With the help of various code snippets from this forum, I have managed to create a working metaball simulaton.  I used a threshhold to get a clean border between the metaballs and the background.

Now, I would like to draw just the outlines of the metaballs. I tried using a min. and max. treshholds to achieve this, and allthough the result (shown below) does show something aking to an outline, it's still not quite what I am looking for. Especially the thicker parts of the line is what I want to get rid of.

As far as I understand it, the above image doesn't actually contain any lines, but just neighbouring white pixels which give the impression of a line.

I am thinking that it should be possible to find coordinates on  these "outlines" and use them to create a bézier curve, wich would then creat a clean, continuously thick line around the metaballs. The problem: I have no idea how to start doing this.

Is this actually possible?

Am I overthinking this using metaballs and there is in fact a better solution?

Does this have something to do with blob-detection?

I'm not trying to create moving metaballs. I just want to draw the metaballs once and then create the clean outlines. So performance isn't an issue for once.

If anybody has any pointers, I'd be thankful.

It's simple to find a character at a specific position within a string: string.charAt(index)

I'm imagining, that it should be just as simple to replace a character at a specific position in said string with another character, but I can't for the life of me find a simple solution. The best I can come up with is to dissect the whole string into single chars, replace one and then join everything together again. But this seems very inelegant.

Basically I'm looking for the PImage.set(x,y,color) equivalent for Strings. Something like string.replace(index, char). Is this possible?

Any pointers are appreciated!

I'm not sure how this topic should best be called, so let me try to illustrate my problem. 

Say I have a collection of points (the white ones), randomly scattered across the stage. Now I take one point (the red one) and try to find the one white point that is closest to it. Until now, I'd just check the distance to EVERY single point in the stage (all saved as objects in an Arraylist), and eventually find the one that is closest to my red point. This works quite fine for 1000 points, like in the image to the right. But this is only a very simplified version of what I'm trying to do, and eventually I'm expecting on having many 10k or even several 100k points in the sketch. That's where the calculations are starting to become quite resource hungry.

Is it possible to limit the "distance-checking" between the red point and it's white neighbors to a certain area (e.g. all points within the red circle)? This way I would only have to check a fraction of the total points, thus saving processing power. Is this possible? No idea if I can limit the search to an elliptical area or if it would be rectangular. How can I know if a point is near to my reference point, WITHOUT having to calculate it's position anyways?

I'd be grateful for any pointers! (See what I did there?)

Thanks!

cheers.

UPDATE: Would it perhaps make things easier, if the points where not randomly scattered, but set up on a grid? I'm guessing that it might be easier to reference the points, since they are written to the ArrayList in a more "orderly" fashion.

I have no idea, if what I'm thinking should be possible, actually IS possible, but let me try to explain:

I want to add new methods (behaviors) to initially "dumb" objects. In my sketch, all objects can have several, very different behaviors, but only one of them at a time. So up until now, I'd just create one kind of object "preloaded" with all methods it could eventually use. And then the object would await certain conditions, switch to the method it's told to use and go with it from there.

But... since I'm expecting my final behaviors/methods to become quite complex at some point, I'm wondering if there is another approach to what I'm planning to do. As mentioned above, I would create initially dumb objects, which don't have any behavior/methods and are completely oblivious of any behaviors they might receive. Then, when needed, a "controlling function" will select one of the objects and add one specific method to it. This way, every object only uses one method at a time and isn't cluttered with the other heavy methods it's not using.

Obviously, the script won't be able to write the method as new lines of code into the object. So I'm thinking there might be some way to tell an object to look for it's behavior with an other object. Some kind of "cross-referencing" of two objects. Sorry if this sounds abstract, but I really have no idea IF this is possible (or if it actually makes sense to try to do it in the first place). And I have no idea what this kind of technique might be called (if it even exists).

Again, I have no clue if my thinking is completely off track. But I'd really appreciate any kind of input on where I could start looking more into this.

Thanks!

cheers,
Greg