Fred Truman
Design & Code Creative
3D Sensing and Visualization – Week 01: Milk Scanner
February 16, 2011
As the first assignment for my 3D Sensing and Visualization class at ITP, we were asked as groups of 3 or 4 to implement some kind of 3D scan – the more DIY, the better. We were given the assignment after the first class meeting (which was delayed a week due to snow) and I’m only getting around to creating a post now because there were logistics issues with my group during this time, and I actually didn’t have a chance to contribute much to the scan that they produced. Since it would feel a little ridiculous to create a post about something I had nothing to do with, I’ve gone back and implemented a milk scanner based on examples we had seen during the first class.
Needless to say that I have the distinct advantage of seeing and discussing many examples from my other classmates’ projects, and getting a far better idea of the different ways to do this. One that I liked the best used sand (instead of milk like the examples we saw) as the filler then they just took the shapes and extruded them in Illustrator. It actually turned out better than what I’m showing here, but I thought I would give it a shot.
So here’s the supplies I used:
- 1 bucket
- 2 gallons of milk (only one was needed, but I thought I’d be safe)
- 1 tape measure
- 1 sharpie marker
- 1 digital camera and tripod
- something to scan
I spent most of my time thinking about which object would make a good scan and then rummaging through my apartment trying to find something adequate. I ended up using the top and smallest portion of a fruit basket that I never use, and although I thought it would turn out neat it really could have been better. The reason being is that it’s hard to get a sense of the basket’s dimensions except for the subtle change in radius as it tapers in down towards it’s bottom and the most interesting portions all happened within the last 2% of the scan – which throw my measured increments off. Objects that I think would probably be better to scan: surfaces with some kind of undulating topography – as opposed to the fruit basket which is really just a flat surface with texture.
To begin I measured the height of the basket and then marked even increments on the inside of the bucket with enough room to accommodate the basket’s height. I then placed the basket in the bucket, and simply poured milk enough into the bucket to fill it up to the first marker. Then I took a photo. Then I poured milk enough to fill up to the second marker. Then I took another photo. Then I repeated this process until the basket was completely submerged in milk.
What I had after was a bunch of images that have high contrast between the object and the milk – making it easy to extract the object information and extrude the pieces in Illustrator. Then put the pieces in proximity to get the general feel for what the shape looks like in three dimensions. I don’t really think it’s very accurate at all, and although I think the technique could be used for greater accuracy, this execution was shotty at best. :-)
I think by doing this exercise and also by viewing some of the other DIY techniques my classmates did, I have only a slightly better understanding of 3D scanning but certainly a better reverence for technology that makes this process easier and produces far* better results. I would in theory like to try to get this scan into a modeled environment to look at it from different angles, but that’s for another day I suppose.
