Color Manipulation

BYOND Forums

Announcements · BYOND Help · Bug Reports · Feature Requests · Beta Testers · Beta Bugs · Developer Help · Design Philosophy · Demos & Libraries · Tutorials & Snippets · Art & Sound · Classified Ads · Game Updates · Contests & Events · Linux Talk · On Topic · Off Topic

Color Manipulation

ID:151770

May 1 2009, 4:30 am

CriticalBotch

I'm not really up on color theory and mechanics. I can use RGB or Hex color codes with no problem, but things like HSL/HSV aren't so clear for me.

My situation is that I want to be able to effect color changes, starting from blue, going into white, to yellow, and finally ending in red.

There would be about 70 distinct color changes, which I could hard code if it really came down to it. In the meantime, I'd like to see what others thought might be a good way to approach the problem.

I've considered starting at RGB 0,0,255 and add 25.5 R and G to the mix until it hits white. From there remove 25.5 B until yellow is achieved, and then remove 25.5 G until red is achieved.

It strikes me that this may be better suited for HSL than RGB though. I found one library that might help [MobiusEvalon.colourfunctions] but I get the following error when trying to download it:

Downloading Colour Functions (9989 bytes).
Failed to download data (HTTP server returned unexpected status: 404 Not Found).

Thoughts and/or suggestions?

May 1 2009, 6:01 am

Lummox JR

I've got color manipulation functions in LummoxJR.IconProcs including HSV, and procs that will handle the blending for you.

As far as the transitions you suggested, It won't make any difference whether you use HSV or RGB since blue⇒whiterArr;yellowrArr;red can all be done with RGB blending just as easily. Because you're modifying only one or two channels and in a uniform manner, HSV would give you the same results. Where HSV is best used is when you're blending between disparate colors like red and green and want to keep the same brightness, where strict RGB blending would give you a darker gold color when you reached the halfway point between red and green.

Lummox JR

May 1 2009, 6:55 am

In response to Lummox JR

CriticalBotch

Lummox JR wrote:

I've got color manipulation functions in LummoxJR.IconProcs including HSV, and procs that will handle the blending for you.

As far as the transitions you suggested, It won't make any difference whether you use HSV or RGB since blue⇒whiterArr;yellowrArr;red can all be done with RGB blending just as easily. Because you're modifying only one or two channels and in a uniform manner, HSV would give you the same results. Where HSV is best used is when you're blending between disparate colors like red and green and want to keep the same brightness, where strict RGB blending would give you a darker gold color when you reached the halfway point between red and green.

Lummox JR

Thanks Lummox. I'm basically trying to automate generating star color based on spectral classification (OBAFGKM), given intervals of 0 to 9 (F5 being half way to G0). F is a yellow star while G is an orange star. Naturally an F5 will be redder than an F0, but not as red as a G0.

I think I've figured out what I'll be doing, and your library will help, thank you. I'll be defining the base color for a given class. I'll get the difference between that and the next class in R, G, and B format, divide by 10, and multiply by the subclass.

So, for an F5, the base color is 255,255,0 and the next color is 255,128,0. The difference is 0,-127,0. 1/10th of that is 0,-12.7,0, and multiplied by the subclass (five) is -63.5, giving me a color of 255,191.5,0.

[EDIT] Or I can just use your blend function with the subclass as a decimal (0.0 to 0.9).

May 1 2009, 7:50 am In response to CriticalBotch
Lummox JR	To handle star spectra, you might also want to consider not using RGB values directly, but using raw numerics. These you can push past full saturation, so drawing a halo around a bright star would just be a matter of pulling the values down a bit. Of course if you're drawing each star as a point source, that's moot. Lummox JR

May 1 2009, 8:00 am

In response to Lummox JR

CriticalBotch

Lummox JR wrote:

To handle star spectra, you might also want to consider not using RGB values directly, but using raw numerics. These you can push past full saturation, so drawing a halo around a bright star would just be a matter of pulling the values down a bit. Of course if you're drawing each star as a point source, that's moot.

Lummox JR

Hmm, please elaborate?

At the moment I'm trying to get caught up on icon manipulation so I can draw the star object at run time (still considering how I want to handle mapping solar radii to the 32x32 - or multiples of such - icon grid). Any advice you might be able to provide would be very helpful.

Relative ranges for radii are tied to the spectral classes (O through M), and range roughly from 0 to 6.6 and up (I went with an 8.0 cap to better work with the 32 base, which roughly maps 2 pixels per radii).

As I understand it, all the attributes of a star can be derived from it's spectral classification. Mass and radii have a floor and ceiling value based on the current spectral class, and we can derive where (with some variance) between the current class and the next in the same manner as I was getting the color change from one to the next.

