ffmpeg how to crop video?

Hi

ffmpeg from git is running OK with cuda 9.1
Video encoding is OK by using following command

./ffmpeg -hwaccel cuvid -c:v h264_cuvid -i /data/1920x800.mkv -vf scale_npp=1280:720 -b:a 320k -c:v h264_nvenc -preset slow -rc vbr_hq -b:v 3M -maxrate:v 4M /data/out.avi

But i need to crop video file to 16:9 aspect ratio.
How can i do this?

I’m trying this:
./ffmpeg -hwaccel cuvid -c:v h264_cuvid -crop 10x100x10x100 -i /data/t1920x800.mkv -vf scale_npp=1280:720 -b:a 320k -c:v h264_nvenc -preset slow -rc vbr_hq -b:v 3M -maxrate:v 4M /data/out.avi
and this:
./ffmpeg -hwaccel cuvid -c:v h264_cuvid -i /data/1920x800.mkv -vf crop=1421:820;scale_npp=1280:720 -b:a 320k -c:v h264_nvenc -preset slow -rc vbr_hq -b:v 3M -maxrate:v 4M /data/out.avi
But crop no worked for me (

From FFMPEG you have this options for this decoder:

Decoder h264_cuvid [Nvidia CUVID H264 decoder]:
    General capabilities: delay avoidprobe hardware 
    Threading capabilities: none
    Supported hardware devices: cuda 
    Supported pixel formats: cuda nv12 p010le p016le
h264_cuvid AVOptions:
  -deint             <int>        .D.V...... Set deinterlacing mode (from 0 to 2) (default weave)
     weave           0            .D.V...... Weave deinterlacing (do nothing)
     bob             1            .D.V...... Bob deinterlacing
     adaptive        2            .D.V...... Adaptive deinterlacing
  -gpu               <string>     .D.V...... GPU to be used for decoding
  -surfaces          <int>        .D.V...... Maximum surfaces to be used for decoding (from 0 to INT_MAX) (default 25)
  -drop_second_field <boolean>    .D.V...... Drop second field when deinterlacing (default false)
  -crop              <string>     .D.V...... Crop (top)x(bottom)x(left)x(right)
  -resize            <string>     .D.V...... Resize (width)x(height)

Try first to make it simple:
ffmpeg -hwaccel cuvid -c:v h264_cuvid -crop 10x100x10x100 -i /data/t1920x800.mkv -c:v h264_nvenc /data/out.avi

If that doesn’t work, enable the debug mode -loglevel debug so you can see how the pipeline is executed.