Recording Videotape

Once all the frames have been computed it is time to record them onto videotape. There are a variety of tools and techniques for accomplishing this task. See [Kennedy91] for a description of some of the variety of equipment which has been used at the Army Research Laboratory for this purpose. Only one of these techniques will be covered here.

The Abekas A60 digital video disk stores 25 seconds or 750 frames of video on a high-performance disk. Frames on this disk can be played at full video speed. The A60 provides video output as both CCIR 601 digital video and an analog signal that is either R,G,B or Y,R-Y,B-Y. This analog signal can be readily converted to a format for input to a video tape recorder.

Frames can be stored on disk either from a video input or by loading individual frames through an ethernet interface. It is the latter which makes the device particularly useful in creating computer-generated video productions.

Under BRL-CAD, access to the Abekas A60 is achieved through the framebuffer library. *|* Conceptually, the A60 has 750 unique framebuffers. The framebuffer device ``/dev/ab'' supports the A60. There are two very important options to this particular framebuffer. The first specifies the host address of the A60 on the TCP/IP network. The host name or address for the A60 is specified by an at-sign ``@'' followed by the appropriate name or address. For example, the framebuffer device ``/dev/ab@vidisk'' specifies that an A60 connected to the network with the host name of ``vidisk'' should be used. The individual frame (or framebuffer) within the A60 is specified by a sharp-sign ``#'' followed by the integer frame number. Thus the full specification of a framebuffer device might be:

This specifies the frame 27 on the Abekas A60 with the host name ``vidisk''. This can be used with any of the framebuffer utilities, such as the ``pix-fb'' utility shown here:
% pix-fb -F/dev/ab@vidisk#27 -w 720 -n 486 ell_vid.pix.27
This loads the image file ``ell_vid.pix'' into frame number 27 on the A60 called ``vidisk''.

Two more options worth noting are the ``o'' and ``v'' options to the A60 framebuffer device. The ``o'' option indicates that this is an ``output only'' access of the framebuffer. The overhead of retrieving the image already in the framebuffer is skipped when this option is specified. The ``v'' option turns on verbose logging of the access to the framebuffer. This is primarily useful to enable the user to watch the progress of the framebuffer access.

This can be generalized into a shell loop to load the 241 ``ell_vid.pix'' images into the A60.

% mv ell_vid.pix ell_vid.pix.0
% foreach i (`loop 0 240`)
? pix-fb -F/dev/abov@vidisk#$i -w 720 -n 486 ell_vid.pix.$i
? end
In this example successive images are loaded into successive video frames (framebuffers) in the A60. As each frame is loaded, the user will see output indicating that the image is converted to YUV format and then loaded into the A60.

Once the entire sequence (or a 25 second segment) has been loaded into the A60, the sequence can be played and captured using a video tape recorder. A collection of 25 second video sequences can be edited together using a television studio or post-production facility to create longer sequences.

Next Section