Our platform: i.MX8M
Video decoder: ISL79987
Analog camera: 2 camera 2 lane
Field: V4L2_FIELD_INTERLACED (NTSC)
Command: yavta -fUYVY -s720x480 -t1/30 --field seq-tb -n4 -c -F /dev/video0
RAW Pixels (1440x480) (left top is first camera top part, right top is second camera bottom part, left down is first camera bottom part, right down is second camera top part)
RAW Pixels (720x480) (left top and right top mix together, left down and right down mix together)
We have the following questions.
1. We capture 720x480 frame but it seems each captured frame is 720x240. Odd field and even field are all 720x120.
Why can't we get 720x240 odd and even field?
2. The captured frame order is camera 1 -> camera 2 -> camera 1 -> camera 2.
Could we get channel 1 (camera 1) total frame and then get channel 2 (camera 2) total frame?
How could we ensure the order is correct?
3. What is the black bar between the frames? Could we have some way to remove it?
4. Our video source is NTSC. Bottom part is transmitted first and then top part.
In yavta, do we need to use interlaced-bt for field setting?
5. If the deinterlace function works, what is the expected frame behavior?
For 720x480 frame, the top part is channel 1 (camera 1) preview and the bottom part is channel 2 (camera 2) preview?
6. Are there any other tools for verifying the captured frame except yavta?
Sorry for my delay reply! see below, please!
In iMX8M internal, the CSI module is used to capture the input camera signal from MIPI CSI2, it can't indetify MIPI virtual channels, so all data will be received in time order line by line. ISL79987 doesn't have buffer in chip, so it can't output whole frame.
So I think ISL79987 can't be well supported on iMX8M. To make it work on iMX8M, the ISL79987 need merge the four camera's VSYNC to one, and even in this case, the captured video is still a line combined image.
Have a nice day!
TIC weidong sun