The Warp Initializer and Warp Filter are independent, configurable IPs. However, the two modules work together to perform the complete warp distortion. They read the distortion parameters from a descriptor in external memory. Based on these distortion parameters, the Warp Initializer generates Warp Filter parameters, which are written into the external memory. After the Warp Filter parameters are written into the external memory, the Warp Initializer generates an interrupt. Whenever the video or warp distortion parameters are changed, the Warp Initializer must be reconfigured and run again before the Warp Filter is triggered to distort images.
The Warp Filter is a configurable core that reads the video distortion parameters and video parameters from the external memory descriptor along with the filter initialization parameters and source image data. The warp processor supports resolutions from 320×240 to 3840x2160, with up to three color components, each of 8, 10, or 12 bits. The Warp Filter reads the image from the source buffer address, using bilinear and bicubic interpolation to get a moderate and high quality warped image, which is written to the destination buffer address. The Warp Filter generates an interrupt after it completes writing the output. The source buffer and the destination buffer addresses can be dynamically changed so that you can point to any buffer address while the IP is running. The new buffer addresses are used from the next frame onwards.