It's the real deal.
Good ref though.
Thanks, zaorr. Good on you for taking on the challenge of a dappled light source! The main problem for me is that the direction of the light source on the ball doesn't seem to agree with the direction in the scene (assuming that those shadows are cast by more or less vertical trees). Also, the highlight looks very bright for the context - I would expect a fuzzy highlight to get dimmer when it spreads to that extent (and it must be spreading because the sharp-edged cast shadow implies a near point-source light such as the sun). It gives the ball a somewhat unearthly look, which of could be very useful when you want it.
I took a real picture and tried to fool you but you spot the over exposure really well. You just reminded me how bad a camera can be compared to real vision. I agree, it looks a bit out, even for a real picture. Didn't notice the direction problem at first but I see now how the overexposed blob makes the direction kinda off looking.
Last edited by George Abraham; February 9th, 2009 at 02:05 PM.
Good one, zaorr! I take it those shadows must actually be cast by something a long way from vertical (fronds?), which would explain the direction thing.
OK, now I get it, I don't understand the shadows either looking at them now. The cast shadows are from a palm tree with low hanging branches. The other shadows makes it a little confusing. I will have to check the scene agian to figure out what that's all about.
Something new everyday, so if this was a painting the shadows would have been to complicated to read. But with knowledge of it being a photo the mind just accepts it as fact without question.
I have read through the tutorial several times but I simply don't understand the rule of uniform saturation. From my experience with digi painting and PS color pickering on photos, low brightness almost always goes with high saturation on skin tones and many other things. How is it possible otherwise? Just pick any photo and it's always like this. Oops!
Great tutorial anyway, shockingly amazing, changed almost everything I know about color. It makes color so difficult.
Well, for a start the rule only applies to the underlying diffuse reflection; in various parts of the lit areas this colour will be modified by specular reflections of the light source (the highlight) and of the environment; for strongly coloured objects in most natural lighting situations the effect of these additions will mostly be to locally desaturate the colour. (Another complicating factor is multiple reflection, which can boost the saturation locally both in the shadows and in the lights).
Secondly, there are a couple of good reasons why just picking colours from photos can give the wrong impression of what is actually happening:
1. Colour noise tends to be greater in the darker areas of an image - try sampling from a larger set of pixels or blurring the image to get a more realistic measure of shadow saturation.
2. Any overexposure of the lights will result in artificially low saturation. Once one of the R,G or B components of a colour reaches 255 there is nowhere higher for it to go, and further exposure just adds more of the other two components, resulting in desaturation. Even in an apparently well exposed photograph some individual colours can easily reach this limit and so be "clipped" and artificially desaturated. This is true of any colours in the photo with brightness (B) = 100.
A uniform saturation series is just as series of colours in which the R/G/B ratio stays the same and only the brightness changes. It's hard to argue with the idea that this is basically going to work as a series of image colours representing a surface of one colour reflecting different amounts of light.
Anyway, I'm really glad you like the site. As far as making colour difficult goes, whenever a student tells me I'm doing this I take out my copy of Colour Science by Wyszecki and Styles to show them what some real writing on colour looks like!
Thanks David. You're really a expert on colors.
So do you recommend studying colors from photos? What should I keep in mind except over-exposure.
The information presented in this thread is an amazing resource for the technical applications of color theory. However, I'm currently TAing an introductory design class at an architecture school, and am curious if anyone has any good resources to share for other applications of color in design. Not how to make color, or what it is, or something so banal as "what this color symbolizes," but an in-depth resource on how artists and designers can use color to enhance their creations.
A webcomic of Apocalyptic proportions.
Thank you Briggsy for this wonderful thread. Color is always an interesting topic among artist who care about why they see "what they see". I'm on my 3rd book on Color Theory and I've concluded that there is no "right" answer, but there is good advice out there.
From what I understand, you are a proponent of the uniform saturation principle. Even though I looked at the example you provided, the Loomis recommendation sits much better with me. It really just looks more natural to me. In the "Spheres from Peer Project shaded on principle of highest chroma in half light" the lower 6 on the right look the most real to me (looks like the same person did those, looks like Easter m+m's). The "uniform saturation" ones below look...Photoshoppy to me. ("digital" is probably the appropriate term). I looked at the red ball you created in Photoshop in Idiot Apathy's thread and painted over it with some minor changes (I hope you don't mind). On your ball example it also seems like the change from half light to full light was only a change in "Brightness" and not value, so my full light does look more pink. I also added the background blue as the adjusted HUE of the shadow side of the ball. In my opinion that would be the color of the light that would be illuminating the shadow side. So would you mine explaining your thoughts on why this is incorrect?
