computeShaderTextureExample

Learning Objectives

This example demonstrates how to find contours in a video using background subtraction. The compute shader is used to convert a video frame into black and white (binary) pixels by the difference between the frame and a given background image. Afterwards the openCV addon for openFrameworks is used to retrieve a contour as a vector of points.

You will learn how to..

• store the pixels of a video frame into a texture using ofTexture::loadData()
• store texture data in an ofBufferObject using ofTexture::copyTo()
• load data from an ofBufferObject to ofxCvGrayscaleImage
• compute the difference between two textures
• use the openCV contour finder in openFrameworks

In the code, pay attention to:

• the handling of GL_READ_ONLY and GL_WRITE_ONLY textures in the C++ code as well as in the shader
• the definition of working groups for parallel computation by setting dispatchCompute() from the C++ shader object and layout() from within the shader
• the functions a compute shader uses to read and write pixels in comparison to a pixel shader

Expected Behavior

When launching this app, you should see a window with two image frames. In the top right corner the framerate is displayed. The left frame shows a video of a while light with a hand that appears in front of it as a black contour. On the right side the difference between the current video frame and the defined background image is shown. It is displayed as a black and white (binary) image. Each contour computed from that image appears as a cyan line on top of the left image, framed by a purple bounding rectangle.

Instructions for use:

• Press the space key to use the current video frame as the background image that will be subtracted from all other frames in order to find contours

Other classes used in this file

This Example uses the following classes:

• ofShader
• ofTexture
• ofVideoPlayer
• ofBufferObject
• ofxCvGrayscaleImage
• ofxCvContourFinder