Hi guys, as part of a community project for a local library and some school credits (not getting marked on my code or processing!)
My goal is to have a kinect set up and have people’s silhouettes made up of words (similar to Mirror Darkly project), however I’d like to integrate the words being said onto the silhouette.
My first struggle is addressing a person’s body map with any static word/words, i saw some suggestions for color to address it. but I was wondering if its possible to address the data points or if it was better to base the word formation off of a skeleton.
I’ve got the option of either OSX or W10 using Open Kinect or the SimpleOpenNI if that makes a difference.
Also, do you want the words to be cropped by the silhouette, or are you trying to pack them into the silhouette, like rectangle packing into a large polygon?
I would say one is always better than the other. You should consider both and determine which one is more convenient and richer.
Most cases it doesn’t matter what OS you use. What matters in the data that you have, your requirements and having an implementation plan: Design, code, test, evaluate, repeat.
Yes. for the first step, I’m just trying to create a silhouette mask and the word cloud is populated behind.
I’d like to not crop it, but I think in this initial step, I don’t mind.
The silhouette mask is easy to do with the values from the depth camera – you just need to set a threshold and loop over the pixels.
You could draw your text to a PGraphics, create a mask from the depth camera, apply the mask to the text image, then draw the masked image on the canvas.
Or, depending on how you are compositing, you could do the opposite and skip some steps. Draw the text directly on the canvas, then loop over the depth image pixels and mask out anything on the canvas that isn’t the silhouette – no intermediate layers required.
I’ve been able to pull an image as the background and create a silhouette mask, it was a bit trickier than I thought, just due to my lack of knowledge on masks.
I’ve been struggling to move onto the next stage and making the text to generated via Voice to Text, so I’m just trying to find a creative solution to making it a bit more interesting with some sort of select a song and the lyrics populate the background whilst music plays. This is deviation is just due to the crappy nature of Voice to Text with Chrome APIs and such.