago, and I had forgot it, but how can this be applied to FF?

Your other two possible suggestions may be good, but I do not know how to do them, sorry

I just thought of a pretty simple solution. Not sure yet if it will work or not. Basically you would generate a list of integers in a script, each number representing a color.

For each step, generate a random number (in the range of colors used) and compare it to values in the list at x-1 and x-row_width (the cells on the left and above the current cell) until a new color is found. Repeat for row width * row height and you should have a sequence of colors that won't repeat left-right-up-down when rearranged onto a grid.

I'm probably missing something obvious here (tired lol), so anyone feel free to point it out. I'll try this out later and see if it actually makes any sense

Just realized FF doesn't render out images with script errors, so this might not be very useful before its fixed. Anyway here's proof

Agreed great work, I figured the only way to do this was in a regular language where you could easily lookup the previous color cell and the cell above. Here is a fix with no errors by doing array boundary checking.

non_repeating_cells_v002.ffxml

Thanks really very much emme for taking the time to make it and Rachel for fixing it, I will see how it works and tell you later. It seems a great work from the example shown

I've modified the script slightly to have the output values in 0-1 range. Works well with map switch now, so you can easily choose all the colors.

non_repeating_cells_v003.ffxml

Emme, I love using your "non_repeating_cells_v003" filter for getting random colors. I have another question for this one - if I wanted to place another set of different colored shapes, say a smaller circle on top of each of the square tiles, how would I go about this, ie. also making the circles random colors, but different colors to the base of the squares which they are on top of?