Priciples of Color Apperance

1. Cone Receptors respond to the quanta catches of their outer segments

     •  Two extreme special cases:

                  a. Beach Scene: Scene Content is mostly sand and sky

                       Maximum intraocular glare

                       Minimum dynamic range of retinal image

                       Reduces 5.6 log unit scene range to 1.5 log unit retinal range

                   b. Stars at Night: Scene Content is mostly night sky

                        Minimum intraocular glare

                        Maximum dynamic range of retinal image

                        Reduces 5.6 log unit scene range to 4.0 log unit retinal range

       • Introcular glare transforms light from scene into a differnt patten of light on the retina

          Optical Glare Spead funtion (GSF) controls retinal image [1]

2.  Color Appearance is neural process caused by spatial comparisons of cone responses

     •  Two extreme special cases:

                  a. First Step:

                       Molecular Physics of cone quanta catch - Colorimetry

                       Able to calculate if two stimuli will match     

                       Restricted to predicting matches of spots of light

                   b. Second Step:

                        Human neural spatial processing

                        Able to calculate the appearance of the entire scene

                        Predicts Color Appearnce of all visual stimuli using entire field of view data

        • Receptors report cone quanta catch; then, neurons use spatial comparisons

          to syntesize appearance. [2]

3.  Edges and gradients:   Color is neural process conrolled by spatial comparisons

      Photographs are moelcular physics records of sensor quanta catches

      Appearances are the sythesis of edge and gradient reports

     •  Two extremes of Human Response Functions:

                  a. Edges:

                       High-slope response

                       Edges cause large changes in appearance

                           from small changes in recepotor quanta catch

                   b. Gradients:

                        Low-slope response

                        Gradients cause cause small changes in appearance

                            from large changes in recepotor quanta catch

        • Edges and Gradient (not reflectance and illumination) are the builing blocks

          of appearance [3]

4.  Human Response function varies with the content of the scene

       •  Two extreme HDR targets with 5.6 log unit dynamic ranges

                  a. Max background: Luminance = cd/m2

                       Maximum intraocular glare

                       Minimum dynamic range of retinal image = 1.5 log unit

                       Appearance white (luminance = 100%) :  Appearance black (luminance = 3%)

                   b. Min background: Luminance = cd/m2

                        Minimum intraocular glare

                        Maximum dynamic range of retinal image = 4.0 log unit

                        Appearance white (luminance = 100%) :  Appearance black (luminance = 0.01%)

       • Eyes response to light varies with the content of the scene (light distribution in scene) [4]

      Variable response counteracts glare and increased apparent dynamic range

5.  If we can model human vision, we can calculate appearances (matching sensations)

      Then, we can write calculated sensation on film, now media

        •  Use best practice to capture gratest (glare-limited) scene range

          •  Scale input information to log luminance

        •  Use spatial algorithm that empasizes edges and minimizes gradients

          •  Transform output color space to match dispaly media's color space

       Calculated sensations make better HDR renditions [5]

6.  Unexpected answers to 9 Questions [6]

          •  Rods are color receptors

          •  Firelight is the ideal illumination for rod- Lcone color stimuli

        •  Rod-Lcone colors share the same color appearance space as Cone-Cone colors  

          •  CIE XYZ predicts matches not appearances

          •  Edges in illuminartion erodes color constancy

           Constancy works best with flat displays in uniform illumination

          •   L* Lightness's cube root comes from introcular glare  

          •  Dynamic range of receptors is 109,

          but on the beach the dynmic range on the retina is 30:1

          •  Calculated sensations make better photographs

              Enhance edges, minimize gradients, increase chroma

           •  Rods are color receptors