RE: PiCasso Design Challenge - POV Globe Blog #13

balearicdynamics — Sun, 02 Jun 2019 06:36:33 GMT

Dale,

sorry if I don't remember if you wrote it or not. Did you include in your project some sort of rotating encoder to know the position of the motor? I think no if I am not wrong as this will add an extra issue to the data transferring, powering the encoder, etc.

I am thinking of a nice idea developed by [mention:b0bc65b9ecdc4307bd967592f00e340a:e9ed411860ed4f2ba0265705b8793d05] to calculate the rotation speed of something just adding a marker. As the LEDs support is done this will be perfect just with a sticker. I mean this project: Cyclops-1000: An Electronic Eye for Rotational Speed Measurement

Well, knowing the instantaneous RPM and the diameter of the circle you know in every moment where the rotating circle is. By the other side, as the LEDs are regularly disposed along the circumference, you have a fixed angular step.

Making a generic function (Python will be perfect, it's only math) that project a point on a sphere, you should just adjust the polar 3D coordinates of every pixel where the azimuth angle is fixed (few data in a table as not only is fixed but is a fixed step - the LEDs angular distribution) and the horizontal if function of sigma-Ro (sorry I don't know how to write Greek symbols) related to the angular moment you can easily update on every rotation. The advantage of the Cyclops is just that it is a small custom sensor that can be applied externally.

IMHO with this kind of approach you are able to project any kind of 3D image enveloped over a sphere, also real-time.

The issue: there are two kinds of approach in converting polar to cartesian and vice-versa coordinates. The first is using trigonometry but I have some doubt that you can reach the needed speed. Instead, as there are a lot of fixed conditions, you can do the conversion very fast using only primary operations (multiply/divide and add/subtract, done fast by the processor directly).

If you are interested in this kind of approach in past I have developed this kind of algorithm for a geodetic problem, identifying less then 1sqm of a surface of the Earth based on the polar coordinates of a larger area provided in coords by Google map. (G-map has a too low resolution on the terrain). It was developed in PHP but it was very fast as had to run on a server. Let me know what do you think.

Enrico

RE: PiCasso Design Challenge - POV Globe Blog #13

dixonselvan — Sun, 02 Jun 2019 02:57:14 GMT

That’s some nice progress [mention:1bc1ed25c74e4385bcbca428cd9ea7e7:e9ed411860ed4f2ba0265705b8793d05]