Rendering skin
Posted: Thu Nov 23, 2006 6:32 pm
Hello all!
With all the expertise on light propagation on this forum I thought I would pose a question to y'all.:
When measuring the ability of a lightsource to render colour a CRI value is used. This value is established by taking 8 colour samples and looking how well the colour is represented in comparison with a reference lightsource with a perfect colour rendering ability -typically an incandescent source. The problem is that the lightsource can be poor in rendering 4 colours and really good in four. This will achieve the same result as a lightsource which is halfway decent in reproducing all 8 colours.
Posing an opinion:
In my view the most important ability for a lightsource is to reproduce the human skin. We are social beings where the face of another person is the most important visual element in an enviroment. So the problem is that perhaps some of these lightsources are really poor at rendering human skin but get the same points for color rendering as one that is good at this task.
Also it is important for many reasons for humans to make assessments about the health of the person we are interacting with, and with poor colour rendering people tend to look sick - I suppose there is some similarities in the way light is propagated through sick tissue as with a lightsource with poor CRI.
Also often the impression of human skin is that it is a solid surface when in fact (at least in healthy skin) a lot of the colour comes from tissue below the surface.
My impression of some fluorescent tubes is that they change the relationship between the direct reflection of the surface and the subsurface scattering that occurs in comparison with an incandescent lightsource.
Can anyone point me toward research in this field. Does anyone have opinions on this subject? I would also be interested in how this is handled in CG.
Peder
With all the expertise on light propagation on this forum I thought I would pose a question to y'all.:
When measuring the ability of a lightsource to render colour a CRI value is used. This value is established by taking 8 colour samples and looking how well the colour is represented in comparison with a reference lightsource with a perfect colour rendering ability -typically an incandescent source. The problem is that the lightsource can be poor in rendering 4 colours and really good in four. This will achieve the same result as a lightsource which is halfway decent in reproducing all 8 colours.
Posing an opinion:
In my view the most important ability for a lightsource is to reproduce the human skin. We are social beings where the face of another person is the most important visual element in an enviroment. So the problem is that perhaps some of these lightsources are really poor at rendering human skin but get the same points for color rendering as one that is good at this task.
Also it is important for many reasons for humans to make assessments about the health of the person we are interacting with, and with poor colour rendering people tend to look sick - I suppose there is some similarities in the way light is propagated through sick tissue as with a lightsource with poor CRI.
Also often the impression of human skin is that it is a solid surface when in fact (at least in healthy skin) a lot of the colour comes from tissue below the surface.
My impression of some fluorescent tubes is that they change the relationship between the direct reflection of the surface and the subsurface scattering that occurs in comparison with an incandescent lightsource.
Can anyone point me toward research in this field. Does anyone have opinions on this subject? I would also be interested in how this is handled in CG.
Peder
