Random crossing pattern

I’m trying to replicate this pattern.

This is as close as I have come.

I’d appreciate any help. Thanks

A few horizontals are straight in the original. Also some boxes have been split vertically while their neighbours have not. That’s something which is probably done in a second step.

I don’t think you’re going to get there by deforming a grid, as each cell in the target pattern has fairly orthogonal boundaries. What you could do, is deform not grid cell outlines, but just vertical centre lines per cell. Then pick some lines and remove their middle third. Finally, convert each line segment into a box.

1 Like

I roughed it out and I’m now point editing. It’s slow as hell but I’ll get a decent result.

Is it cheating to use Image Sampler? I adapted this model posted three weeks ago to make peaks where the black spots are, then slice them with an adjustable height plane and cull unclosed curves. It can be refined in various ways…

Image_Sampler_2019Nov17a.gh (449.7 KB)

Due to a very old GH bug, the Image Sampler will likely be missing when you open this file, so you need to replace it using the image you posted. Note the “Color brightness” setting.

P.S. Image dimensions are 512 X 930 so 512 is the maximum on the ‘resolution’ slider. Turns out that the Boundary surface is incredibly slow at this resolution; eight minutes on my old laptop in R5. Otherwise, except for Brep | Plane (Sec, 40 sec.), it’s reasonably fast so I suggest disabling Boundary until you get the cross sections you like first.

I tried cleaning it up in Photoshop then tracing it in Illustrator. I got similar results as with your image sampler. The corners loose too much sharpness. Thanks

Added Round to act as a “threshold” (zero or one, cyan group) and lowered the ‘Z cut’ cutting plane from 0.75 to 0.4. Looks better to me? As I sad, the idea can be refined in various ways. What is the picture? An overhead view of a city?

Image_Sampler_2019Nov18a.gh (454.4 KB)

Compare to yesterday’s version (below):

Baked and rendered surface with ‘Z scale’ = 2.0 and ‘Z cut’ = 0.4. White lines show “cuts”:

This is expectation.

I’m getting close with random curves and offset then using Pull back onto a Heightfield surface.