This is basically visualization 101. 'Turning data into images' isn't exactly a new concept. Sometimes I think the next generation of programmers has gotten too good at thinking up fancy names for reinventing the wheel and hasn't spent enough time looking at what has been done in the past 50 years so they can build off of it.
Also there are links to 'visualizing big data' and 'visualizing billions of points'. These too have been problems solved a long time ago with a lot less resources. The book advanced renderman are is a good place to start, and of course, opengl uses basic z-buffer rendering.
Unless you need sorted transparency (and accumulating points for brightness definitely does not), then you can rasterize as much as you want in any order, without holding anything in memory except for the image buffer. It doesn't matter how many point/particles you have, you don't have to do anything to make it scale.
Also there are links to 'visualizing big data' and 'visualizing billions of points'. These too have been problems solved a long time ago with a lot less resources. The book advanced renderman are is a good place to start, and of course, opengl uses basic z-buffer rendering.
Unless you need sorted transparency (and accumulating points for brightness definitely does not), then you can rasterize as much as you want in any order, without holding anything in memory except for the image buffer. It doesn't matter how many point/particles you have, you don't have to do anything to make it scale.