The street as platform

UTS Master of Digital Architecture class November 2008

1. mitchell 11.20.08 / 12pm

   Great! How are you digesting / outputting the results of this process? Can you probe / sample the activity in the two domains (car/pedestrian) and usefully save it out as a numerical value?

   I’d also like to see a simple slitscan that focuses on those two areas – as in my watching the street thing (nb latest images now up at http://www.flickr.com/photos/mtchl/tags/watchingthestreet/
2. mitchell 11.20.08 / 12pm

   p.s. average colour doesn’t need blob detection though right? Blobs are computer-vision objects with attributes like location, used for doing face tracking, for example.

   For average colour all you need to do is pull out the pixels you’re interested in, and do the maths…
3. bcoorey 11.20.08 / 1pm

   The blob is there to help refine the image to grayscale, and the fading will allow us to recognise more in time wether or not the cars are coming. Having a bit of trouble trying to analyse the pixels.

   I have uploaded a zip file of the sketch to the basecamp website. Could you have a look at it. I have put a function in there ‘getColour’ which attemps to get an average colour for a region. Doesn’t seem to be working properly though. Perhaps if you could give us a script that gets an average colour for a region.
4. mitchell 11.20.08 / 4pm

   OK I started to rebuild all this but it started to get complex when I busted your file loader. Here’s the function you need – nb you’ll need to comment out your global vars totalcolour and averagecolour, I have made them local to this function.

   float getColours(PImage inputImage)
   {
   //loadPixels();
   float totalcolour = 0; // these can be local variables – they get reset each time the function is called
   float averagecolour = 0;
   println(inputImage.width);

   for(int x = 0; x < inputImage.width; x++) // use width & height, that way we can pass in an image of any size
   {
   for(int y = 0; y < inputImage.height; y++)
   {
   color c = inputImage.pixels[x + (inputImage.width*y)]; // flipped this around! y = rows
   totalcolour = totalcolour + brightness(c); // just using brightness for now
   }
   }

   averagecolour = totalcolour / (inputImage.width*inputImage.height);
   //float newcolour = map((averagecolour), – 200000, 200000, 0, 255);
   //println(newcolour);
   return averagecolour; // just return the raw brightness value for now
   }

   Also I think you can just turn off all the blob stuff – this can run on the raw pixel data – the blob detection will just slow it down needlessly.
5. mitchell 11.20.08 / 4pm

   P.S and you call the function like this:

   PImage segment1 = street.get(44,183,176,85); // “get” a new PImage from a rect within street, and pass that to getColours
   float averageFill = getColours(segment1); // then get a new float value – next, use this to fill a rect, put it on screen
6. Ben 11.20.08 / 4pm

   thanks.

   got it recognising colours!

Blog at WordPress.com. | Theme: Hemingway.

Entries RSS Comments RSS