MAT594G Techniques, History & Aesthetics of the Computational Photographic Image

George Legrady | http://vislab.mat.ucsb.edu

Course materials are protected by US Copyright laws and by University policy. Contents of this course may not be reproduced, distributed, or displayed without my express prior written consent.

Elings Hall, lab 2611, CNSI Building 2nd floor - Tues-Thurs 1-2:50pm

Course Information
An interdisciplinary course that examines, thorugh case studies, the state of the photographic image, its history, the theoretical, conceptual, and philosophical underpinnings. The course bridges studio arts, engineering, and humanities. This course may be of interest to artists, humanities researchers or programmers as there are three directions to explore:
  1. The aesthetic creative applications, opportunities and constraints,

  2. The computational processes from image processing to machine-learning, and

  3. The social considerations – to what degree can we believe in the image given the considerable software-based processing and autonomous image detail generation we are seeing through machine-learning processes.

The end goal is to investigate the photograph’s transformation through weekly presentations of projects, methods and discussion leading up to the impact of machine-learning on the creative process resulting in computational generated artworks.

Course Workload

Attendance and participation at zoom lectures
Weekly contribution to course journal at Student Forum | Legal agreement
Final presentation pdf documentation of either a research paper OR project


Course Overview, Introductions



Apparatus Fundamentals
An overview of the optical-mechanical image capture machine

Photographic History Selection (1830-1990)

A range of explorations from documentary, pictorialism, composite assembly, photograms, formal composition studies, etc.



Raster/Pixel Digital Photographic Explorations (1960-1990)
The image as a 2D matrix data structure consisting of numerically assignable pixels

Image Processing Fundamentals
The manipulation of the image through mathematical filtering



Material, Machine-Generated Images
Examples of the artistic application of the analog/chemical based materiality of the photograph and electronic delivery systems such as screens

Data, Signal & Noise/Glitch
Data for artistic exploration, Information Theory’s noise and signal, randomness, Brownian motion



Volumetric Data Points, Photogrammetry
Motion and depth sensing laser-based devices, artistic applications of photogrammetry image stitching (Weidi Zhang presentation)

Computational photography
The transformed camera through computers embedded within it



Computational Aesthetics
An engineer’s approach to understanding and quantifying the rules of aesthetic processes

Generative Art
Rule-based artistic explorations



Vision Science & Perception
Human/Animation visual perception and how it determines what we see and the design of the machines by which we see and capture images

Machine-Learning, CNN, Deep Learning
An introduction to machine-learning, convolutional neural networks (Weihao Qiu presentation)



Aesthetic Explorations of ML, CNN, DL, GANS
An overview of some artistic applications of machine-learning

Fabian Offert Presentation
Guest speakers Fabian Offert lecture – with a Humanities, critical perspective on image classification and machine-learning applications, followed by an overview by Mert Toka of the Xavier Snelgrove’s texture neural synthesis demo software



Various Deep Fakes, Social Implications
Humanities perspectives, news perspectives, use in arts and entertainment and business, applications of Deep Fakes




Individual Meetings

Concept Presentation



Open Studio, individual meetings

Final Class Presentation (Reports due December 15, 2020)

Alex Ehrenzeller
Chad Ress
Katie Parker
Mert Toka
Weidi Zhang
Weihao Qiu
Yichen Li

Pet Adoption Powered by AI
Ten Thousand faces
Computationally Recognize the Subconscious
Deep Reflection
Image Representation with CNN
A visualization of how Google Maps constructs one’s subjectivity through automated decision making (ADM) in its map labels