Tag: 3D printing

Modeling Cisterns

How do you construct a 3D model of something that lives underground and only exists in a handful of pictures taken from the interior? This was my task for the Cisterns of San Francisco last week.

The backstory: have you ever seen those brick circles in intersections and wondered what the heck they mean? I sure have.

It turns out that underneath each circle is an underground cistern. There are 170 or so* of them spread throughout the city. They’re part of the AWSS (Auxiliary Water Supply System) of San Francisco, a water system that exists entirely for emergency use.

The cisterns are just one aspect of my research for Water Works, which will map out the San Francisco water infrastructure and data-visualize the physical pipes and structures that keep the H2O moving in our city.

This project is part of my Creative Code Fellowship with Stamen Design, Gray Area and Autodesk.


Many others have written about the cisterns: Atlas Obscura, Untapped Cities, Found SF, and the cisterns even have their own Wikipedia page, albeit one that needs some edits.

The original cisterns, about 35 or so, were built in the 1850s, after a series of great fires ravaged the city, located in the Telegraph Hill to Rincon Hill area. In the next several decades they were largely unused, but the fire department filled them up with water for a “just in case” scenario.

Meanwhile, in the late 19th century as San Francisco rapidly developed into a large city, it began building a pressurized hydrant-based fire system, which was seen as many as a more effective way to deliver water in case of a fire. Many thought of the cisterns as antiquated and unnecessary.

However, when the 1906 earthquake hit, the SFFD was soon overwhelmed by a fire that tore through the city. The water mains collapsed. The old cisterns were one of the few sources of reliable water.

After the earthquake, the city passed bonds to begin construction of the AWSS — the separate water system just for fire emergencies. In addition to special pipes and hydrants fed from reservoirs for hydrants, the city constructed about 140 more underground cisterns.

Cisterns are disconnected nodes from the network, with no pipes and are maintained by the fire department, which presumably fill them every year. I’ve heard that some are incredibly leaky and others are watertight.

What do they look like inside? This is the *only* picture I can find anywhere and is of a cistern in the midst of seismic upgrade work. This one was built in 1910 and holds 75,000 gallons of water, the standard size for the cisterns. They are HUGE. As you can surmise from this picture, the water is not for drinking.cistern(Photographer: Robin Scheswohl; Title: Auxiliary Water supply system upgrade, San Francisco, USA)

Since we can’t see the outside of an underground cistern, I can only imagine what it might look like. My first sketch looked something like this.

cistern_drawingI approached Taylor Stein, Fusion 360 product evangelist at Autodesk, who helped me make my crude drawing come to life. I printed it out on one of the Autodesk 3D printers and lo and behold it looks like this: a double hamburger with a nipple on top. Arggh! Back to the virtual drawing board.IMG_0010I scoured the interwebs and found this reference photograph of an underground German cistern. It’s clearly smaller than the ones in San Francisco, but it looks like it would hold water. The form is unique and didn’t seem to connote something other than a vessel-that-holds-water.800px-Unterirdische_ZisterneOnce again, Taylor helped me bang this one out — within 45 minutes, we had a workable model in Fusion 360. We made ours with slightly wider dimensions on the top cone. The lid looks like a manhole.


Within a couple hours, I had some 3D prints ready. I printed out several sizes, scaling the height to for various aesthetic tests.


This was my favorite one. It vaguely looks like cooking pot or a tortilla canister, but not *very* much. Those three rectangular ridges, parked at 120-degree angles, give it an unusual form


Now, it’s time to begin the more arduous project of mapping the cisterns themselves. And the tough part is still finishing the software that maps the cisterns into 3D space and exports them as an STL with some sort of binding support structure.

* I’ve only been able to locate 169 cisterns. Some reports state that there are 170 and others that there are 173 and 177.

First three Data Crystals

My first three Data Crystals are finished! I “mined” these from the San Francisco Open Data portal. My custom software culls through the data and clusters it into a 3D-printable form.

Each one involves different clustering algorithms. All of these start with geo-located data (x,y) with either time/space on the z-axis.

Here they are! And I’d love to do more (though a lot of work was involved)

Incidents of Crime
This shows the crime incidents in San Francisco over a 3-month period with over 35,000 data points (the crystal took about 5 hours to “mine”).  Each incident is single cube. Less series crimes such as drug possession are represented as small cubes and more severe a crimes such as kidnapping are larger ones. It turns out that crime happens everywhere, which is why this is a densely-packed shape.


