??? 05/01/09 07:54 Read: times |
#164979 - Today, rasters are hard for microcontrollers Responding to: ???'s previous message |
Richard Erlacher said:
We do have to cover both options, and, while the vector method is brighter with less power, the raster is, IMHO, easier to generate, particularly for complex images. The raster concept is very simple when looking at external hardware, but it is normally only easy to generate for static or semistatic (such as scrolling text) images. Moving video requires huge amounts of data. And it takes a lot of processing power to decompress an MPEG-encoded video stream. Directly rendering vectors or polygons into a frame buffer would normally require a very fast processor, low resolution or a graphics accelerator because of the large number of pixels involved to get a reasonable resolution. For the type of images that a laser projector is good at (given the limitation of a laser to do area fills without loosing intensity and frame rate) a vector display is really excellent. Home-built vector displays maybe manages up to 5kpoints/seconds with a few exceptional home-built galvos reaching 15kpoints/second, while commercial systems manages 30k+. Within that physical limiation, you can get very bright images with very low processor load from a vector display. A PC is normally not suitable because of the real-time requirements, but it can output the data to a microcontroller, and an 8051 or small ARM7 can manage to store one or more frames in RAM. A microcontroller can also manage to store full animations on a memory card, without having problems with either the total size of the data, or the bandwidth to extract it from the memory card. The vector output system will basically form a graphics accelerator, because of the low-pass filtering introduced in the system when turning the mirrors to try and reach the next target point in the graph. And even as little as 8-bit data outputs will look like almost infinite resolution since the emitted data isn't gridded into a visible raster. Between the target coordinates, you will see smooth (but possibly not straight) lines, instead of the jagginess you get for diagonal lines on a raster display. So while CRT, LED, LCD or plasma technologies are excellent for implementing a raster display, we still have a bit to go until laser projects will be suitable for generic raster data. I'm pretty sure that will will reach a stage when it will be possible to build high-quality HD-resolution laser projectors for front or back projection. But we will need stronger lasers at an affordable price. For such a display, a rotating mirror is excellent. It is very easy to monitor a rotating mirror and instantly cut the laser output if the mirror isn't keeping the correct speed, thereby protecting from eye damage or fires. Right now, the current state of the art - and the current state a hobbyist can reach - a raster display wouldn't be a good choice unless run at low resolutions - such as the limited vertical resolutions you can get away with for a text-only display with few text lines. |