[florin]

I hope I'm posting in the right place.

I'm writing a simulation software related to gravity and general relativity. The software is supposed to start by rendering the image of empty space, with stars as background.
I am planning to start with an image size similar to HD (1920x1080 pixels) and a number of stars around 3k or so.

I will have to render each star as a small disk, a few pixels in diameter. The goal is to create a photo-realistic image.
The image should look like it's captured by a very sensitive HD camera on a spaceship, somewhere in our galaxy but far away from any particular star. I don't want to render nebulae or galaxies, just stars at this moment.

My question is: when looking at the image of a single star captured with a digital sensor, how does the brightness of the pixels decrease from the center of the star image to the edge? Is it something like a bell curve? (bright in the middle, then decreasing to the edge)

How should I decrease the brightness in a natural way from center to edge? Does it decrease in a uniform way, just a smooth uniform decrease from center to edge? Or is it more or less constant and then it drops off suddenly?
Also, at some point there has to be a cut-off point, where I should stop rendering if the pixel is too far from the center. Something like a radius of the star image. Is that a sharp edge, or does it fade off gently to black?
I have no idea how a star looks like on a digital photo, that's why I'm asking these questions which should be pretty simple for y'all.



My other question: the colors of the stars should also be photo-realistic. I will store each pixel as an RGB triad, so I just have to assign various hues to each star. But the thing is, what are the realistic hues, and how many stars of each hue should I generate?
I noticed there aren't any green stars. Most of them are reddish, blueish or white. The saturation is pretty low. But I have no idea how to create a realistic distribution of hues and saturations.


I guess that's it for the moment. If I could get some info on these two crucial issues, I could start implementing the initialization routines and actually start rendering some images.

[sgcullen]

Some answers:

Q1. Yes, the star brightness is generally a flat topped curve (white clipped) if you were to graph it from the center to the outside. You should check out some of the astro images here and also on LightBuckets.com. This will give you a good idea of how they appear to a CCD. When imaging stars, keep in mind you are only imaging a point of light, not the star itself. Stars are too far away to actually image their disk. I have attached 3 images - an image of a typical star, the star's brightness curve, and the area plot of the star.

Q2. There are no green stars. Star color is a function of the star's temperature. There is some good reading on the subject at The Definition of Color in Astronomy. Keep hitting the Next button to go to even more information. To get some idea of the distribution of colors read Hertzsprung-Russell diagram - Wikipedia, the free encyclopedia. The H-R diagram of 22,000 stars is a small sample but should give you some empirical data to support your rendering.