we try to realize an art project using vvvv. There is a canvas which is watched by a webcam and on that people stitch something in different colors and shapes. ( ex.like in image1) After they put the canvas back to its place and a screenshot will be taken. Afterwards the visual composition on the canvas is screened on to a monitor in different possibilities of the current visual (like in image2). How can I start with this algorithm? would colorkeying be an option?
Thank you very much,
there is no image…
Yes some images would help to understand your idea.
Meanwhile check this > Video Analysis/Tracking
I started to compare two screenshots using trautner and background subtraction but could not go on, because I could not answer questions:
- Background Subtraction: How can I separate the lines or dots which are detected on the empty background image?
- Trautner’s frame difference image is black and white, how can I have colored image as an output?
Did you try with Contour, here simple patch I put together, if I understood you right.
UPDATE: The filetexture/dynamictexture/blend part, pls replace with your videoin+background substractor ;) to the input of the threshold node.
contour tracker (26.7 kB)
I did it. But there are some red nodes missing: Blend/difference and threshold in your patch.:(
I think you are missing the Addons.
@homig: yes, you need the addons also try the background subtraction shader, instead of blend - sites/default/files/user-files/BackgroundSubstraction_0.zip. I think this one is the latest one, or search the forums for the latest one if this one does not work. Adjust your threshold values for best results. you can convert the x,y from the contour node to texture space and with pipet you can take the color of the shape, and apply it to the drawing nodes.