May 1 2009, 8:12 am

In response to CriticalBotch

Hobnob

CriticalBotch wrote:

Thanks Lummox. I'm basically trying to automate generating star color based on spectral classification (OBAFGKM), given intervals of 0 to 9 (F5 being half way to G0). F is a yellow star while G is an orange star. Naturally an F5 will be redder than an F0, but not as red as a G0.

Interesting. You might find this page useful. Even if you just use a linear approximation between stellar classes, it could give you some useful RGB values to interpolate between.

May 1 2009, 9:03 am (Edited on May 1 2009, 9:28 am)

In response to Hobnob

CriticalBotch

Hobnob wrote:

CriticalBotch wrote:

Thanks Lummox. I'm basically trying to automate generating star color based on spectral classification (OBAFGKM), given intervals of 0 to 9 (F5 being half way to G0). F is a yellow star while G is an orange star. Naturally an F5 will be redder than an F0, but not as red as a G0.

Interesting. You might find this page useful. Even if you just use a linear approximation between stellar classes, it could give you some useful RGB values to interpolate between.

Hmm, thanks for the site. I've been using the Wikipedia entry which seems redshifted from the colors on the site you provided. I'm not sure if that's due to atmospheric interference on the part of the Wikipedia entry or something else, but I believe I'll stick with the Wikipedia values (I like my "red dwarfs" to be red, lol).

I ran a test, using the following values for each class' base:
O - #000080
B - #8080FF
A = #FFFFFF
F = #FFFF80
G = #FFFF00
K = #FF8000
M = #FF0000
X = #400000

X, obviously, isn't a true spectral class. I include it to provide shifting from M0 to M9 and generate a brownish color for brown dwarfs.

In any event, using Lummox's library's BlendRGB(Base, NextBase, SubClass/10) worked wonderfully.

 Class O0 Star [#000080]
 Class O1 Star [#0d0d8d]
 Class O2 Star [#1a1a99]
 Class O3 Star [#2626a6]
 Class O4 Star [#3333b3]
 Class O5 Star [#4040c0]
 Class O6 Star [#4d4dcc]
 Class O7 Star [#5a5ad9]
 Class O8 Star [#6666e6]
 Class O9 Star [#7373f2]
 Class B0 Star [#8080ff]
 Class B1 Star [#8d8dff]
 Class B2 Star [#9999ff]
 Class B3 Star [#a6a6ff]
 Class B4 Star [#b3b3ff]
 Class B5 Star [#c0c0ff]
 Class B6 Star [#ccccff]
 Class B7 Star [#d9d9ff]
 Class B8 Star [#e6e6ff]
 Class B9 Star [#f2f2ff]
 Class A0 Star [#ffffff]
 Class A1 Star [#fffff2]
 Class A2 Star [#ffffe6]
 Class A3 Star [#ffffd9]
 Class A4 Star [#ffffcc]
 Class A5 Star [#ffffc0]
 Class A6 Star [#ffffb3]
 Class A7 Star [#ffffa6]
 Class A8 Star [#ffff99]
 Class A9 Star [#ffff8d]
 Class F0 Star [#ffff80]
 Class F1 Star [#ffff73]
 Class F2 Star [#ffff66]
 Class F3 Star [#ffff5a]
 Class F4 Star [#ffff4d]
 Class F5 Star [#ffff40]
 Class F6 Star [#ffff33]
 Class F7 Star [#ffff26]
 Class F8 Star [#ffff1a]
 Class F9 Star [#ffff0d]
 Class G0 Star [#ffff00]
 Class G1 Star [#fff200]
 Class G2 Star [#ffe600]
 Class G3 Star [#ffd900]
 Class G4 Star [#ffcc00]
 Class G5 Star [#ffc000]
 Class G6 Star [#ffb300]
 Class G7 Star [#ffa600]
 Class G8 Star [#ff9900]
 Class G9 Star [#ff8d00]
 Class K0 Star [#ff8000]
 Class K1 Star [#ff7300]
 Class K2 Star [#ff6600]
 Class K3 Star [#ff5a00]
 Class K4 Star [#ff4d00]
 Class K5 Star [#ff4000]
 Class K6 Star [#ff3300]
 Class K7 Star [#ff2600]
 Class K8 Star [#ff1a00]
 Class K9 Star [#ff0d00]
 Class M0 Star [#ff0000]
 Class M1 Star [#ec0000]
 Class M2 Star [#d90000]
 Class M3 Star [#c60000]
 Class M4 Star [#b30000]
 Class M5 Star [#a00000]
 Class M6 Star [#8c0000]
 Class M7 Star [#790000]
 Class M8 Star [#660000]
 Class M9 Star [#530000]

Obviously there's some spectral classifications here that don't really occur with any frequency, but given that it's for video game I think some concession can be made in favor of ease of use.

[edited as I changed the base O and X values to provide a more useful spectrum, and included the hex values being used]