Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

You can send an email to if you have questions about how to use this forum.

Oil Paint Color Temperature - Split Primaries

I couldn't find a conversation and/or reference to the split primary color temperatures subject and the available Geneva Paint primaries,  so I thought I'd throw this out there. 
Wondering how the limited palette / 3 Geneva primaries (Cad Yellow, Pyrrole Rubine & French Ultramarine) are related to with respect to the split primary story.  

Any insights and/or references are appreciated.



  • Split primaries seem a bit of a stretch here.  I suggest you watch the DMP videos. DMP is a primary color palette leaning system. 
  • SummerSummer -
    edited December 2016
    @Albert ; All I know from reading online is that they, not DMP, are still trying to agree on whether some of the colors are either warm or cool or not.  For instance is ultramarine a cool or warm color?  Is phthalo blue a cool or warm color?  Don't know if this topic will gather force here on the forum but there are online resources that delve into the questions you ask.  Summer 
  • edited January 7
    Split Primary idea arose from a misunderstanding a couple of centuries ago. It started with Newton's observations of mixing colours using prisms. From this people got the idea of a small number of 'primary' colours which could be used to create all other colours. When artists tried this, however, they found that there were colours that they could not mix (not hues, specifically, but they could not match the hue/value/chroma of all colours from their three 'primary' colours, for some hues, at the highest chroma).

    Why was this? Well, what they decided was that there really were primary colours, but the pigments they had to work with were not 'pure', they were 'contaminated' by some of the other primaries. So in trying to mix a high chroma violet from red and blue (or magenta and cyan come to that, the chroma that can be achieved is higher with these two but still has limits) they thought they failed because the red (or blue or both) contained a bit of yellow, which being complementary to violet decreased the chroma.

    So, they decided that as they don't have pure primaries in their pigments, they would just control which of the other primaries they were including. By having a red that was contaminated with blue (a 'cool' red) and a blue that is biased towards red (a 'warm' blue) then both would be contaminated by the other, and not by yellow. Similarly, they could have a yellow that was biased towards blue and one that was biased towards red, and use the 'warm' yellow with the 'warm' red to get better oranges, for example.

    So, we are talking about how to get the largest gamut, and in fact by using these 'split' primaries they could increase the gamut, but not by much, and not for the reason they thought. The reason they could not get the highest chroma in secondary and tertiary hues is not because of 'contamination' as such, but because of saturation costs. Think of a colour wheel (should really be 3 dimensions but we'll take a slice at a particular value, so you have all the hues with neutral in the middle for a particular value of neutral grey). Mark your red hue/value/chroma on the wheel and your blue likewise. Draw a line between the two. Basically, that line is the maximum chroma you can make at any hue angle by mixing those two colours. You can get a lower chroma at any particular hue angle along that line by adding some of the third primary. Put in the point of your third primary (yellow) and draw lines and you have a triangle. That triangle encloses all the colours that you can make with those three colours (not exactly, the lines aren't actually quite straight, particularly in the red/yellow portion, but this is a useful simplification).

    You can see that at the mid-point between any two primaries your line is closest to the centre of the colour wheel. In other words, this is where you can achieve the least chroma. The three colours themselves are the highest chroma you can achieve, because obviously you cannot make a blue be more blue by adding yellow or red. This fact, that the line connecting the points of any two colours you chose inevitably gets closer to the central neutral value in between those two points is called 'saturation cost', and this is the real reason why no three primaries can match all colours. In fact, no three 'primary' colours that can exist are true primaries (ie able to mix all colours), and any three true primaries cannot exist in reality (but can in mathematical colour space).

    To minimize saturation cost you choose two colours that are closer together in hue. This is actually what you are doing when you choose a 'split primary' palette, nothing to do with 'contaminants'. In your mind, move your red and your blue points on your colour wheel closer together and you see that the line between them doesn't dip towards neutral grey as much. (and if the two colours are 180deg from each other the line passes right through the centre, ie these are complementary colours). To get the largest gamut from 6 colours you should choose them equally spaced, ie Cyan, Blue, Magenta, Red/Orange (eg Cadmium Scarlet), Yellow, Green. This is a 'secondary palette'. If you add more colours equally spaced between the ones you already have and at highest chroma you can get, you push the gamut still further (a 12 colour tertiary palette). I believe you can mix all the colours in the Munsell book of colour (nearly 1600 colour chips) with about 19 carefully chosen oil colours. You can also see that to get the largest gamut from any set of colours those colours should be ones with high chroma. eg if you have yellow ochre, perylene maroon and indanthrone blue, you will not be able to mix as high chroma colours as with Cadmium Yellow Pale, Quinacridone Red, and Phthalo Blue G/S. As a further nail in the coffin of Split Primary theory, consider that purple is not a wavelength of light between the red wavelengths and the blue wavelengths, and yellow isn't either, they are necessarily mixtures. If you picked a wavelength exactly and narrowly where we think 'yellow' is it would be very dull, because our visual system does not work by having three sensors specific to certain wavelengths like a digital camera, it works by an opposition system. Yellow is seen when BOTH green and red is present but blue is not. So the idea of a 'pure' yellow pigment without any green or red present is just nonsense. Split Primary theory assumes that you could mix all colours from three primary colours if only you could get 'pure' pigments, but this idea is simply false.

