🚀 go-pugleaf

I'm doing a little research as to what it would take to design a
scan-doubler/enhancer.  The idea would be to feed line-level video
(or S-video) in one side, perform some digital data-amplification,
and output RGB out the other side.  The end result would be a
non-interlaced picture with twice the number of scan lines displayed
at 60Hz on a workstation monitor.

The tasks involved would be:

    * translate the line level video or s-video to RGB

    * digitize the RGB signal for storage (off-the-shelf chip from
                                                       Brooktree)

    * store scanlines in high-speed static RAM for later reference
        (simpler than VRAMs for this work, and faster too)

    * use dedicated digital circuitry to alternately reconstruct
        the original scan lines and generate new ones using interpolation

    * output the converted video stream using a RAMDAC (again, an
        off-the-shelf part from Brooktree).

Now, the only way I can design interpolation circuitry is to correctly
model the sampling operation which produced the original video signal.

Several CCD image sensors I have read about generate the interleaved
NTSC signal by simply dumping the same 200+ scan lines for each field
of a complete frame.  This means that the second field of the frame
is actually offset in time from the first field.  Is this how all
video cameras operate?

I know that motion picture film is converted to video by periodically
re-using frames to reconcile the 24-frame/sec motion picture rate with
the 30-frame/sec video rate.  Does that mean that in this case the
second field of the frame is NOT offset in time from the first field
(i.e. both fields have been recorded from the same static movie frame)?

Regards,
Glen
  
-----------------------------------------------------------------------------
Glen Sandness      Bell-Northern Research Ltd.   The above opinions are those
sandness@bnr.ca    Ottawa, Ontario, Canada       of the author, not BNR.