To generate them equally spaced out based on the number of colors wanted. Something that looks like this if 8 is given for the count specified:
List
You can also take a look at Rich Newman's HSLColor class. He has a series of blog posts on it, starting with this one.
Using this code I was able to generate a series of colors evenly spaced around the color wheel, but you'll have to add additional logic if you want to include shades of grey.
private void button1_Click(object sender, EventArgs e)
{
listView1.Items.Clear();
int step = 240 / 8;
for (int i = 0; i < 240; i += step)
{
HSLColor color = new HSLColor((double)i, 240, 160);
listView1.Items.Add(i.ToString()).BackColor = color;
}
}
You'll find colors easier to work with in these sorts of problems if you use HSV instead of RGB.
"equally spaced colors" almost always means "equally spaced hues". So, for [0,360) in hue, you just equally space by dividing that range equally.
Now you have a hue, and you just need to find the "pastel" version of that hue. To me, this means desaturating the color a bit. I'd say to 80% saturated for starters.
In my tests, I used 100% for value. Then just convert to RGB. Here's what I've been playing with:
<body>
<script>
// taken from http://ariya.blogspot.com/2008/07/converting-between-hsl-and-hsv.html
function hsv_to_hsl(s, v)
{
var ss, ll;
ll = (2. - s) * v;
ss = 1. * s * v;
ss /= (ll <= 1.) ? (ll) : 2. - ll;
ll /= 2.;
return [ss, ll];
}
function do_colors(sat, light)
{
n = 15;
document.write(n + " colors at " + sat + "% sat, " + light + "% lightness<br />");
for(var x = 0; x < n; ++x)
{
hue = 360.0 / n * x;
html_chunk = "<div style='width: 50px; height: 50px; display: inline-block; background: hsl(" + hue + ", " + sat + "%, " + light + "%);'> </div>";
document.write(html_chunk);
}
document.write("<br />");
}
do_colors(100, 50);
do_colors(95, 75);
do_colors(75, 50);
do_colors(100, 35);
// rudimentary averages from your colors
sl = hsv_to_hsl(.7, .9);
s = sl[0] * 100;
l = sl[1] * 100;
do_colors(s, l);
do_colors(75, 60);
</script>
</body>
Not C#, I know, but just trying to nail the light & sat down.
Otherwise, you could look at your sample colors, and see if there is any correlation in the HSV/HSL values, and try to derive an algorithm from that. If you plot S/H and V/H, you'll see a large dip in the graph at the grey color --- it seems to be an outlier. (Third from left on bottom row.) Ignoring that value, S is about at 75% and value is just under 90%. Using those values probably gave the nicest result.
Link: http://jsfiddle.net/ZHyAQ/