DAM Technology

The multi-touch tables are back in Denver after a successful run in Atlanta. (There’s a summative evaluation of them that I’ll turn into a separate post which speaks to the user experience (a success!) as opposed to the hardware (they worked!))

Turns out that the problem with bespoke solutions (and believe me, these are bespoke, although easily reproducible) is that they’re very finicky in setup. Little things like the metal frames that hold the mirrors were each produced individually so each mirror and the projector angles all need to be tweaked and fine-tuned. The electrical setup has been different in each venue, the lighting is different, the museum schedule is different, whine, whine, whine. There’s also that unholy terror of realizing that I’ll get to see them every day and invariably nitpick at whatever’s not absolutely perfect.

So, yeah, a time-suck, but golly, they’re still cool. And, every time I touch one I have that “holy crap, it works” moment. Which, honestly, is kind of fun.

Exhibition, Hardware, Installation

I had ordered LaCie’s 2big Triple a few weeks ago and it just arrived today. It’s a mini RAID array (1 or 0, meaning that the drives can mirror each other for redundancy or be striped, writing data on both at the same, so it feels extra fast) and I got it with 2 1TB drives.

I’m blown away. It’s hard for me to imagine that disk space has gotten so cheap that I just added 1TB (with redundancy) to my desktop. I’ve always had a bit of a drive fetish, never quite feeling like I’ve had enough, but I think I can pause for a few minutes here, especially when I think back to my first LaCie drive when I paid $700 for 700MB.

In any case, it’s a nice bit of engineering, took just a few seconds to flip a switch to select the RAID style, and then formatted it with Disk Utility. All seamless and transparent. Oh, and it’s got a huge honkin’ blue light on the front which all technical people know — the blue LED is the sure sign of coolness.

Hardware, Medium Picture

The multi-touch tables are done and have been playing with the public in Atlanta for about two months.

We had two late-in-the-game gotchas that were a bitch to solve: - halogen lighting - surface stickiness

We’d done the initial testing and development here at the Denver Art Museum in our offices and gallery space. They have flourescent (and indirect sunlight) and incandescent lighting, respectively. The indirect sunlight was a pain in the butt — it adds a lot of infrared spectrum light (we’re looking at 945nm-ish) that we didn’t expect and adds bright areas to the camera that either become dead spots (because of background differencing) or variable “light shadows” that look like touches. We fiddled with the sensitivity of the general system and tightened up some of the shape detection that takes place to better differentiate touch vs random sunlight and got things to a pretty good point. The gallery incandescents also added a bunch of random light, but as part of the indirect sunlight solution, we got to good spot that as long as the incandescents didn’t shine directly on the table, we were good.

During the install in Atlanta, we learned that Atlanta uses halogen lighting which is as bad as direct sunlight on the table. It’s incredibly bright in our infrared range and throws everything for a loop. In an intense two week period, we became experts on window surface treatments (we couldn’t solve it exclusively through software), learning that you can’t apply stuff to acrylic (which we use for the top surface) because it out-gasses and will eventually bubble up any surface application. Likewise, we couldn’t switch to regular glass because there’s a lot of iron content in normal glass (view it on edge, it looks green. That’s the iron) which totally blocks our edge-lit IR LEDs. We eventually settled on making camera filters from the a couple of the different window surface IR blocking treatments that got us close — we were never able to perfectly solve it because we just ran out of time, but we got about 90-95% of the way there and given more time we could have dialed it in better.

The second unexpected gotcha was that the tables are too damn popular (net, a good problem) and fingerprints quickly build up during use. The museum staff are finding that during heavy use that they need to do a quick clean of the table surface with windex or similar every couple of hours. We eventually found some cleaning solutions online that help reduce fingerprint buildup which has helped a lot, but still requires some periodic low-key intervention to keep things working well. I’m thinking something bold like rain-x is a better solution, but we’ll wait until the tables are local before we really start messing with them.

It’s funny. The technology was hard and yet it’s been some decidedly low-tech issues that have kept things from being as awesome as I would like. That being said, the tables have been incredibly popular and for all reasonable intents, they’ve been a success. We’re already starting to noodle on their future life once the exhibition is finished travelling.

