Archive for the ‘Openframeworks’ Category

Watercolor and charcoal drawing in realtime 3D

Saturday, June 28th, 2014

Experiments on realtime painting/drawing looking rendering. Examples are all done with the same set of shaders with slightly different parameters.
Thom Yorke one was a bit slow because the data is provided as a series of csv files and it’s loading one file per frame. I could make it faster when rendering the video but kept it slow since the noise around his hair was too jarring in 30fps.

point cloud to drawing from kynd on Vimeo.

point cloud Thom Yorke from back in 2008
http://creativecommons.org/weblog/entry/8476

charcoal drawing study from kynd on Vimeo.

3D data: Walt Disney Head by David OReilly
http://blog.davidoreilly.com/post/18087489343/disneyhead

Auto portraits

Monday, March 31st, 2014

Automatically generated portraits based on the Flickr photos shared under Creative Commons Attribution 2.0 Generic License (credit).
Resulted portraits are necessarily looking alike the models, even some might be hardly identifiable as faces or some appear too photographic to me, but I just let the codes do the job and didn’t manually filter anything out.
The photos are randomly download by a PHP script and scanned with ofxFaceTracker to locate the faces. The challenge is to reduce the number of strokes while making them look more deliberately placed. A good artists knows which lines are more important than others and is able to capture a shape with fewer touches, which is not very easy for computers.

portraits from kynd on Vimeo.

01

02

03

04

05

06

07

08

09

10

stroke tracking from kynd on Vimeo.



bluebird and diamonds

Sunday, December 1st, 2013

bluebird

diamond

diamond

diamond from kynd on Vimeo.

Original (2560 x 1440)

Abstrakt

Sunday, September 1st, 2013

abstrakt

abstrakt

abstrakt

abstrakt (oil paint study) from kynd on Vimeo.

Should simulated painting look good? Yes, at least to some extent it should be somehow interesting to look at. I’m spending some time to make them look OK but since they are experimental studies, I’d rather prefer to move on quickly to the next trial than sticking to one project too long.

Simulated painting could possibly be very beautiful visually. Yet I wouldn’t feel like spending a lot to buy the resulting pieces even if some genius invented an algorithm that could perfectly reproduce Van Gogh or Picasso.

The real fun is in the process of studying how people, including myself, paint, see and react to or evaluate paintings through trying to reproduce it almost from the scratch. the medium and the method, that are usually thought better to be invisible so that the viewer can focus on the art, are actually the subjects here.
It is more like playing with a very complex puzzle where tweaking a little part can change the composition of all other pieces. I believe it is, to me, impossible to make a human painter completely into code, nor probably it is not what exactly I mean to do. There always needs to be abstraction occur in some level, and the game is to find several rather simple rules that can produce seemingly complex and realistic phenomenon.
This is analogous to Boids(example) being able to reproduce the behavior of flocking birds or fishes from only a few rules regardless of whether the rules are what real birds are following.

And a quick comment to add is that this is not such a serious project but rather is a fun study of openGL and openframeworks trying out what I can do with them. It just goes random.

Related Posts with Thumbnails