top bar
QuickTopic free message boards logo
Skip to Messages


Trainable imaging

02:19 PM ET (US)
I wonder if you would 'break' the filter if you tried to create an analogy filter from an original model and a Picasso abstract. Also it begs the question, when does an apparently painted image stop being art? And is the artistry in the technical proficiency of the filter generator's code, or in the end product? I suppose another interesting (far-fetched) ap would be to create this theoretical two-way filter on a given Van Gogh, using the model and the product, then take another of his products, back it thru the filter, take the output and run it thru a Rembrandt filter to have a work of Van Gogh as interpreted by Rembrandt ...
Cory Doctorow
12:42 PM ET (US)
I was thinking the same thing -- also wondering if you could used a blurred image as your original and a sharp as your transformed image, and get something that auto-sharpens its input.
Erik V. Olson
12:40 PM ET (US)
Of course, you need damn near unghodly processing power to pull the trick. Of course, damn near unghodly processing power will cost about $300 next year, and $30 the year after that.

This is really, really cool. I'm wondering if it can turn around -- set up several real still lifes matching Van Gogh and Monet paintings, run the software, and then get a Van Gogh or Monet filter....

Print | RSS Views: 977 (Unique: 505 ) / Subscribers: 1 | What's this?