Hello,
We are working on 1080I capture with gstreamer and V4L2 driver. Our EZSDK version: 05 05 02 00.
Our aim is to capture 1080I video, encode and transmit to other terminal, then to decode and display the video.
Now we can capture the 1080I video. It seems good but it has some problem.
In TI ezsdk, when the v4l2 driver captures the 1080i video, it performs field merge. So it converts 1080i video to 1080p in field merge process(We do not have any doc about this procedure. It looks like the driver adds two fields to one frame without any further processing. We may be wrong about this due to lack of doc.)
Our dataflow:
1080I->V4L2 capture(field merge)->1080P->encoder-> … >decoder->1080P->Display
This process generates the “combing” artifacts in interlace mode. Although DM814X has de-interlacer module, it can not be used under current driver architecture. The reason is that after the v4l2 capture driver, the video signal is already progressive.
Our question is:
How to use the de-interlacer module in DM814X with current V4L2 capture driver to remove the combing artifacts?
Or can this DEI module be used independently?
So we can change the data flow as the following:
1080I->DEI module->1080P->V4L2 capture->encoder…
On the TRM 12.3 It clearly claims it can support. But how to realize it? (we use gstreamer+v4l2)
"The DEI is primarily used to convert interlaced video source material to progressive form, which is based on the motion-adaptive de-interlacing. It performs both temporal and spatial interpolation via edge directed interpolation. It also incorporates four fields of motion detection to produce high quality output. The DEI can perform de-interlacing on up to 1080i video input source, producing 1080p video output."
Thank you very much.