I build a vision machine, and I have a source of 1080p/10bit image sensor, working in 30fps.
The frame injected to the DM8168 directly from proprietary FPGA (not through TVP5158/7002) , into video port 0, which has the following signals: 24 bit data, DE, Hs, Vs.
The problem is that if I would like to send RGB of 1080p frame RAW data, 10bit depth, each pixel requires 1.25 times than it was regular RGB888.
My thought was to send the end of line sync signal a bit later than usual, while setting the VPSS driver to RAW data.
The local FAE warn me that there might be a buffer overrun issue since I'm overrunning the limitation of 1080p/8bit per frame (max buffer limitation).
My question is: Can I change the current settings or buffer size of the VPSS to support 1080p frame, 10bit depth, to accumulate larger frame size?
The reason for my need is to be able to run image processing algorithm on 10bit image (1920x1080) using DSP core on Netra. This is a mandatory request from the system.
Thanks,
David