The Dimensions of Colour - a colour theory discussion thread

[curtmantle]

David - just a thanks for all the effort and hard work put into this site.

I have to admit that for someone who knows absolutely nothing about the concepts (and was rubbish at physics at school), it's extremely hard-going but that isn't your fault at all... it's a complex subject. Ever thought of doing a "Complete Idiot's Guide"...

... you're going to tell me that this IS the complete idiot's guide aren't you?

I will persevere - I'm sure it'll all start to click in place eventually and I have already learnt much from it. Well done.

[senany11]

Thank you, this looks wonderful
some really great info you've put together!

__________________
makepolo

[briggsy@ashtons]

Thanks again for the replies! Here's a colour exercise that I thought some people here might like to try.

I've just finished teaching a short course on colour and light at Billy Blue in Sydney, which was the first one I have given specifically focused on the digital environment. (The workshops I give at Ashton's cater more for traditional media). Towards the end of the course I introduced a variation on the old sphere painting exercise, where I got the students to analyze photographs (ripped off from Flickr) for lighting and atmosphere, and then paint in a sphere so that it looked compatible. You need to decide on the directions, sizes, colours, and relative brightnesses of the main and secondary light sources, and then paint the modeling, highlights, form shadows and cast shadows accordingly. I asked the students to colour the spheres so that they looked bright coloured, that is, neither glowing nor greyed, but you could go on to make them look luminous as a variation. I also worked out an OK way of faking spherical reflections based on the content of the photos. Later we did some misty scenes that required addition of atmospheric perspective. You might also want to look at the image characteristics of the photo and perhaps add some grain and/or blur
to really get the spheres to blend in, though this was not really the focus of my course.

I found this exercise to be really good for drawing together the stuff we had covered on light and colour, as well as all the ways of choosing and adjusting colour in Photoshop.

Anyone care to give it a try?

[Form]

wow briggsy, what a cool exercise! i wish we had such a cool teacher when I was there at billy blue. it warms my heart to see that you are still teaching there and that you have such responsive students!

i have a lot of news to share, i will have to come see you sometime and give you the lowdown!! would be good to catch up. are you around during any of the drop in sessions at ashtons?

[briggsy@ashtons]

Hi Adam, it'd be great to catch up! This week I should be at Ashton's on Wednesday night and all day Saturday and Sunday, if any of those times suit. After that there's a break over Christmas until my first workshop there on the 29th.

Unfortunately this year may have been my last undergraduate class at BB, at least for a while. The new course has a lot less drawing in it, and with Debby, Edwina and a couple of others they have more drawing teachers than they need. Hopefully something will turn up there eventually as I really love the place, and the students have been great. I'm still down to teach my colour "masterclasses" for the general public there next year, so at least I'll have some involvement.

[GDonaldson]

woah I had no idea colour could be so complicated! very well documented resource you have here briggsy. Honestly I dont get it, I just draw what seems right to me, no identifiable method or theory...so when I read this stuff I have no clue how to apply it in a practical manner. I did art at school and have an honours degree in design for interactive media but no one has taught us anything like this about colour.

I assume that I have got some theorys or methods working in my head but I just havent really identified what they are myself. I do alot of 3D so I suppose some of my ideas come from trying to be a human 3D rendering device haha.

[briggsy@ashtons]

For something so important it's amazing how badly colour is taught - almost universally - which is why I put the site together. Of course, how much you need to know depends on how awesome you want to get at it, and what you're satisfied with. I think an exercise like the one I just described is a good test of what someone really knows about colour and light. I'm sure anyone who tries to get a really realistic fit will see where most of the stuff I talk about comes in.

[dose]

Oh- great exercise. I'm going to try to do it on my breaks between work...

[briggsy@ashtons]

Great, Tim - please post what you come up with!

Anyone else - don't be afraid to give it a shot. Happy to help if I can.

[dose]

Work has been crazy lately, but I've been chipping away at the exercise. Will hopefully have something to post soon.

On an unrelated note, I have two rather technical questions. First, what is the function to convert from linear radiance to nonlinear brightness? I saw the charts on the "Effect of Inclination to Light" and "Effect of Distance from Light" pages, but I'd like to play around with some values other than what you have listed (and it's clearly not a simple function). I tried googling it, but it mostly left my head spinning...