Hi Jason, really great to see you on this thread. I'm in the process of switching to a new ISP contract at home, so I'll give you a proper answer as soon as I'm connected again.
Dementist, someone else might be better at answering that one - most of the books I've seen along those lines seem much of a muchness to me. The New Munsell Student Color Set has a concise introduction to the basic concepts, but if you want to look at the subject in more depth I'd be looking at some writing on colour in art history such as John Gage's books (Colour and Culture, etc).
The way I understand it now is that we have been studying a highly polished cue ball.
In theory the object will have a constant saturation as it moves into becoming brighter, I like fully lit better as it avoids confusion and implies working with the diffuse effect alone. In Photoshop this will be a movement parallel with the line between black and white in the Photoshop color screen (Bottom right black and top left white).
The movement will be kind of parallel and not straight towards white implying constant saturation. Some colors will hit the roof or top edge or the color square screen so they are forced to move directly towards white getting more saturated, the same with shadow, they will hit the right edge and will have to move down towards black directly. Are these examples existent in reality, is there some exceptions or are these colors too bright or not possible at such a purity for example?
I have named or identified a movement in our examples directly towards white as moving directly towards the color of the light source without changing the energy/diffuse of the surface, thus this movement can only be specular effect. So you cant do a movement like that on the white side without adding the effect as specular. I don’t know about shadow though, I guess shadow should only add a constant drop in value but then to look at the new darker environment as a lit area and shade that accordingly, but what if the drop has hit the side or the color screen. Should that be possible?
The red ball above displays a specular dot and specular effect arround it(Saturation change) making it look pinkish. But if you look closely that could be seen as a faint specular light impression of the room’s roof area around the light source or the sky. Was this intentional or accidental specular effect?
It’s just that I when I heard specular in the beginning I was thinking bright dot and perhaps some have stuck to the same idea. I think studying a shiny ball and a textured ball would have cleared certain things up faster. Anything not polished is textured, skin looks smooth but it's textured and the more common surface type you will encounter.
Regarding shadows I have been a moron, I keep trying to render around an object, I think I picked that up from loomis, not that it's what he taught, I just assumed that movement around the ball even for the shadow.
Something I only recently started to grasp and perhaps now solidify is that there is no shadow only change in lighting environments.
So to correct my own shadow behavior on the color picker, The shadow will have a once off drop(Indicating the absence of the main light). The drop on the color picker would be a move directly towards black in a straight line as the point of zero light will be the point of zero diffuse(Assuming the ambient is also white)
This drop should then be treated as a new environment where the ambient or room bounce light source should render the shading and appearance of that part of the ball.
I think a few exercises is in order, Shading a ball in an environment where the rays of the main light source is blocked from it, and once I have convincingly rendered the ball, remove the block and then render the lit side according to the new lighting environment. I should also try to get the idea of the terminater and drop shadow hardness as being an indicator of the light strength. I noticed this effect thankfully on a day standing outside for a smoke, where light clouds would move infront of the sun, Then I noticed how edges and terminators harden and soften.
Hi Jason, sorry it's taken a while to get back to you.
The very first step in understanding the appearance of objects (which Loomis does not take) is to make the distinction between specular and diffuse reflection. The uniform saturation principle applies to the diffuse reflection, but it's important to remember that the latter is almost always modified by specular reflections of the light source(s) and of the environment. These specular reflections are not changed in colour by the object, and so they tend to desaturate the diffuse reflection coming from coloured objects in the areas where they are present. For example, if a small main light source is surrounded by a bright diffuse cloud or surface, then the specular reflection of that cloud or surface will desaturate the colour of the sphere over a relatively large area. This in fact is the way your version of the red sphere reads to me, and apparently also to zaorr.
I'm afraid I can't think of any reason why the six spheres you mention should look right to you - to me they are all great examples of what goes wrong when people follow Loomis' advice! Some of the other spheres (including my 2006 sphere) may look "photoshopped" because they have been painted using a very high saturation series, which gives them an excessive chroma for a real sphere.
On my 2006 sphere I was thinking of the blue as a background rather than a completely encircling environment, and I envisioned that otherwise the shadow area was lit by white light. The hue of the light from the background is very close to being the exact complement of the local colour of the sphere, and so very little of it would be reflected in a diffuse manner from the sphere (actually none from my perfectly red sphere), and thus it would have no real effect on the (diffuse-reflection) hue of the shadow area.
