[bre] passed along this portrait drawing robot from the same people that brought you the chief cook robot. Like the previous project, the portrait drawing bot emphasizes man machine interaction. The subject sits down and is asked for the quill. The robot captures the subject’s image and runs it through several processing steps. One set is to determine the contours and another is to determine the fill. The robot then draws the the contours and fills returning to the ink pot each time it reaches a trajectory threshold. Quill drawing can be quite hard for novice illustrators, but the robot’s smooth trajectory calculations prevent drips and stains.
6 thoughts on “Portrait Robot”
Leave a Reply
Please be kind and respectful to help make the comments section excellent. (Comment Policy)
This is quite a feat – just the image processing is pretty hard, getting a (probably grainy) image of the person, seperating it from the background, detecting edges etc. is a real challange.
The voice synth could do with a bit of work – even Microsoft Sam is better than that!
Where is the finished image? This seems like it would be pretty easy to do. The processing looks like the same to make a pumpkin image from a normal image. But it’s awesome that it uses a quill and ink pot, wow.
I worked in an arcade for several years and we had a “photobooth” like machine that would plot out portraits for people. you’d basically pose and take your photo like a regular booth but you could watch a cartoon hand pretend to draw your picture on the screen and then it would print out. The quality was quite good and you could do color if you wanted too.
If you were to take the algorithms from that, have the robot decide when to take the photo and plot the image in real life instead of using an ink-jet you’d have a much better version of this robot.
I’m not sure but I think the machine we had basically took the photo and applied a photoshop like filter to it and then pretended to draw the resulting image. I think it could also determine facial features because it would always start with the eyes then the lips and nose then the outline and hair then go back and touch up the facial features and then do the background.
haha almost as good as the NS-5 on iRobot :)
I’ll give that it is no small piece of work, tho (and perhaps I’m just old school) I was much more impressed with Maillardet’s Automaton (built around 1800)
http://www.fi.edu/learn/sci-tech/automaton/automaton.php?cts=instrumentation
In case anyone wants to check it out.
I guess this sort of technology is just getting more and more common place but if I was to stick a plotter on a photo booth I’d get essentially the same result. I don’t wan to sound too jaded I’m sure more work went into this than that and it is still interesting.
what would be cool is if the bot could see you doing the things it asks instead of you having to talk back to it because it seems like the voice recognition isn’t very clear; he had to speak pretty loud and clear. it also seemed a bit slow to respond but that’s to be expected. it’s cool nonetheless.
but i shouldn’t really say anything because i’m pretty technologically challenged; i can only so stuff with simple circuits with no circuit boards or programming or anything like that.