Dave Tweed wrote: > From: Stef Mientki > >> Dave Tweed wrote: >> > The 18" Eyeball spins at 30 revs/sec and can put about 500 pixels >> around >> > the circumference, giving a pixel rate of 15000/sec. or 67 us/pixel. A >> > turn-off time on the order of 10 us would not be too slow. But it >> uses RGB >> > LEDs (no phosphors), so it wasn't even an issue for us. >> > >> > I don't recall how this thread started, but if the application is >> for a >> > planetarium (projected on the inside of a dome), >> >> That's exactly the idea, >> although we're not sure yet to use projection, because we'll loose a >> lot of brightness. >> >> > you'll probably want a >> > *lot* more pixel density than that -- at least 10x linear density, >> if not >> > 100x, which begins to approach the spatial resolution of the human eye >> > (about 200 pixels/degree, or 6e-9 sr/pixel). This means somewhere >> between >> > 10e6 and 1e9 total pixels in a half-sphere. >> >> If you look at a star-map on a computerscreen, the stars vary >> (depending on the brightness of the star) from 0.02 .. 0.5 degrees (I >> mean star-degress, but don't know if that's a good english word). >> Looking at a starmap on paper, the size varies from 1 .. 2 degrees. > > > I was thinking about this a bit more after I sent my previous message. > My laptop has a pixel density of about 75/inch, and I sit about 18" away, > giving an optical density of about 0.042 degree/pixel, or 24 > pixels/degree. > Such a resoltion would be nice: 24 LEDs per degree. But also due to the large beamwidth of LEDs this seems not very oppetune. >> So I think we get a quiet good image if we set our goal at 1 degree >> for the brightest stars. In that case we need about 100 white LEDs >> over a 90 degrees arc. To also display moon and planets, we need >> another 100 yellow or orange LEDs. > > > But it isn't just a question of how big the stars are, but also how > accurately you can position them in the "sky". > > And a planetarium show isn't just a big star map; the goal is to > reconstruct the actual appearance of the night sky. Excessive pixellation > will destroy that illusion very quickly. > I agree, but it depends on what your goals are, and what is practical realizable. Our main goals are (some are just dreams for the moment ;-) 1. show the major stars at the current location at the current time (In Netherlands you cann't see too many stars ;-) 2. show the moon and planets (maybe even some satalites) 3. show the accelarated movement of stars, moon and planets 4. show the effect of light polution 5. with some kind of pointer, point an area at the dome (I read somewhere that LEDs can be used as light sensor, or the old PC-pen trick), and the telescope outside moves to the selected area The point I'm worried at most is a smooth accelarated movement of the stars, thus the 1 degree resolution. >> If we're aiming at a sphere with a diameter of 40 cm and use 3 mm >> LEDs, we can't get 100 LEDs on a quarter, so we need in totally 4 >> rows of each 50 LEDs. > > > Huh? You should be able to place 419 x 3mm LEDs around the > circumference of > a 40 cm circle. If you take 1 arm of the propellor clock, it covers 90 degrees of a half sphere. With a diameter of 40 cm, thats radius * pi / 4 = 20*pi/2 = 30 cm. The 3 mm LEDS have en border of 3.9 mm. If want to drilll holes in an arc, I need some inbetween space and must count the LEDs are not parallel, taken up some extra space. So I estimate the central distance of he LEDs at 5 mm. So I can place 60 LEDS at an arc of 90 degrees. > >> Now fortunatly stars twinkle, so we can choose a lower rotation >> speed, than in normal propellor clocks, maybe even as low as 5 rev/sec. > > > But stars twinkle randomly, not periodically. We've noticed that, so an improvement would probably be to add random noise (to the on-time). > I've seen the Eyeball running > at 15 to 20 revs/sec., and it gets really hard to maintain the > illusion of > any sort of continuous image at that flicker rate. Well maybe we should increase the speed. I think I read somewhere that the eyeball, iball, or whatever they are called, runned at 30 rev / sec. Is that correct ? Stef Mientki -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist