45.32 The Art and Science of Depiction⧉
Before this book, Frédo Durand and Julie Dorsey taught The Art and Science of Depiction at MIT, a course on how pictures work: how a painter, a photographer, or a renderer turns a three-dimensional world into a flat surface that someone else will read back as a scene. Its lectures on perception, contrast, tone, and perspective are where much of this book's material on those subjects began. The whole course is archived here, syllabus and slides, hosted with the book rather than left on an aging server. The original site is mirrored in full at /appendix-slides/depiction/.
An overview talk and its accompanying survey paper give a shorter route into the argument.
45.32.1 Introduction⧉
Presentation of the class and its motivations, and a brainstorm of the issues a theory of depiction has to face.
45.32.2 There is no passive recording⧉
Photography and cinema are hardly passive recordings. Make-up, lighting, film processing, and editing all shape the image, and realism is not always what a picture is after.
45.32.3 What is a picture?⧉
The chain that runs from reality to a picture, and the double nature of an image: at once a flat 2D object and a depiction of a 3D world.
45.32.4 Introduction to human vision⧉
Low-level human vision: the photoreceptors, the visual pathways, and the specialized analyzers the brain runs on the retinal image.
45.32.5 Vision solves problems⧉
Vision as tuned to extract information from the environment: invariants and constancy (an object read as the same across lighting and viewpoint), and vision as information processing.
45.32.6 Picture organization and Gestalt⧉
How the eye groups a picture into wholes: the Gestalt principles, and what they imply for composition.
45.32.7 Gaze movements and focal points⧉
Eye movements, saccades, and the focal points a picture is built around, with an introduction to color vision.
45.32.8 Color⧉
The physics of color (wavelength, spectrum, reflectance) and human color vision: color spaces, color appearance, and the cultural side of color.
45.32.9 Limitations and compensations I: the picture is flat⧉
A picture cannot give real depth, so artists have developed cues and tricks to compensate for its flatness, or to exploit it.
45.32.10 Limitations and compensations II: viewpoint and frame⧉
The viewpoint is unique and the frame is finite. How pictures work with, and around, those two constraints, with Byzantine icons as a case study.
45.32.11 The image is static; contrast and gamut are limited⧉
Three more limits of the medium, a still image, a bounded contrast range, and a bounded color gamut, and how depiction lives within them.
45.32.12 Contrast⧉
Managing contrast in a picture: what the medium can hold, and how tone is spent to convey a scene.
45.32.13 Representation systems⧉
A classification of the technical problems of depiction into three systems: the drawing system, the denotation system, and the tone system.
45.32.14 Perspective⧉
Linear perspective and its distortions, and the alternatives (cubist, Byzantine divergent, topological). A flat picture cannot preserve every property of a 3D scene at once; different projection systems make different choices.
45.32.15 Drawing and line drawing⧉
Drawing systems beyond linear perspective, and the theory of line drawing: which lines a picture shows (silhouettes, discontinuities) and why.
45.32.16 Denotation systems⧉
The primitives a picture is made of (regions, lines, points) and the marks that stand for them (brush, stroke).
45.32.17 Tone system⧉
How the tones and colors of a picture are set. Fidelity to reality is one option, but composition, ambiance, and symbolism drive the choice as much as accuracy. This lecture also covers the optics of the real scene: light, shadow, inter-reflection, and atmospheric perspective.
45.32.18 Non-photorealistic rendering, and style⧉
Non-photorealistic rendering, and a closing discussion of style: using the course's vocabulary to describe how a picture is made, its style rather than its subject.