    How does this relate to the Carder Five limited palette? The point is that although you could make a larger gamut with more colours you actually rarely need very high chroma colours for accurate realism, so instead the colours are selected to mix well and logically. eg Phthalo Blue G/S will mix higher gamut greens and purples than French Ultramarine, but Phthalo is such a strong tinting colour that it is hard to control and work with, whereas Ultramarine is not. For portraits, especially, the colours are actually pretty muted, and it is more useful to have some earth colours than mix the same thing with expensive cadmiums. You also need to be able to produce the full range of values, so you need a set of complementary colours in your chosen palette, in this case Burnt Umber + French Ultramarine. 

    Every possible palette has advantages and disadvantages. Eg, if you had a palette of 20 paints you could mix more colours, and many colours with just one or two of them, whereas with the Carder palette you might need to mix using four of your paints. But you also need a clear mixing strategy, a way to accurately zoom in on any colour you want, and that is what Mark based his palette on, plus much less waste as you only ever need to buy the same 5 tubes and not have some unused year after year.  You will also notice that Mark's palette has red and blue, not magenta and cyan, which are more 'true' primaries. The effect of this (apart from pigment behaviour) is to increase the gamut on the yellow/red/purple/blue side of the colour space, and reduce the gamut (maximum achievable chroma) on the green side. However, this matches very well with the sort of realism that Mark, and I think most of us, are looking for. In nature greens are actually not that saturated on the whole. Of course, if you have a flower or some fabric or something that you cannot achieve with these five (you would be able to match hue and value, but can't reach the chroma) then you would add something for that particular case, eg Phthalo Blue G/S, or Phthalo Green B/S perhaps, or Quinacridone Magenta, or Cadmium Orange, or whatever. But you would very rarely need them, so it is pointless keeping them on your palette all the time.

    Hope that helps.
  • SummerSummer -
    edited January 6
    Thank you @AlunapR for taking the time to create this interesting explanation about the color wheel, hue, value, chroma, bias (warm and cool concepts), and for putting them into perspective as they relate to painting with pigmented color, the Carder palette, and science.  Summer
  • Thank you @AlunapR :)

    I agree with what you said which corresponds very well to what I've read on

  •  I'm gonna get a PHD in art just from reading this forum. Thanks @AlunapR and thanks to all. this is really inspiring but I think I'm going to take Kingston's advice because Mark's course keeps things simple  and practical for me. As I get more experience I'll revisit the split primary question.
  • edited January 6
    @Richard_P, Handprint is an excellent resource, that guy really understands his stuff. Recommended. Also, since 'colour' is a psychological phenomenon, not a physical one, I could recommend for an in-depth discussion. The Munsell book of colour is extremely expensive, but the Munsell Student Color Set is a lot cheaper, and well worth the money for a good understanding of how to think about colour. But keep in mind that the main thing you are interested in is how to create in paint a colour that matches a particular point on your subject, and to place it on the correct place on the canvas. This is what Mark teaches, better than anyone else, I think.

    And in case anyone gets the idea from my earlier post that because you can get a larger gamut by choosing each of your colours to be the highest chroma possible (which is true) that you should do so, rather than some of the more muted colours that Mark chooses (like Burnt Umber and Rubine Red/Permanent Alizarin Crimson/Perylene Maroon and similar) remember that Marvin Mattelson originally used Frank J. Reilly's palette which had lots of Cadmiums, but found that his students struggled to control these strong colours. He then took William MacGregor Paxton's palette of Yellow Ochre, Indian Red, Terra Rosa, Raw Umber, Ivory Black and Flake White and found both he and his students had a much easier time getting to the colours they needed for portraiture. Like Mark, he kept a few stronger colours 'in reserve' (in Mattelson's case, Ultramarine Blue, Cadmium Yellow Light, Cadmium Scarlet, Alizarin Crimson and Viridian) for just those times when he needed more chroma.

    So there are real advantages in having a limited palette that offsets the theoretical disadvantage of not being able to achieve every colour possible with paints.
  • FlattyFlatty admin
    Simple is good for me :-) paint what I see. Mix and match with DMP work flow. 
  • AlunapR

    Welcome back to the Forum. Where ya bin?


  • edited January 6
    @dencal Been busy with many things, but not too much painting. I have been drawing, and some watercolour and gouache, but this year I plan to do lots of oils, starting with a portrait of my wife. So, I'll probably be hanging around here more now. For those not aware, I used to be active in the original Carder Method forums some years ago, and in fact started oil painting using that method. Whatever else I do I always come back to the principles of that method.
  • edited January 6
    Excellent points @AlunapR

    Also the lower chroma colours you listed (and other inorganic pigments) tend to be more opaque which suits Mark's single layer approach as well.
  • Wow!  Your explanations of these elusive color concepts are worth their weight in gold.  Finally I understand how and why Mark's system so successfully accommodates the many complexities of oil colors and color mixing. 
  • Thank you @AlunapR !
    and everyone else.
    Really appreciat the shared insights, dialogue and exploration.

Sign In or Register to comment.