UTCS Colloquium: Ganesh Ramanarayanan/Cornell University: Visual Equivalence - A New Standard of Image Fidelity for Computer Graphics ACES 6.304 Friday September 19 2008 3:00 p.m.

Contact Name: 
Jenna Whitney
Date: 
Sep 19, 2008 3:00pm - 4:00pm

Type of Talk: Colloquium

Speaker/Affiliati

on: Ganesh Ramanarayanan/Cornell University

Date/Time: Friday Sep

tember 19 2008 3:00 p.m.

Location: aCES 6.304

Host: Keshav
Pingali

Talk Title: Visual Equivalence - A New Standard of
Ima

ge Fidelity for Computer Graphics

Talk Abstract:
Determining the
visual fidelity of an image is a
fundamental problem in computer graph

ics. When
is an image good enough; i.e. when does it convey
a conv

incing representation of a scene? Most graphics
algorithms either aim t

o compute a physically accurate
solution matching the real world or th

ey leave judgments
of fidelity entirely up to the end user. The former

is often
computationally intractable and the latter is ad-hoc since it cannot be generalized or predicted.

In this talk I will introd

uce our work charting a new course
between these two approaches. We pro

pose visual equivalence
a new standard of image fidelity that focuses

on what is visually
important to the observer: the appearance of the sc

ene consisting
of impressions of shapes materials and lighting. Unde

r visual
equivalence an image with noticeable pixel-by-pixel differen

ces
from a perfect reference can still be a high fidelity representatio

n
of the same scene provided it conveys the same impression of
app

earance. Our work is to our knowledge the first approach to
image fid

elity that permits judgments of this kind.

I will present an end-to-

end psychophysical and algorithmic
investigation of visual equivalence
and its impact on scene
modeling and rendering in computer graphics. F

or natural
illumination we measure the degree to which representations

of lighting can be manipulated without affecting object appearance. We demonstrate how the resulting metrics can motivate new
rendering a

lgorithms for scalable rendering and compression. For
complex aggregate
geometry we investigate how different
combinations of object shapes a

nd colors affect appearance
and derive thresholds that can be used to

reduce scene complexity.
This research takes some important first steps
into a large new
space of perceptually based graphics to address funda

mental
challenges in modeling and rendering complex scenes.