DBG Project Feedback System

One of the visual highlights of the Botanical Gardens gig was the use of video feedback for a large section of the dance piece. However, unlike most feedback systems, ours had to have a slower flow than you can typically get, and had to be able to self-stimulate when a dancer entered the visual area.

The system we came up with - which we dubbed "pixel glide" - was a combination of a few techniques that turned out to give good (if still somewhat uncontrollable) results. A camera fed the current image into a Jitter-based patch that provided some brightness, contrast and saturation adjustments (which I could perform from the central computer). From there, it went into a jit.slide object to slow things down a bit. The result was mixed (at about 75% to 25%) with a highly contrasted image of the sculpture's rough surface. This was then sent to the projector to complete the loop (since the camera was pointed at the projection surface).

The sculpture-texture was dark enough that the areas unoccupied by dancers offered little in the way of feedback action - the jit.slide and the low-light saw to that. However, when a dancer entered the area and was illuminated by some of the highlights of the texture, the system would kick into speed and begin flaming, roiling and twisting. Some of the activity was based on the location of the camera (on a platform about 4' high, mounted on a tripod so it was about 8' in the air), the offset of that location from the projection location, and any zoom and panning that we did. Some of the panels flamed straight up, others twisted in an arc and others created filmstrip-like image feedback along the length of the projection area.

It was a highlight, and people seems to love it. I'm sure I'll be using this one a lot more...

[ddg]

DBG Project Network Topology

Above is a basic overview of the layout for the DBG/Moore performance. The projection system featured 9 projectors (four for the walls, four for the illuminated cubes, one for the main scene background), driven by 9 computers (4 G5's, 4 Intel Mac Minis and 1 Intel Laptop). A tenth computer, my personal laptop, was the central driver.

Basically, this central control computer sent OSC messages to the other 9 computers which forced them to conform to scene changes. The videos were not hard sync'd, so there was no need to have a timing source. However, the central computer did send out "ping" messages, since otherwise we were finding that the network interfaces would fall asleep on the G5's and would stop responding after a while.

We created our own network using a Netgear router not connected to anything. My only mistake: we named the network "DBGONLY", which apparently was too interesting for some hacker-types to withstand. All afternoon I was fighting with people hopping on the network and dumping my systems off. I finally locked down the network and everything was fine.



It felt pretty cool to sit along the side of the room and control this whole system from behind a podium. Felt like God - or at least like the Wizard of Oz.

[ddg]

Relax...

The review (in class) of the networks seemed to go well. Tim had a lot to add to the discussions, including a significant amount of information on the Mayan water use model. Since that's the one that I'm going to be working on for the final project, some of the things that he talked about (instrumentation and music types, sacred water sources, current number related to the model) will help me make a more robust and interesting bit of art by the end.

Ahhhh...

[ddg]

More on self-presentation

Following on with the information received from the Self-presentation sketch...

I realize that the sketch may not represent that much on the surface. However, it is at the heart of a project that I'm working on. Basically, I want to make an Arduino-based board that will be used as a melody-making device, then use a feedback system that will allow this system to input listener preference of phrases and alter its output to suit those preferences. The circuit board is almost done (see above), and the Arduino sketch is in process - it is basically split into three sections: generation of phrases, storage and sorting of phrases, and feedback loop based on user input.

The phrase generation and storage will probably be rather simplistic, but it should allow me to test out a generative sequencing system that will be reactive to listener feedback - and thus give me in-performance change-ability without being totally tied to pre-built melodies.

Sounds fun, eh?

[ddg]

DBG visit signals change

Went over to the Denver Botanical Gardens today with a couple of projectors to test out the location. We had decided to back-project everything, but that took a great big fall when it turns out that the corners of the room have these enormous columns that eat a bunch of space we would have needed for proper project throw. Thus, we are back to projecting from the center of the floor. Luckily, there is a 4-foot riser that will hold us and the equipment, so we won't have to worry about interacting with the patrons(!).

We've nailed down the content necessary for the various sections, so Andrew will start banging out video. I've got to come up with some sort of control system for the projection devices (10 of 'em!), so I'll be busy. Hopefully, I can get Cory to give me a hand.

I'm hoping that I have section 1 done before mid-week. That will feel like maintaining a sane schedule.

[ddg]

The board work continues...

... on the Arduino music board. This will be paired up with the "repression and self-presentation" sketch that I did as a proof-of-concept for user-determined improvisation (although probably not as my final project for the networks class), but is going to be an amazing thing nevertheless. I should get a chance to present my networks in class on Wednesday, and I'm going to see what people think about this being a useful direction to go.

[ddg]

Population Update!

Spent a little time this morning adding population support and feedback to the Mayan Water Storage system. It was cool because it produced a bunch of new variables for output - the population is a lagged mirror of storage amount, and the population stress level is a jittery cool output for something that might need to be more active.

Now I feel like this system is generating a good combination of related-but-different values for the next level of abstraction.

Side note: It wasn't until I inserted the population controls into the system that I realized some of my basic calculations were off by a factor of 10. I've mopped up the previous post, and changed the patch so it does what it is supposed to.

[ddg]