Dear TI,
I have been engaged in debugging OV10640+OV490 camera module based TDA2x EVM-5777G for several days.
My setup:
OV10640------> OV490 ------> UB913 ------> UB914 -------> TDA2x EVM
( Self developed PCB ) (SAT0089 ch0)
OV490 output info:
- 1280x720,
- 12bit YUYV
- hsync, vsync, pclk (72Mhz)
- 30 fps
Usecase i used:
4CH VIP Capture + Mosaic Display
I choosed single channel mode for my experiment. And i need not to access more than one camera in the future.
I have learned the VIP section in the TRM.
As far as i know:
- All the data processing links provided in the visionSDK is based on 8 bit yuv420sp or RGB dataformat.
- VIP instances are used to split color component. Input data interface can be 8/16/24 bits.
- And then VPDMA is responsible to write these separated color component to DDR. The datatypes supported by VPDMA is limited as described in TRM.
- I can not get all the 12 bits output of OV490 correctly but configure the VIP IO to 12/16 bits raw format.
- Only 8 bit packed data format is supported by dss.
My questions:
1. Are there any mistakes of my understanding?
2. Can you kindly providing me the reference software configuration of VIP to acquire the 12bit data output of OV490.
I cannot find ov490 related configuration in file \ti_components\drivers\pdk\packages\ti\drv\vps\examples\vip\captureVip\src\CaptureVip_main.h.
Maybe it's what i need.
3. After i get the 12 bit valid YUV data, what shoud i do to convert it to 8 bit packed as supported by display. (YUV422I_UYVY ---> NV12 maybe)
12bit length per component)
Should GPU be used here?
If yes, are there any openvx libraries or any other tools can help to do this easily?
Regards,
Liu Gan.