Turns out I forgot I stripped out my image of all its color to only produce white when the selected colour was present!
So instead of looking for the color in a 4 plane (ARGB) matrix, I should be looking white (val 255) in a 1-plane matrix containing just black (val 0) and white.
But I've still got a weirdly large box in that lower right window, so I zoomed in and guess what?
Turns out there really was some #noise. Raised the #threshold from 223 to 241 to see if that would help and... badabing badaboom! We've tracked just the dancing blue pencil crayon!
Let's try with another color/dancing pencil crayon:
Thhe "gold" pencil crayon seemed to require an even higher #threshold of 251 to be properly detected. This project might require lots of precision.
Future:
Seeing how sensitive Jitter is with detecting and tracking these colors, I'll have to see just how different my colors need to be for CHIMIRA to accurately track.
Now that I've gotten Max/Jitter to successfully track one color, I can move onto tracking several colors, experimenting with what colors are easier to track, and answering all my previous questions:
Questions: What are my limits? How many instances/colors can be tracked at once? What colors are tracked better? How large do the objects need to be in order for them to be detected for tracking? Is it possible to track multiple instances of the same color?
Onwards!
Comments