Big Picture, Exhibition, Hardware, Software

So, we got a Kindle. We wanted to try it out, experience the display, see if it felt like the revolution for e-books was at hand, and see if it was something hackable. Heck, maybe it could even be our device for the ultimate reference library of technology related books in the office. (20 O’Reilly books on a little transportable device would be awesome.)

I’m underwhelmed.

Don’t get me wrong, it’s a nice piece of kit, but after a couple of days and reading a bunch of stuff on it, I haven’t picked it up again. I’ve picked up lots of books since then, though.

Not to echo the rest of the detractors out there, but it bugs me that the device is a destination rather than a tool. Anything I ever purchase on the Kindle is going to stay on the Kindle. I can’t send it to another Kindle, I can’t print anything out. If I switch to another user’s account on the Kindle, the books go away (until I log back in as the original account).

And, the customer experience is just weak. It would have been awesome if Amazon had remembered all of the books that I’ve previously ordered with them and offered me a crazy discount to duplicate them onto the Kindle (those that have been Kindle-ized). But, charging me money to subscribe to a blog, and even having to cough up a credit card just to try out a subscription to a magazine or newspaper is annoying. Rather than playing gotcha, hoping I forget to cancel one of the ten or so things I’d like to try out, ask me for the cc at the end of the trial period. As it is, I’m not trying out anything since I know something will slip through the cracks.

The E-ink display is great, but that’s always been a strong point of e-ink’s stuff (random point of interest (mostly to me), my daughter’s first daycare was across the parking lot from E-ink back in Cambridge, MA). The refresh is slow — the whole screen flashes dark and then redraws — which throws a bit of a speedbump into the reading process, although I think you just sort of get used to it.

Oddly, the thing I’ve been most intrigued by is the little lcd silver slider. It’s cool — it’s silver.

The hacking community is just beginning and while Amazon has given a favorable nod to people hacking things up, there’s not a lot there yet. It’s also a shame that if I wanted to put my own content onto the device that’s yet an additional charge. It feels like there’s potential, it’s just not that friendly yet.

(And I especially hate that every time I try to pick up the device I grab one of the next / back page buttons on the side and change the page.)

Exhibition, Hardware, Installation, Medium Picture

Well, not really. It’s just a pain when the problem was painfully obvious once you realized what the problem was in the first place.

We’ve been working on our interactive table for most of the summer (unfortunately as a part-time gig, rather than the full-time attention that it demands). It’s a hard project, but we’ve been slowly chipping away. There aren’t any of those breakthrough moments where everything just falls into place, it’s just an iterative grind.

I was choking on the guts of the table for about three weeks where we need to have two projectors, two mirrors, two cameras, and a large computer. I kept moving things around in Sketchup trying to figure out the projector throws (an aside rant: why is there no 3D software that lets you model light? Not just the results, but the actual beam of light and have it visually bounce off a surface? everything is a hack.) and coming up with a solution in software that wasn’t working in real life.

The important bit to know is that most projectors don’t project evenly outward. They’re designed to sit on a table or a desk, so one side of the projection is pretty horizontal rather than an even cone.

So, I took a left turn and applied my bold papercraft skills finely honed in kindergarten. I printed out the projector throw and cut it out of a piece of paper. It was great because there was no way for the throw to not work or bend incorrectly. Even better, it was quicker to make changes than doing it in some 3D app (arguably I was using the premiere 3D app, real life).

After about 10 minutes of playing, I stopped and gave a long hard stare at what I was working on and just started laughing since it was painfully evident what my problem had been all along for the previous couple of weeks. I had one of the projector throws rotated 180 degrees and while I had thought I’d correctly measured that angles of reflection, there was no arguing with the paper that I had been getting it wrong all along.

So, lesson of the day: Got an impossible problem and no solution is working correctly? Get out the paper and scissors and do it old-school. Ironic that this would have been easier to figure out 100 years ago.

