Need help with creating a "No intersect scatter" FX-map/graph

sltrOlsson

I know I'm not alone in wanting a node like this and I assume there's a reason we don't have one yet. But I'm seeing the need for one the more I advance and I feel like it can't be impossible, at least not as a highly unoptimized but working graph.

My idea is fairly simply. FX-map(1) randomly place input1 a few times. Output to blur->history scan->invert(output1). Connect input1 to FX-map(2) and randomly scattter. Use otuput1 as mask color/luminosity. Cause I assume that is the only one that can mask? Repeat. If the blur is big enough the mask should be big enough to take away any shapes that is scattered on top of a already existing one. 

This creates a very big graph and can't iterate without adding the whole function over and over. So it's pretty hopeless.. but the basic idea of creating a scatter and bluring the result to use as a mask MUST work! Somehow. 

Any ideas?! 

NicolasW

I've tried to achieve this multiple times myself, it appears there is no other solution at the moment..

sltrOlsson

Aww.. I was afraid that was the case. 

What do you see as the main component that's missing for something like this? 

synergy11

I have to jump in here as well. As far as I'm concerned this is the last problem that needs to be solved in Substance Designer before it becomes the perfect program for creating procedural textures.    There has to be a way to procedurally lay down some random brick patterns without having them overlap. There should be a "pixel awareness" node that looks around itself and sees the black pixels from the mask and adds your new input brick in the exact center of the white pixel space, at this stage an auto-scalling in the new input brick would be the key.   I hope some math wiz figures this out and makes a node for it. Right now Im using my brick generators but having to hand place them to get that random effect. The Tile Sampler node has a similar feature where you can select use mask and there is also a center option for distribution but the results are not what I expect. So here's my vote for Allegorithmic to get on this problem! haha.