Transcript Color
Computer Vision for Visual Effects CVFX 2017 This lecture is based on the course materials of the Computational Photography courses given at MIT (Prof. Durand & Prof. Freeman) and CMU (Prof. Efros) Computer Vision for Visual Effects CVFX 2017 Color 3 Color [Wikipedia] 4 Spectrophotometer [Wikipedia] 5 Electromagnetic Spectrum Human Luminance Sensitivity Function [Wikipedia] 6 The Physics of Light % Photons Reflected Some examples of the reflectance spectra of surfaces Red 400 Yellow 700 400 Blue 700 400 Wavelength (nm) Purple 700 400 700 © Stephen E. Palmer, 2002 Question › If the reflected lights from two objects have different distributions in spectrum, then, is it possible that the colors of the two objects look the same to a human? 8 Different colors normally have different spectral reflectance, but different spectral reflectance may cause the same perceived color. the same perceived color white petal white flower Forsyth, 2002 9 Question › If the reflected lights from two objects have different distributions in spectrum, then, is it possible that the colors of the two objects look the same to a human? › What are the disadvantages? › Advantages? › How to exploit it? 10 Reproducing Colors › Projectors and LCDs › Does a projector need to emit lights of different wave lengths? 11 Additive and Subtractive Color Mixing 12 yellow Additive Color Mixing When colors are combined by adding the color spectra, e.g., three-color projectors. 500 600 700 nm 400 500 600 700 nm 400 500 600 700 nm white blue 400 13 When colors are combined by multiplying the color spectra, e.g., pigments. 500 600 700 nm 400 500 600 700 nm (blue pigment) 400 green cyan yellow Subtractive Color Mixing [Wikipedia] 400 500 600 700 nm 14 Blue Pigment [Wikipedia] 15 Color Science › How to make the projector reproduce the color we saw? 16 Human Visual Perception Cone cells function in relatively bright light. There are three types of cones responding to different wavelengths of light. light cone cell rod cell Rod cells can function in less intense light than cone cells. Rod cells are more lightsensitive, thus responsible for night vision. [Wikipedia] 17 Human Perception [Wikipedia] 18 Human Perception [Wikipedia] 19 20 Color Matching 21 Trichromaticity › Empirical finding › We can use a combination of lights of three wavelengths to create almost all the colors we can perceive (the coefficients may be negative). › For most people, the combination coefficients are the same. 22 Color Matching Experiment 1 23 Color Matching Experiment 1 p1 p2 p3 24 Color Matching Experiment 1 p1 p2 p3 25 Color Matching Experiment 2 26 Color Matching Experiment 2 p1 p2 p3 27 Color Matching Experiment 2 p1 p2 p3 28 Color Matching Experiment 2 A negative amount of p1 is needed to make the match: we add it to the test color’s side. p1 p2 p3 p1 p2 p3 p1 p2 p3 29 CIE 1931 RGB Color Matching Functions 700 .0nm 546.1 nm 435.8 nm [Wikipedia] 30 The CIE 1931 xy Chromaticity Diagram with CIE RGB [Wikipedia] 31 RGB › 32-bit mode › Only 24 bits are used › 8 bits each channel B R G 32 Early Color Photography › Sergey Prokudin-Gorsky › Optical color projections (1905) › http://en.wikipedia.org/wiki/Prokudin-Gorskii › Also check out Prof. Lewin's demonstration of color projections on YouTube http://www.youtube.com/watch?v=FJVvtOy-ukE&feature=related 33 Color Sensing in Digital Camera (RGB) -- Bayer Filter › Estimate RGB at ‘G’ cells from neighboring cells [Wikipedia] 34 Color Sensing in Digital Camera (RGB) › 3-chip vs. 1-chip: quality vs. cost › Why more green? › Why three colors? http://en.wikipedia.org/wiki/Bayer_filter 35 36 Yung-Yu Chuang 37 38