Jason, it would be really great to see you what you could do with the exercise I described on the preceding page here, of painting a sphere into the specific lighting situation judged from a photograph. Personally I think you would soon discover for yourself the limitations of the Loomis advice for realistically representing appearances. Of course if you still prefer the Loomis "look" for aesthetic reasons, that's a different matter.
Zaorr, I'm not sure I follow everything you say in your first post, but the attachment shows what I think is happening in terms of the colour picker in Photoshop.
On the subject of shadow hardness, the tonal contrast depends on the relative strength of the main and secondary light sources, while the softness of the terminator and penumbra vary with the size of the main light source (small light source > sharper edges).
Aaron, sure I think you can study all sorts of questions about colour using photos - after all, a digital camera is really a sort of cheap and portable colourimeter that is intended to have a response closely comparable to human vision. Of course there are limitations but it's hard to comment on what factors to keep in mind unless you tell me what specific kind of question you want to investigate. But obviously the more you learn about how digital cameras work (as well as about light and colour), the more confidently you can use them.
Last edited by briggsy@ashtons; April 13th, 2009 at 08:33 AM.
Thank you Mr Briggsy for the gem.
I just had to see it on "the squre".
Hi Briggsy and Zaorr,
I haven't ditched the thread I just wanted to reply well. But there were some things that did not add up to me when I think about some of the information. And this just shows that no matter if you think you got it right, listen to what others have to say...cause I've been learning quite a bit here. What really made me think is why did you draw the area in full light just brighter and not lighter in value? I then realized that "maybe" it is backwards...that maybe loomis is considering diffused light and yours is specular light. The reason I say this is that "matte" objects are considered as having a diffused quality and the areas in full light, as the planes reach perpendicular to the light source, get lighter in value...Generally. When rendering matte objects...Form change = Value change. Well what gave me that "aha" moment is that I did a search of "red balls" on the internet and I noticed something. Almost all of the balls that I found, the "full light" areas to the "Half lit" areas were close to the same value...they looked pretty flat. And they were all specular...very shiny... and it makes sense that the half lit and full lit areas would be of similar value because the environment is being reflected to the viewer. Zaorr said this about my red ball and I agree that he is correct. So the values of the sky or room make up a somewhat "flat" plane and thus making the reflected half lit and full lit areas appear flat as well. Here are some photographic examples of what I am talking about. I also believe the the "lines" of the color picker for a "diffused" object would be curved towards the top and then head to white...given the earlier scenerio...Now that I look at the top "diffused" ball I think that the irregular shape of the half lit area is actually reflections of surrounding trees. I'll try the exercise on the earlier page and see what I come up with. Thanks guys!
Jason. you're quite correct that the red colour does not change much in brightness within the full light area, but this is just what we expect of a diffuse reflection. If a plane facing the light source catches x amount of light energy, the same plane turned say 45 degrees from the light source will catch cosine 45 degrees or 0.7071 times x. You may also know that our perception of brightness is related in a nonlinear way to light energy, so that a surface that looks middle grey reflects only about 18% of the light reflected by a white surface. By the same relationship, our surface reflecting 0.7071 of the amount of light energy will look about seven-eighths as bright as a fully lit plane. So in turning halfway (45 degrees) from facing the light source, the perceived brightness of the light hitting and being reflected from the surface will drop by only one eighth. For a red or mid grey object this amounts to a change of only a little over half a point on the ten-point Munsell scale.
So - form change creates value change, but only very slowly until we reach the half lights.
On top of that, because our visual system pays less attention to gradual changes than abrupt changes, we tend not to notice even the small amount of change in lightness that does occur in the full light. (? which may be why you thought I had made the full light brighter but not lighter ??)
I would describe what you've illustrated as the difference between a small/medium-sized light source and a very large light source, but it's true that the latter creates a saturation pattern like that described by Loomis in the full light and half light (although as you can see this tends to come with an indistinct terminator and shadow shape).
Last edited by briggsy@ashtons; April 15th, 2009 at 09:49 PM.
This thread has taught me not to look at specular as a surface condition but to look at the surface as it it's a mirror when dealing with specular regardless how dull it is.
Those are nice balls Jason. Try looking at each as if they are mirrors and feel the light source and blue skies reflected in them. This will help you seperate surface diffuse from the blending reflection.
