Posted by gever on December 26, 2008
Using only information in the image, is it possible to algorithmically choose angles for brush strokes? I started fooling around with this idea back in 1988, and have been resurrecting it off and on over the years – this version runs in Processing.
Get the source here.
|
Strange Things Around the Eyes
This has always been the problem with this approach – no attention to the anatomical details. I wonder if facial-recognition data could be used to scale the strokes around and on the eyes to preserve some detail?
|
|
More Facial Distortion
I like the treatment of the ceiling in this one, but the distortion of the facial features is disturbing again.
|
|
Sleeping Dog, Window
Swung the laptop towards the window and caught Kai sleeping. Stroke direction on Kai just happens to coincide with fur.
|
|
Un-natural Directional Choices
Vertical strokes on the hand? Feels like those should follow the vectors of the fingers…. Might try reducing image to a few colors, creating a polygonal representation, and then use a vector field (seeded with the lines from the polygons) to determine brush angles.
|
|
Self Portrait – Eyes Closed
Looking for good angular distinctions between areas of similar brightness but different hues.
|
|
Monochrome Brushing
Ignoring the color information to see the angular details more clearly.
|
(added 3.jan.09)
Interesting to note what kind of traffic you get from posting to the Exhibition section on processing.org:

I posted to processing.org on Christmas day (25.dec.08) and then get an almost perfect decay curve in blog traffic.
This entry was posted on December 26, 2008 at 2:08 am and is filed under Uncategorized.
You can follow any responses to this entry through the RSS 2.0 feed.
You can leave a response, or trackback from your own site.
Nicolas de Vallière (Geneva) said
Very nice !
May I suggest a look at the SUSAN ALGORITHM, fine for detecting gradients with noise rejection ?
Sincerely yours, N.