h a l f b a k e r y
Expensive, difficult, slightly dangerous, not particularly effective... I'm on a roll.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
Take the images on amihotornot as a
starting "population". Use the
amihotornot rating service as an
evaluation metric. Combine two
photos by using the "morphing"
algorithms that were all the rage a
few years ago (remember that Michael
Jackson video?). Submit the results
for evaluation. Repeat.
goal here is to combine existing
photos to synthesize new photographs
of fictional people who are as
attractive as possible.
Morphing algorithms require human
involvement and work best with
similar backgrounds, but these
problems can be overcome,
particularly with the aid of modern
machine vision work that
automatically detects faces and
facial features in images.
(I know, amihotornot is yesterday's
fad, but I just had the idea. I'm
not sure this is categorized correctly.)
Evolving faces from principal components
In support of crime witnesses, not Internet voyeurs. Definitely departs from morphing towards a more structural separation of a face into texture on a grid ("eigenshape"), created manually. The texture is morphed, the grid parameters change, and the resulting random images are much more convincing. [jutta, Mar 02 2001]
Searching for genetic similarities by the principle that certain events have predictable anticedents. [reensure, Mar 02 2001]
If They Mated
[rapid transit, Oct 05 2004]
||There's something really creepy
about the grids of
disembodied synthetic heads in
that paper. But, yes, that's
clearly a superior technique.
||From the title of this idea I envisaged a creepy eugenics organisation which would force amihotornot.com 'winners' to reproduce.
||ever seen the dave gorman collection? well see what the average dave gorman looks like and youll soon see wht a bad idea it is... that reminds me, i havnt been on amihotornot for a while... bye
||This was done in a Discover magazine a while back. They showed people pictures of women's faces, then added the features highest-rated together, created a run of faces with those, and so on. Eventually it came out to a short-haired blond blue-eyed girl with a very pointed chin.
||Steven Pinker mentions composite attractiveness, and in general, the factors that go into attractiveness in, *How the Mind Works* (pp 483-7). He cites a number of interesting papers.
||People might want to search the web for the 'Most Average Facial Appearance' effect - it's a tendency to prefer faces in which the eyes, nose, lips and other features are close to the average of a population.
(from 'Universal Principles of Design, Lidwell et al, 1-59253-007-9)