Second, how would one apply those brightness values in Photoshop? I understand that in the absence of any ambient or reflected light (i.e. space) the brightness for any surface would drop to zero. But how does the scale shift in the presence of ambient light? Does it just shift up proportionally? And how do you account for the local value of the surface?

(I guess that's more than two questions...)

[briggsy@ashtons]

On an unrelated note, I have two rather technical questions. First, what is the function to convert from linear radiance to nonlinear brightness? I saw the charts on the "Effect of Inclination to Light" and "Effect of Distance from Light" pages, but I'd like to play around with some values other than what you have listed (and it's clearly not a simple function). I tried googling it, but it mostly left my head spinning...

Technically, radiance is the raw amount of light energy, so to convert to perceived brightness this first needs to be weighted, wavelength by wavelength, to the sensitivity of human vision, giving CIE luminance or Y.

The conversion from luminance to perceived brightness has been modeled using power functions ranging from a cube root (i.e. an exponent or gamma of 0.33) to a square root (an exponent of 0.5). Measurements on one of the original Munsell atlases (1915) and on the first Munsell Book of Colour (1929) have shown that their greyscales quite closely fit exponents of 0.43 and 0.40 respectively. For the 1940's renotations, greyscales were experimentally judged against white, grey and black backgrounds, and the (very varied) results were smoothed into a complex polynomial that someone later noticed was very close to a cube root relationship. This cube root relationship was incorporated into the formula for CIE lightness (L*) and presumably in its equivalent (L) in Lab space in Photoshop.

On the other hand the conversion from linear to nonlinear RGB brightnesses, I gather from Charles Poynton's excellent Color FAQ, is based on a simple power function with an exponent of 0.45. Consequently, a series of greys that are evenly spaced in brightness (B) in Photoshop are not exactly evenly spaced in L.

I've taken you to or perhaps beyond the outer fringes of my current knowledge of the subject, so if anyone understands this stuff better than I do please chip in. In any case, as I say on the site, you only need to think about this conversion if you want to exactly calculate the fall-off of light with distance and inclination. For most purposes, including this exercise, you can just eyeball.

Second, how would one apply those brightness values in Photoshop? I understand that in the absence of any ambient or reflected light (i.e. space) the brightness for any surface would drop to zero. But how does the scale shift in the presence of ambient light? Does it just shift up proportionally?
(I guess that's more than two questions...)

You can emulate the effects of multiple light sources in Photoshop by superimposing layers showing the effects of each source in either screen mode or linear dodge mode. Linear dodge is more accurate, but screen is less inclined to reach the limit of the RGB gamut and "clip".

And how do you account for the local value of the surface?
(I guess that's more than two questions...)

(As I suggested to what must have been another dose about a page back) you can apply this with a top layer in multiply mode.

[dose]

Awesome- thanks David! That gives me a bunch to experiment with.

For most purposes, including this exercise, you can just eyeball.

Yeah, I totally agree that I don't need to know this stuff for practical painting purposes. But I suffer from a terrible need to understand things, so I just couldn't leave this alone!

I also think the model I have in my head for shading series is incorrect- in my head I think the steps are much more even than they really are. For example, you mention that the full light area can be rather large, but in my head it's pretty small. I think a couple very precise renderings could really clear up the flawed model in my head, and after that I probably wouldn't ever need to be so accurate again.

Further, I've been experimenting with some scripts for Photoshop that will generate strings of swatches to color pick from. It's pretty trivial to do evenly stepped swatches, but I might like to experiment with generating strings based on the inclination to light- so each swatch corresponds to a certain inclination to the light. I'm not sure what will come of it, but I'm curious.

I don't imagine I'll be using much of this for the photo exercise, but like I said, I can't leave it alone...

[dose]

Here's my first attempt. I made a green sphere for variety, and to be able to see the effects of the yellowish/orangish light better...

[briggsy@ashtons]

Great work on the colours, Tim, but looking at the shadows of the trees I'd make the cast shadow on the wall wider. The alignment of the terminator and highlight also look just a little out.

[dose]

Yeah, I'll admit I was a bit sloppy on the drawing side of things- I just eyeballed pretty much everything. That's a really interesting method to find the placement of things, though- never would have thought to apply something like that here. It would have been probably been much less effort than eyeballing.

