SPRADE5 February 2025 TDA4VM , TDA4VM-Q1
The Surround View application shows a 4-image sensor 3D Surround View image. The four cameras capture a raw image that is processed by the on-chip ISP. The four images are given to the GPU to do the 3D Surround View rendering. This image is given to the DSS to display on a screen.
The Surround View app contains two separate graphs. The first graph is simply used for generating the GPU LUT that contains the SRV bowl mapping. This graph is only executed once with the given application; however, if the GPU LUT needs to change depending on the scene, this can be run multiple times. The second graph contains the full GPU SRV, capturing using the capture node, giving the raw output to the VISS and AEWB nodes to perform the image processing. The output of the VISS is then given to the GPU to perform the rendering of the SRV image. Finally, this output image is given to the display node to be shown on the screen. Figure 1-1 shows the detailed block diagram for the SRV application.
In the current flow, the time taken to run this SRV demo using MMCSD as boot media is around 20 sec. The detailed timing diagram is as shown in Figure 1-2
This application note addresses the optimizations methodically at every stage of boot:
Each of these stages are elaborated in the coming sections.
Thus, eventually leading to running a surround view demo in approximately 3 seconds from power ON. The detailed timing diagram is as shown in Figure 1-3.