[gst-devel] Grab video

[Andrey Nechypurenko]

Hi,

> One final question: I'd like to resize the data, e.g. from 720x480 to 360x240
> before the encode, but none of the pipelines I've tried (yes, I'm new to this!)
> seem to even want to run.  Mostly, I get messages about not being able to connect
> one endpoint to another, but I really don't understand either the messages or the
> process.

I am doing similar things and the following pipelines work for me on
x86 with x264enc and on BeagleBoard correspondently:

v4l2src do-timestamp=true !
video/x-raw-yuv,width=320,height=240,framerate=30/1 ! videoscale
name=qos-scaler ! capsfilter name=qos-caps
caps=video/x-raw-yuv,width=320,height=240 ! x264enc name=encoder
byte-stream=true bitrate=300 speed-preset=superfast subme=6 !
rtph264pay pt=96 ! appsink

v4l2src always-copy=FALSE !
video/x-raw-yuv,width=320,height=240,framerate=30/1 ! videoscale
name=qos-scaler ! capsfilter name=qos-caps
caps=video/x-raw-yuv,width=320,height=240 ! ffmpegcolorspace !
video/x-raw-yuv,format=(fourcc)UYVY ! TIVidenc1 codecName=h264enc
engineName=codecServer bitRate=30000 genTimeStamps=TRUE
byteStream=TRUE ! rtph264pay pt=96 ! appsink

I've also manage to run the pipeline which uses TIVidResize (hardware
resizer on DSP) instead of videoscale which runs on ARM. However it
works only if I precisely specify a lot of caps (such as fourcc
format, etc.). It makes it difficult to change between different
cameras, that is why I am currently not using it.

Regards,
Andrey.