My high school technical drafting teacher would be disappointed!

[briggsy@ashtons]

I've already written to them.

Now for some aerial perspective:

[Ian Miles]

Just found this Thread!

Here´s one of my first trys at painting!! ^^
Made with painter. I know is not the best for photorealism, but is what I have and the painter look is my fav!

[Elwell]

Ian, I'd pick another photo to work with. That one's been highly manipulated, the sky doesn't match the lighting of the ground.

[Idiot Apathy]

Hey Briggsy, hope you are well!
I still have a hard time lighting warm local colors with cool light, if you ever had the time to demonstrate with a sphere or two I'd be in your debt (again). Understandably in many situations the local color of the sphere wouldn't be apparent in the slightest, but I suppose I am looking more for the situations where there is an obvious transition - perhaps multiple lightsource situations? Tricky business...

[briggsy@ashtons]

Sure, Tom. Just tell me the colour of the main light, secondary light and local colour that you want, and I'll give it a go! My own personal Peer Project exercise, yay!

[Idiot Apathy]

Haha, yayyyyy!

Hmm, let's see. Cyan main light, red secondary and an orange local colour. If that gets too ugly due to the complements maybe the secondary should be white or orange?

Thank you!

[briggsy@ashtons]

How's this look? I've left out the complexities of the reflected light since we're mainly looking at the colours. Here's the PSD file so you can see my logic, and tweak the hues to your heart's content:

http://djcbriggs.googlepages.com/fortom.psd

[Delmar]

This is absolutely amazing!!! I wanna thank you so much for this site and your effort and knowledge you've put in.

I'm also one of that guys who tries to get along with that little bit of color theory knowledge hoarded along the time....wondering why my colored stuff isn't quite working......

I should be working right now, so I'm not able to get through the whole stuff...but i'm really looking forward to read it!!!.

Thanks a lot, i guess i will be back with some questions soon.

Cheers

Andi

[theTooninator]

You're a big man for trying to help us shmucks along. Thanks briggsy.

I've recently been hit with the colour bug. It's all I see anymore. Trying to observe light and colour closer than I ever have before, in hopes to apply it to my work. This thread will serve nicely in helping me along. I'll try to follow along.


-Matt

[germ-X]

awesome color works. impressed. love to see more of your work.

[Idiot Apathy]

Briggsy Sir, thanks so much for that sphere. It's opened my eyes to a lot of things. A bit of a renaissance going over here in my head lately

[George Abraham]

Thanks MR Briggs.

Your site is a really good resource that saves alot of time.

I had some raytracing stuff I did yonks ago and played around with words that didn't make sense to me at the time. Now they do.

Just one question. On a dull ball without a specular hilight but with a diffuse effect of saturation change(Closer to the color of the light source).
Does the degree of change have to match the half light and full light direction of the sphere or does it follow the direction of reflecton the specular would have had? In otherwords to simply see the effect as a large dull specular?

[EDIT]
I looked at my hand and the answer came to me. Because the texture is distributed (little bums on the back of the hand's skin texture) each bump has it's own specular almost like a bazillion little spheres stacked together, distributed across the surface of the hand. So the direction of the specular on each bump would confirm to reflection or specular rule but because they are distributed each of them would give a little specular over a wide area but thus only the planes with texture (little balls) in range will display the mixed or combined effect of saturation change:A little bit of light color blended with the object color and illumination of that color resulting in saturation change. Depending on the texture of the surface then the effect would be limited only to certain planes that has it's texture in range and that might not be the fully lit plane, and this effect could diminish down the length of a flat or constant plane as the relationship to the light and eye changes. Therefore the saturation painted on a lit side of a ball and the brightness change would not share the same relationship of degree into the same direction.

[briggsy@ashtons]

Very glad if the site has been of help, everyone - just keep the questions/discussion/crits coming!

zaorr I not sure if I quite follow all of that but it sounds to me like you're on the right track. I'd expect to see an additive combination of the fuzzy specular reflection (the colour of the light source, and centred on the point where the specular on a polished sphere would be) and the diffuse reflection (the colour of the object, and reaching its maximum at the full light plane).

[George Abraham]

Thanx.

[Dile_]

very cool thread.. i had no clue that you could translate so much to the digital medium... very cool , thank you so much!