Construction Permits
This shows current the development pipeline — the construction permits in San Francisco. Work that affects just a single unit are smaller cubes and larger cubes correspond the larger developments. The upper left side of the crystal is the south side of the city — there is a lot of activity in the Mission and Excelsior districts, as you would expect. The arm on the upper right is West Portal.  The nose towards the bottom is some skyscraper construction downtown. 



Civic Art Collection
This Data Crystal is generated from the San Francisco Civic Art Collection. Each cube is the same size, since it doesn’t feel right to make one art piece larger than another. The high top is City Hall, and the part extending below is some of the spaces downtown. The tail on the end is the artwork at San Francisco Airport.



EEG Data Crystals

I’ve had the Neurosky Mindwave headset in a box for over a year and just dove into it, as part of my ongoing Data Crystals research at Autodesk. The device is the technology backbone behind the project: EEG AR with John Craig Freeman (still working on funding).

The headset fits comfortably. Its space age retro look aesthetically pleases except that I’d cover up the logo in a final art project. The gray arm rests on your forehead and reads your EEG levels, translating them into a several values. The most useful are “attention” and “meditation”, which are calculations derived from a few different brainwave patterns.

eeg_headestI’ve written custom software in Java, using the Processing libraries and ModelBuilder to generate 3D models in real-time from the headset. But after copious user-testing, I found out that the effective sample rate of the headset was 1 sample/second.* Ugh.

This isn’t the first time I’ve used the Neurosky set. In 2010, I developed art piece, which is a portable personality kit called “After Thought”. That piece, however, relied on slow activity and was more like a tarot card reading where the headset readings were secondary to the performance.

The general idea for the Data Crystals is to translate data into 3D prints. I’ve worked with data from the San Francisco’s Data Portal. However, the idea of generating realtime 3D models from biometric data is hard to resist.

This is one of my first crystals — just a small sample of 200 readings. The black jagged squares represents “attention” and the white cubes correspond to “meditation”.


Back to the sample rate…a real-time reading of 600 samples would take 10 minutes. Still, it’s great to be able to do real-time, so I imagine a dark room and a beanbag chair where you think about your day and then generate the prints.

Here’s what the software looks like. This is a video of my own EEG readings (recorded then replayed back at a faster rate).

And another view of the 3D print sample:


What I like about this 3D print is the mixing of the two digital materials, where the black triangles intersect with the white squares. I still have quite a bit of refinement work to do on this piece.

Now, the challenge is what kind of environment for a 10-minute “3D Recording Session”. Many colleagues immediately suggest sexual arousal and drugs, which is funny, but I want to avoid. One thing I learned at the Exploratorium was how to appeal to a wide audience, i.e. a more family-friendly one. This way, you can talk to anyone about the work you’re doing instead of a select audience.

Some thoughts: just after crossing the line in an extreme mountain bike race, right after waking up in the morning, drink a pot of coffee (our workplace drug-of-choice) or soaking in the hot tub!


* The website  advertises a “512Hz sampling rate – 1Hz eSense calculation rate.” Various blog posts indicate that the raw values often get repeated, meaning that the effective rate is super-slow.


3D Data Viz & SF Open Data

I’ve fallen a bit behind in my documentation and have a backlog of great stuff that I’ve been 3D-printing. These are a few of my early tests with my new project: Data Crystals. I am using various data sources, which I algorithmically transform data into 3D sculptures.

The source for these is the San Francisco Open Data Portal — which provides datasets about all sorts of interesting things such as housing permit data, locations of parking meters and more.

My custom algorithms transform this data into 3D sculptures. Legibility is still an issue, but initial tests show the wonderful work that algorithms can do.

This is a transformation of San Francisco Crime Data. It turns out that crime happens everywhere, so the data is in a giant block.


After running some crude data transformations, I “mined” this crystal: the location of San Francisco public art. Most public art is located in the downtown and city hall area. But there is a tail, which represents the San Francisco Airport.


More experiments: this is a test, based on the SF public art, where I played with varying the size of the cubes (this would be a suggested value of artwork, which I don’t have data for…yet). Now, I have a 4th axis for the data. Plus, there is a distinct aesthetic appeal of stacking differently-sized blocks as opposed to uniform ones.

Stay tuned, there is more to come!random_squares