All along the Bull wall, Art Deco shelters face the water. Simple structures, beautiful forms in lurid turquoise, make everything else purpley-blue.

It seems to be coming together at last, my dots are actually quite adorable! Today I started to tweak the values to make the dots sensitive to different aspects of the music… trying to find the right speeds, the best frequency of bangs for certain functions. Eg, they get more excited and turn faster with high pitch notes… they get closer together in their pairs when getting closer to another pair… they each have different sensitivities to attacks and frequencies in the music so they change direction and the center point about which to turn in their couples at different times. They have really come alive. It’s incredibly simple to look at – just coloured dots on a white background, but the their movements are full of personality! A screen shot wouldn’t do it justice, their beauty is in the way they move. I tried to take a video screen grab to post an example of them dancing to some music, but my laptop is having none of it… I suppose she has been overworked today.

At the beginning of this project, when I was thinking about using flash to generate the pattern that would be created, I found a website that inspired me (I talked about it in the post ‘What to do with the data‘). Working with a friend (Mitchell DeBruyn) I wrote some action script trying to achieve a similar result. Subsequently I decided not to use flash for the project, mainly because I couldn’t find out how to get information from the real world into and out of the program in real time, plus I already had those systems in place with maxMSP from the previous project. Anyway, I’m very pleased (because it took us bloody ages!) to announce that the code has been given life elsewhere. I used it as the basis of the home page navigation device on the MA catalogue website, http://maexhibitions.com which I created in collaboration with Jonathan Harvey (who graduated from the same course last year) and Charlotte Vogel (who graduates next year).

Another website that I liked when I was looking around thinking about visuals was storynest. It has some really amazing flash animation and sound design work on it, the page I have linked to is one I particularly liked and which actually reminds me of some of the patterns created in the dance piece ‘glow’ – the video I posted yesterday.

Actually, that expression calculating a mid point about which to turn (that I mentioned in a previous post) wasn’t useful in the end. Instead, I used sin and cos of an angle multiplied by a variable radius to describe points on a circle that can change size, and a counter to change the angle over time so that by using S=O/H C=A/H T=O/A (another blast from the past) I could animate a circle moving on a circular path. The programing on the turquoise background describes how the circles orbit each other within their pair, and on the blue background how the pair move on a circular path around the space.

By changing the direction the counter counts (up or down) I can reverse the direction of travel. Around the space, I used the equation I mentioned before (the one that determines when two circles get close to each other) to initiate a change of direction around the room for the trailing couple, this prevents collisions. But, since in a milonga the direction of travel around the dance floor is anticlockwise, I also added a delay instruction on the message so that after a second the couple turn back and continue in an anticlockwise direction.

Within the pair, I used the objects ‘bonk’ and ‘fiddle’ to determine when to initiate a change of direction. Bonk listens for attacks in the music, which worked well with the percussive sounds of the piano. Fiddle listens for pitch and different types of sound, which worked better with the bandoneon. The idea is that the circles will move in accordance with selected parts of the music, as would an improvising dancer.

Below is the last version of boids I used before I made the decision to give up on them completely and begin to program from scratch a set of rules for pairs of coloured circles to obey. I just found the boids object to be too chaotic. No matter how I calibrated it, the results looked like flocking insects or animals (not surprisingly) but not like a model of human behaviour in a milonga. Back to the drawing board. Not entirely wasted though, I took forward a couple of nifty equations worked out for calculating distances between two objects in a 2D space – the expression: expr sqrt((($f1-$f2)*($f1-$f2))+(($f3-$f4)*($f3-$f4))) is based on the trig equation: a squared + b squared = c squared, to give the hypotenuse of a right angle triangle. (dredged that one up from the recess of my GCSE maths mind. Never thought it would be useful again at the time!). Also the expression: expr (($f2-$f1)/2)+$f1 to find the coordinate midway between two other coordinates on one axis. this is the point about which the pair of dots orbit. actually, I could vary this value so that they don’t always orbit about their combined centre. Hmm. must do that tomorrow.

Anyway, the version shown in this screen grab had five pairs of boids, each pair had an attract point governed by the position of one of the five boids in the boid6 object. The 6th boid of boid6 would have been invisible, a space for the real dancing couple. This one was intended to keep the other couple representations away from the real couple.

My programming isn’t up to much, but I imagine the way to go about simulating a dance floor would be to come up with some simple rules to govern decisions made on the direction taken by each dancing couple.

1. Dancing couple comprises two people. Person A, person B.
2. A is the leader, therefore most of the time A walks forward, B walks backward, either diagonally or directly in front of A.
3. Sometimes B moves around A, very occasionally A moves around B.
4. Due to the embrace being uneven, moves are easier on the open side, which means there is a tendency towards anti-clockwise rotation.
5. Each dancing couple has to maintain the distance between them and the couples in front of them. If a space opens up in front, slightly to one side, they move to fill it. (still following the movement rules 1 – 4)
6. If the couple in front is dancing a little bit wildly, the distance left between them and the next couple would be greater.

I don’t know, maybe it’s interesting, maybe not… Any dancers reading this who would agree or disagree with those rules?
I might film a milonga from above and trace the paths of each person, see if they hold true, see if the paths make a nice pattern.

I think new development in the visuals is less of a comment on the sensation or internal aspects of the dance, and more about the social, external experience of a dance floor. Its a representation of action and reaction between me and my partner, and other dancing couples, and also to some extent the music. The freezing aspect is a musical accent: when the music freezes, the dancers may well do too, and so it makes sense that the visuals do the same. The various objects orbiting the tracked object are representing other couples on the dance floor, and so it also makes sense that when a grand gesture is made, for a example an energetic boleo (kick), the surrounding objects (representing other people) would get out the way. My challenge with the patch is to make the boids behave in the manor of other couples on the dance floor. To do this, I need to program each object to orbit in a pair, then each pair to orbit the space (anticlockwise is the tango convention) whilst avoiding each other and reacting to each other in other ways (for example staying further out of the way of more energetic pairs).

I’ve altered the patch so that instead of sending the coordinate information to a graphic window, it gets sent to a matrix which loads colour pixels into a jit.window – a video window. This was to make it more compatible with the existing patch (the one with the ripples), and also to give me more freedom to process the information with video filters.

In these screen grabs, the pattern has been built up by a pair of boids orbiting the mouse position (which will eventually be the performing couple’s position on the floor) and leaving trails of their paths.

I was thinking that perhaps the dancers should erase the pattern with their actual paths so that they are always dancing in a clear space. I’d also like to make the orbiting boids leave a smoother and more delicate trail… I guess its something to do with the metro timing but I haven’t figured it out yet.