[theora] Indexless interpolation search seeking code
chris at pearce.org.nz
Sun Apr 25 20:57:42 PDT 2010
5-7 is pretty typical in my experience, it sometimes gets worse,
particularly for videos which have large or irregular sized keyframes,
where it's hard to get an accurate bitrate estimation on your
bissections. I asked if I could see the seek code to see if there was
some trick I wasn't doing. :)
I agree that with non-exact seeking, or with low latency storage,
indexes aren't as necessary.
All the best,
On 26/04/2010 3:45 p.m., xiphmont at xiph.org wrote:
>> sure, but you're looking at around 5-7 seeks for a random spot,
>> whereas you pretty much have 1 by definition with an index ...
> The first seek in an index gets you within a few seconds, then you
> have to search from there anyway. An interpolated bisection search
> using that same granularity would vary depending on how varied the
> bitrate changes in the file would be. 5-7 would be on the high side,
> but the number would be variable. The index would give a more
> predictable starting performance.
> Again, this all assumes that we're seeking to find some specific frame
> number, audio sample position, etc. That's considerably more work for
> both seeking cases than finding an approximate point, or moving
> forward/backward an approximate amount.
More information about the theora