Exhibition, Hardware, Medium Picture

• Interactive Multi-touch Cocktail Bar Video 1, Video 2
  Nicely done with vvvv (an interactive toolkit worthy of a post of its own).
• Beat Blocks
  The user mixes an 8 track beat rhythm by moving blocks via a tangible interface.
• 2D into 3D images
  Software is getting better at making 3D images out of 2D. I’d seen a variation of this a few years ago out of Stanford, and MS has been doing research along these lines. Be interesting to take all of those museum paintings of architecture and see how they turn out in 3D. (Or take pictures of our new building and see if the software comes even close)
• MS Photosynth
  And here’s the MS research project. 3D environments from photo sets.
• RSS feed telegraph
  I love alternate output devices for digital stuff.
• Ikea Interactive Bed
  I don’t know that it adds to the experience, but a projected person on a bed rolls off when the mattress is touched.

Big Picture, Hardware, Software

I’m going to recycle an answer gave on the ASTC mailing list last year. The information is still valid, so apologies for anyone who’s read this before:

A question for exhibit designers. I am in charge of creating an exhibit that will have a 6 minute video loop as part of the display. Can anyone reccomend what equipment will allow me to play the video/DVD and have it loop continuously?

I’ve used a bunch of stuff over the years — starting with just DVD players then mpeg players (the adtec edje was a nice little piece of hardware) — but we’ve been moving to using mac minis as a platform for all of our video delivery.

The price point is a little higher, but it gives a lot of flexibility. We’re able to deliver multiple resolution video from the same device, if we need interactivity, it’s still the same device, if we want to change the content, we can do it remotely over the network, etc,

The software that we use was developed in-house (and we’re releasing the code as open-source sometime “soon”) but it’s essentially the quicktime player at heart, with a software wrapper.

Here’s what we’re doing, all with the same device:

• run a single video, looping, in perpetuity (we’ve had one of the machines doing this for about 12000 hours at this point, never being turned off)
• run a single video, looping, with a pause at the end and displaying a couple of still graphics of information before starting the video again (yes, you could just make that part of the video, but when we want to change that content, we change one graphic as opposed to re-rendering the whole movie)
• run a set of 16 videos in random order, with never any back to back randomness
• run different audio tracks from the video, that aren’t necessarily synchronized (a music track doesn’t need to be timed or edited to match the video).
• run a single video after receiving a touchscreen input. While the video is playing, we’re adjusting a lighting control to make an object visible in a case and then fading down the light when the video is done
• playing one of 25 videos using input from a tangible object interface (you place something in different locations on a table and it plays a different video)
• playback of three different videos synchronized (audio and video) across three different screens.

it’s also pretty small and light. I can velcro it to the back of a large plasma display or tuck one up in a drop ceiling with no problem. And heck, if I did just want to use it as a dvd player, it has a dvd drive built in, so no worries.

you get the idea. While the mini can do very simple, straightforward playback, it also opens a lot of possibilities that wouldn’t be available with just a traditional dvd player. Best of all, if I don’t need to use one in an exhibit any more, I can give it to a staff member and it’s immediately useful to them.

Hardware, Installation, Medium Picture, Software

Over the years a topic that seems to periodically resurface in people’s wish lists for museums are having electronic label displays that can change. For example, if someone had a lapel pin or visitor card that indicated a particular language preference, the label of an object would automagaically change to the new language as they approached (and presumably revert to the default when they left).

I’ve yet to really see prices drop to a price point where the idea of having a ‘universal label’ is really viable — ideally the UniversalLabel ™ would have a simple cpu, wireless connectivity, and have fairly low power requirements (batteries should, of course, last in near perpetuity). Probably the most important element would be the display technology itself and while things are getting better, we’re still a ways off.

Even more important, I’d ultimately want the display technology to be flexible. Anything that can be integrated into the experience rather than sitting on top of it is appealing to me. For example, having a label that flows along a series of objects or can bend around a pole or even can scale into a banner like experience is really interesting. To that end, here’s where flexible displays seem to be:

• E-ink
  E-ink’s been around for at least the last 6 or 7 years but only recently have they been used in any real world products (Sony’s eBook is probably the best known). Interestingly, it looks like they have a dev kit available for $3k.
• Flexible OLED (FOLED) displays
  One of the advantage of OLED over regular LED or LCD displays is that they’re surprisingly bright and readable in a variety of lighting conditions. I used to have a Motorola cell phone that had an OLED display and it was great. Universal Display Corporation seems to be at the forefront of making this available.

