The idea is to make a 3D canvas for drawing purpose which is free from the hassles of mouse or electronic slates and pens. Its happens more so often when we are sketch something and want physically move into the 3D space which is rendered in front of us on the computer screen and so that we can draw it in a better way.

​

The concept for the idea eventually build-up while exploring the kinect and its various functionality to what extent it can be hacked to make a 3D Canvas for quick sketching.

​

I started off by making a small 3D canvas in Processing (which is used as a e-prototyping tool) which can be manipulated using the mouse. The basics were making points first then selecting the two points to join them with a line. The entire program was then added with functionality of making a surface with more than two points selected.

​

Then comes the part of introducing 3-D and NUI into the picture using kinect. The kinect actually gives me a hand coordinates which can be mapped onto the application. Furthermore, the depth in the scene here the user is interacting can be used to add the third dimension.

​

Once the part was over I need to map how the user is going to select options available to for drawing i.e., making points, lines, planes and moving points. So I started off with exploring options to make the switch from one option to another using hand and finger tracking. Every digit on my hand stands for one option above. I explored ways to make the hand and finger detecting more efficient using kinect, borrowing examples from the OpenCV and Kinect. I figured out many options to do so but none of them was actually implemented for Processing. So I had to write the libraries myself and make them to work. The finger tracking was efficient but it wasn't reliable enough so that I can map them to the main application. Although, I tried hard to do so.

​

Furthermore, the next challenge was to adding the functionality of rotating the canvas in X & Y axis to ease up the drawing purpose.

 

The final prototype was done using the Touch OSC for selecting the options and Kinect for point moving and tracking. The final prototype lacks the functionality of rotating the canvas.