I have an application running on a DM3730 using DVSDK 4.02.00.06 and the linux psp included with it (2.6.32) with a 12-bit greyscale camera attatched to the parallel input. I would like to get the raw frames captured from the CCDC since the greyscale data cannot be processed in the rest of the ISP. I have a setup now that uses V4L2 with the omap34xxcam driver to capture video. Looking at the ISP driver it appears that is the input format and the output format is set to one of the RAW formats (e.g. V4L2_PIX_FMT_SRGGB10) that only the CCDC is used. So I have my driver configured to be inputting "V4L2_PIX_FMT_SRGGB10" and I ask V4L2 to capture in the same format. The resulting frame data appears to be 16-bit greyscale values (not the 12- or 10-bit data that I would expect). Is there a better way of getting the raw data using V4L2?
Isn't the data 10 in a 16 bit data ... I.e. a word ?! 10 or 12 bit are not standard data, so it is packed into a 16 bit word.
Regards
Brahim
That is what I would expect a zero padded 10 or 12 bit value, but upon inspection of the data all 16 bits are varying.
Check for DATSIZ in register CCDC_SYN_MODE this has to be 0x6: cam_d is 10 bits with bridge disabled i.e. PAR_BRIDGE = 0x0 in ISP_CTRL. Also check if the INPMOD =0x0: Raw data in register CCDC_SYN_MODE.
Hello Jeffrey,
I also want to capture grayscale frames with CamISP and started a thread http://e2e.ti.com/support/dsp/omap_applications_processors/f/447/t/207989.aspx
Are you trying to capture only one color component or more than one color component. Because I checked V4L2_PIX_FMT_SRGGB10 that you mentioned and at this link http://linuxtv.org/downloads/v4l-dvb-apis/pixfmt-srggb10.html
it seems that this format is not a pure monochrome (one color component) capture format.
I want to capture 12 or 14 bit grayscale data. You can read my thread I linked above. I would be very glad if you tell me what you think about my situation.
Regards.
I am trying to achieve the same functionality as you are with the CCDC. I want to capture 12 bit greyscale data to memory.
After studying the ISP drivers provided in the DVSDK, I noticed that using the SRGGB10 RAW format for the input and output of the camera "should" cause the ISP to be configured for a pass though mode. I have had limited success with this though, because the data is coming across as 16-bit values (not 12-bit padded in 16-bit).