All the examples I’ve seen seem to use one of these two underlying technologies, especially E-inks.

Big Picture, Hardware

One of the approaches that we’ve tried to take with technology is to not necessarily get something absolutely perfect, but instead get it out into the wild and be prepared to iterate on the execution of the technology. It’s interesting — no matter how much we try to anticipate or however much we discuss how someone is going to use something, there’s always some level of surprise.

Now, I’ll be the first to say that we’re not ever going to make the ideal experience for everyone — that’s one of the reasons that I strongly believe in offering a variety of experiences (just like we offer a variety of art). But, we can certainly try to make an experience that works for as many people as possible.

For example, in the Hamilton building we created the “Select-A-Chat” which is essentially an interactive table which displays different videos when the user places an object on the topic in which they’re interested. It’s a Tangible Object Interface (a discussion for another post) which just means that instead of creating something on a computer screen, we use real world components (virtual space, meet meat space).

In the original release, we had embedded 25 RFID sensors in the table which worked pretty well except that in software we had to cycle through the sensors and proximity of an active sensor could interfere with the read at another sensor. We could cycle pretty quickly, but it was probably a full second or second and a half before we could get through all of them and start the cycle anew. In practice, while the interface worked, there was sometimes an awkwardly long pause between moving the object and the new video starting to play. Painfully long if you were the developer and wanted it to seem like magic. But, it worked.

Fast forward a few months, we realized that our problem was not stepping back and considering an alternate approach. RFID didn’t specifically buy us much in this instance since the uniqueness of the ID didn’t do anything, it was just a trigger. Realizing this, Aubrey found some magnetic sensors that could all be on simultaneously and with some decent rare earth magnets in our object, the table became almost instantaneous. The response time is amazing and for all the time that I spent agonizing about the interface that the table itself didn’t give any feedback (such as lighting up the selected topic), it became moot because the change is so fast. It’s all the feedback you need.

The first picture above shows the new simplified interior of the table, the other pictures show what we were able to rip out — big argumentative sensors and a nest of wiring.

There was also the added bonus that the guts of the table greatly simplified. So, it’s a good example of having something that worked, but being prepared to not just leave it alone and being ready to improve it when we could.

Big Picture, Hardware, Opinion, Software

While working on the table prototype, we’ve come to the conclusion that while we could wire up the power circuits and LEDs in place that it’s not necessarily a very durable design. We know these tables are going to get shipped across the country at least three or four times and the more modular and durable we can make things, the less repair and maintenance time (we hope) we’ll need over the duration of their lives.

So, we’ve decided to develop PCBs to hold the LEDs and power bus. Each PCB will hold 6 LEDs, the necessary resistor and a power bus that can be daisy-chained to additional boards. Each board will be 6″ long which lets things be pretty flexible in use, no matter how long the surface is that we’ll need to illuminate.

I’ve never designed a PCB before so it’s been interesting over the last few weeks finding decent software and some possible manufacturers. The biggest problem I’ve had with the software is either inconsistency in the parts / components that are available for each (I only need some holes, LEDs, and a big resistor) and then trying to figure out how each works. The defining requirement for me ended up having the software have some direct connection to a manufacturer so I could get an immediate price estimate for our production run.

End results? I like Pad2Pad’s software because I can create a ‘logical net’ of the circuit and then let the software define the optimal trace. But, I’ll end up using ExpressPCB because their production costs and times are almost half of Pad2Pad. It’s a shame because the Pad2Pad software really is a bit better, but I just ended up copying the design manually into the ExpressPCB software. (Of course neither saves in a format that the other will recognize. sigh)

In any case, my initial 18 boards will cost about $145 and arrive sometime late next week. These will let us do a nice job on the next prototype and make any corrections we might need to the board (watch, I’ll have designed the circuit backwards or something) before we do the final run.

Exhibition, Hardware, Little Picture