[MOC] image transformer

[Fissl]

Hello everybody!

Last winter I started to study 'digital media' at the university of the arts in my hometown. I took part in a course where the task was to develop a generative design process.

So I've built this copy machine that transforms black & white images into some kind of bar code.

The color sensor of the NXT 'reads' the image concentrically and transmits the information to a motor which then lowers a felt pen on a paper. After every complete circle the sensor moves a bit away from the center and the paper moves a bit forward.

Beside the felt pen and two little magnets which keep the image in place it's pure lego ;)

Facts:

approx. 1900 parts

1 x NXT 2.0

3x NXT motor

2x color sensor

2x touch sensor

1x PF XL Motor

1x PF switch

1x PF extension wire

[DrJB]

Very nicely done. While I definitely admire the lego adaptation, i must say I'm thinking primarily of the math behind the transformation. If I understand correctly, the transformation is purely reversible, correct? Also, and with a simple Matlab code, one can take an image, put the origin of a reference frame in the middle, use polar coordinates, and the circles in such frame become your bars in the print. Thus, this transformation is not limited to black and white only but any color image can be transformed (conformal mapping). I'm thinking a simple iPad app where one can experiment with various images and transformations ... but I must stop as I'm getting way off topic.

Again, very nice contraption.

[Fissl]

Yes theoretically its reversible, but beacuse the image is read in only 40 coaxial circles and the accuracy of the color sensor isn't the best a lot of information gets lost . Also with every step away from the center the radial velocity increases thus the accuracy decreases even more. Initially I wanted to transform the printed bar code again and then transform the new bar code again and again and again... but the sensor didn't recognize the lines because they were too thin.