[Bf-committers] Problems with time in do_nla()

[Stephen Swaney]

On Thu, Jan 05, 2006 at 11:46:02AM -0700, Matthew Fulmer wrote:
> 
> Video (and audio) are fundamentally continuous, analog signals,
> broadcast throughout the real world. 

For audio, this is true.  For video, the fundamental unit is the
frame.  Yes, broadcast video is sent over an analog signal, but that
is merely the carrier.  The information is essentially a packet
whether delivered in parallel as a film frame or serially as broadcast
scanlines. ( Let us ignore interlaced video since the fields are
basically half-frames )

> 
> In film, the ideal signal is what passes into the camera and is
> then sampled into the computer. In cartoon, there really is no
> original signal, just the frames. However, blender (and any
> computer animation software) is different. Blender does not work
> with the frame by frame samples of the video; it creates them;
> it creates the ideal, continuous video signal, and can create a
> frame at any moment in the video by taking a snapshot of the
> continuous video signal and rendering it. 

A film camera is essential an analog-to-digital converter.  It takes
continuous (analog) events in the real world and samples them as
frames.  We have all seen the Nyquist consequences of this process in
the form of spoked wheels turning 'backwards'.  Cartoons, as you point
out, are frame-based and digital from the start.

> Thus, since we have access to the original video at every moment
> in time (not just a finite set of frames), frames should be more
> of an afterthought.

I am not so sure.  Our output devices are frame-based, and therefore
bandwith limited.  Sampling at multiples of the Nyquist limit does not
buy us much extra fidelity and increases the data rate greatly.  For
some effects like motion blur, being able to interpolate between frames
can be handy, but this is not the same as shifting into the analog domain.

- 
Stephen Swaney			
sswaney at centurytel.net