4 ms·
I would look at tomographic reconstruction (https://en.wikipedia.org/wiki/Tomographic_reconstruction https://en.wikipedia.org/wiki/Tomographic_reconstruction) a
by Someone 10y ago
I would look at tomographic reconstruction (https://en.wikipedia.org/wiki/Tomographic_reconstruction https://en.wikipedia.org/wiki/Tomographic_reconstruction) algorithms for this for inspiration. This is a very similar problem: what projections best approximate a given image?
- saboot 10y agoThis is a good idea, each thread is basically a single point in the 2D fourier transform of the image so an approximate image would sample that space accurately. Not sure what accurately means exactly since the fourier space isn't 1 or 0 (thread or no thread). Maybe include dithering to approximate?
- murbard2 10y agoI think you're thinking of the Hough transform, not the Fourier transform. https://en.wikipedia.org/wiki/Hough_transform https://en.wikipedia.org/wiki/Hough_transform
- saboot 10y agoNo, I am not https://en.wikipedia.org/wiki/Projection-slice_theorem https://en.wikipedia.org/wiki/Projection-slice_theorem
- wyager 10y agoHow do you figure that each thread is a point in the 2D Fourier transform? A thread looks nothing like the ifft of a single non-zero pixel in Fourier space.
- theveloped 10y agoSweet! I had not thought of tackeling the problem from this angle. This might indeed give for some very interesting results!
- etrautmann 10y agoI had the same thought - VERY cool project and fantastic that you took it all the way through the automated fabrication. Impressive already, but I bet you can improve the optimization to get even better results.