I did another rendering study the other night and found myself placing the specular and some of the roof lit around the light source, I do believe now that this is the correct way to render speculars, for your mind to shift looking at form as if it's a mirror and then rendering the "source scene" as it may appear on the surface. No more turning a piece of the object into a bright spot mechanically like loomis would have.
I baught some blue scrap books a while back and a nice white color pencil not knowing exactly what I should be doing with it. I guess now is a good time to take those out of the drawer.
Last edited by George Abraham; April 16th, 2009 at 12:07 PM.
Thank you for this valuable reference. How generous of you to make this available to us.
Briggsy (and others)- you might be interested in a Photoshop script I wrote that generates a digital approximation of HVC strings in the swatch palette:
Great work Tim! It actually generates hue-value-saturation strings, but they are at least as useful as HVC strings, in their own way, for physical paint as well as digital.
Quite apart from its practical use, the swatches should have a great educational use in helping people to learn to think in terms of an absolute tonal scale of Munsell values.
I look forward to giving it a try!
I'm glad you like them David! I'd love to hear any feedback you might have on the script.
I'm contemplating a version which would let you specify saturations on a per-hue basis (rather than applying the same saturations to every hue). It would require the user to learn a special syntax to enter the desired values, but it could be a little more useful & flexible for those willing to brave it (I'd probably choose something close to the HVC slash notation). I'm also contemplating a really fancy version that would only work in CS4, but would allow you to "mix" between swatches on the fly- so that if you wanted to knock the saturation of the current color down a bit you could click on the corresponding gray a couple times, which influence the current color but not replace it. This would be pretty complicated, but a great simulation of painting with strings in oil. But that won't happen for a while- I've been tinkering with the tools too much lately and not painting enough with them!
It looks like it's been down for most of the weekend our time. I've just sent off an email to ibiblio, so hopefully I'll know something soon. Glad to know it's missed!
Mr. Briggs sirrrrrr, I have opened pandora's box again and have become rather confused. I've been going over some of the old Peer Project exercises recently for CA's mentoring and got stumped by one of my own questions! Well, it's not that I don't know the answer I wanted at the time; it's that I think I was wrong!
It's an hypothetical exercise with an object of a pure hue in an environment with a single pure and direct lightsource.
The question is as follows; The lightsource is pure red and pure yellow. Our object is green. What do you see?
In my old train of thought I would have said you would see as the end result a yellow sphere. But now I can't help but shake the feeling that you would see a green sphere though perhaps a bit dark...
Can you help untwist my brain as you always have?
The exercise for reference if necessary.
I think the crucial question is what we mean by a green object "of a pure hue". Probably not an object that reflected only monochromatic green light - such an object would appear very dark or black under any other light source. There is a theoretical concept of "optimal" colours that reflect all wavelengths in a given band of wavelengths, and none outside that band. Optimal green surfaces range in theory from darker more saturated greens (reflecting a relatively narrow band of wavelengths) to brighter but less saturated greens (reflecting a broader band). So the answer really depends on which of these it is - the former would reflect neither red nor yellow monochromatic light, but the latter might well reflect yellow. How does that sound?
This is an excellent resource. I am glad that something like this can still be offered for free in this day and age, and it seems that you are very devoted to your subject. Even though it's difficult for a relative beginner like me to take it all in in one setting I still feel I learned something while reading through it the first time, and I'll be sure to keep coming back to it as I learn more.
The chapter on colour constancy I find especially interesting, it's quite amazing how your perception of things needs to change in order to faithfully reproduce reality on canvas, the problems faced when trying to copy what is perceived rather than what is actually there is something I'm sure we've all had to struggle with.
The only thing I could say I would've liked to see that isn't on the site are some analyzes of real-world conditions and how the principles learned can be practically applied in painting, as well as how they have been utilized by others before. This may be something that I'm alone in finding useful, but I think it would have been valuable to see both photographs and master painter's works deconstructed, with the colour dimensions individually analyzed and interpreted, and then (partly) reassembled in paint, revealing not only the information itself but also how it can be used. I know it's too much to ask of a website like this, but it could be an idea if you ever make the site into a book or a DVD.
Anyway, I just wanted to extend a big thanks for the work you've put in, the site is incredible.
Just saying that I'm finally trying to intergrate with all this computer stuff.. he he you'll see
Hope you're well
Your very first CA post - I'll treasure it always!
But don't forget to link to your blog in your sig!