Blistering Barnacles: 3d printing smackdown

So now I have TWO Barnacle lamps.

After reading my post on 3d printing a lamp on a Zcorp 450 out in ADSK land, Jesse Roitenberg from Stratasys was kind enough to print a version on a Dimension uPrint. So I wired it up with a Sylvania portable LED work light . . .

which is nice because it has a magnetic base that I can stick to stuff.

There is a lot to say in favor and against the uPrint and Zcorp 450 printing process. In the end, which process you choose really depends on what surface you are going for, what your facilities are, and how much time you have available.

Post Processing

Out of the machine, the piece appears much more durable. I didn't finish this one myself but Jesse said:
"Once the model is taken out of the printer you just put it into an agitation bath with heated water and a solution and the supports dissolve away. I believe this model was in the tank for 2-4 hours. Then you remove the model and the supports are gone. A quick rinse and I shipped it off . . ."


I'm still amazed that I managed to get this one finished without breaking it. After excavating the barnacle from the powder by delicately vacuuming around the edges, and blowing the remaining surface powder off the exterior, I started blowing out every individual cell. This quickly became boring, so I took it outside for a good blasting with the shop vac on reverse, which worked well (although my car looked like a mime). I don't have access to a larger dipping or spraying facility to cure the final product in what is essentially superglue (I also don't like the finish this gives, a bit too shiny for me), so I've been using a clear enamel Rustolem spray finish. I emptied an entire can on this piece, in about 3 or 4 coats. The result is a very smooth, durable matte finish that has held up great for almost 2 years hanging from my ceiling. However, it took over 2 weeks of off gassing before I left the light on for more than a few minutes.



The object is quite sturdy, and comes to seriously pokey edges. My desk is littered with fairly fragile Zcorp prints that my colleges have learned to handle carefully after a few tragic drops and smashes. I toss around the Strata print with abandon, which freaks out those who associate it with the more delicate counterparts.

Using the Rustoleum finish, the object is still pretty fragile. I make sure that people are on alert when they interact with the thing. Again, the company recommends using a superglue finish, which is VERY durable, but I find the application process really unpleasant and the finish too plasticy. After a few coats of spray finish, I am able to do things like drilling and sawing the objects.

Surface Finish


Closer inspection reveals a pretty rough surface appearance. The dimension's printing process is similar to coiling a pile of thread, with one thin continuous extrusion of plastic being coiled upon itself over and over. Back in grad school about a thousand years ago I spent some time with a similar machine. Both this newer model and the older have the same thready surface, and in sharper area you can see/feel protruding loops . From a distance this is not noticeable, but is a significant aspect of the tactile experience.

The other distinguishing characteristic of this print in this application is it's translucency. With back lighting , the dimension transmits significantly more light.


The surface of this print is quite smooth. Not autobody smooth, but like a piece of well loved sidewalk chalk. Part of this is due to the brittle product that comes out of the machine, where very sharp edges simply don't survive. But this is also a result of using the spray finish, which just slightly melts the granular surface.

The 450 is essentially printing in a gypsum based material, and is translucent only at its most thin areas. The result is a much higher contrast lamp.

I'm posting the original stl file here, so if anyone else would like to give it a whirl, send me a copy. Laser sintered steel anyone?


Bouncin' and Behavin' Building Volumes

Quick quiz: what do these shapes all have in common?

Boxes? Well, yes boxes, and we at Buildz Internationale usually shun boxes. No, the bigger idea is that they are all boxes with about 50,000 SF of floor area. But the fun part is that they are all the same family.

So lets say I have a program calling for 50,000 sf of floor area but I have freedom to make the space a tall and skinny or short and fat (I know, I know, it doesn't usually work that way, just bear with me, sheesh!). Let's also say I'm going to nail my floor to floor heights at 10'. This means at 50,000 sf x10', I will need a volume of 500,000 cubic feet of building in which to stash my floors. 500,000 CF can look like a lot of different things. However, for a boxy building, it is basically going to be width x length x height = 500,000 CF

Back to our boxes. These are each the same loaded family, with some yanked instance parameters. Looking at the parameters of the family:

There is an instance parameter for the length and width of the box, the result of which drives the height. But the height is mediated by 2 factors. The first is that I always want my length x width x height to be the same, but I also don't want my top floor to come out at some stupid height like 3.14159'. So hInteger is an integer parameter, which means it will only return whole numbers. This is the "close enough" part of the equation. I take my total volume/(length x width) to get the optimum height to get 500,000 CF, divided by ten and rounded to the closest integer gives me the number of floors in my building. Multiplied by ten feet, this now gives the resulting height of my building.

But the fun part is dropping the family into your rvt file, giving it floor area faces, and stretching it around.

Or, glom a few together to make a 200,000 sqft development.

The principles involved for making a simple box are the same for then making something more complex, like this.

All Approx 11 million sqft. For this you have to get a little more inventive to figure out how to get the top and bottom to work together, and there is a balancing act with the pleats, but the basic ideas are all the same.

Download the files and have some stretchy volumetric fun.


"Geometry is Trivial, What are you trying to accomplish?"

You can model goddamn anything in Revit.

[mmmm . . . except this:

Regardless . . . it isn't interesting to ask "Why the hell can't Revit make [insert shape]?" Because Revit can. It is YOU that is inadequate. Also, your shoes are ugly. Gaudi drew the Sagrada Familia with a ruler, a compass, a french curve, and a bunch of string. Man up.

[Oh, and it would be a major pain to make this:

The more interesting thing to ask is "why do you even WANT to make it in Revit?"

Do you want to create variations on a theme?

Do you want to understand the logic of the form?

Do you want to analyze it?

Do you want to optimize it?

Do you want to document it?

Do you want to make changes late in the game?

Do you want to build it?

Do you want to build it and still be able to pay the mortgage?

Do you drink Moxie, run marathons, or date Charlie Sheen? [This is just a calibration question to control for masochists. If you said yes, go ahead and make your goofy ass stuff in Revit. Don't forget the alligator clips and candle wax.]

If you said yes to any of these, then your life might have just got a little harder (but perhaps more interesting). If you said no, or even said yes to a couple of these, stick to your non-Workplane based, nurby, subD mesh 3d package with more robust booleans and fillets. Workplanes are a pain in the butt for freeform modeling. Make it in your unconstrained environment (without meshing it if you can), import or link the damn thing into an in-place mass, and hang some walls on it. Done.