Hi,
Our sensor is sending RAW 12 ,but only 10 bits are valid. So
data need to be shift left by 2 bits.
video pipeline is currently setup as, sensor -> csi -> isp.
How and where can I have data shifted left by 2 bits in this video pipeline?
Thanks
Hi
Did you try to declare the video to be Bit in the kernel driver and in the isp-imx wrapper?
I suppose the Bayer pixel will be transmitted as 16bit/pixel anyway and the IP will do the shifting automatically according to the bit depth you have declared in the driver.
malik_cisse,
Thanks for the suggestion.
How do you "declare the video to be Bit in the kernel driver and in the isp-imx wrapper"?
Could you point out the files and format?
Thanks
Hi,
The full steps to port a new camera driver are described here:
https://community.nxp.com/pwmxy87654/attachments/pwmxy87654/imx-processors/225459/1/AN13713(8MP%20IS...
As for the bit depth in the sensor kernel driver, here is an example for ar1335 sensor:
https://github.com/nxp-imx/isp-vvcam/blob/2102360b58d9d1b36bc0c654c8301e4014b33951/vvcam/v4l2/sensor...
.bit_width also needs to be set as shown in this example:
in isp_imx\isp-imx-4.2.2.11.0\units\isi\drv\OV2775\source\OV2775.c
static struct vvcam_mode_info pov2775_mode_info[] = {
{
.index = 0,
.width = 1920,
.height = 1080,
.fps = 30,
.hdr_mode = SENSOR_MODE_LINEAR,
.bit_width = 12,
.bayer_pattern = BAYER_GBRG,
},
Note that no git repo exis for isp-imx. Only binary is provided as explained in pdf doc above. See example yocto recipe:
https://github.com/Freescale/meta-freescale/tree/master/recipes-bsp/isp-imx
IMHO these are the two only code where bit depth is set.
Experts, please correct if I am wrong!