[theora-dev] Suggestions and questions
Colin Mckellar
c.mckellar at student.murdoch.edu.au
Wed Oct 2 19:16:29 PDT 2002
I am not a developer either, but I have played around with video a
bunch, so I may be able to answer some of these questions:
On Thursday, October 3, 2002, at 02:48 AM, Philip Jägenstedt wrote:
> Some ideas:
> 1. There could be the option of having a -q switch (just like oggenc)
> controlling BOTH audio and video so that for each level of q the sum
> of video and audio bitrates could be known (sort of like the nominal
> in Vorbis). In this case there should also be some switch controlling
> how much to prioritize audio and video respectively (for example:
> 0=top audio priority, 1=equal, 2=top video priority). The point would
> be the ease of predicting file size and perhaps more important the
> ease of matching corresponding video and audio qualities.
This would be a really difficult thing to do for a number of reasons...
the main thing is that quality is very subjective. Also, audio and
video quality cannot be related very well. For example, with audio
quality, any audible deviation from the original, is often
unacceptable, while with video, people can handle quite a bit of loss
in quality... mainly because there is no way to completely confuse the
eye (even with a DVD, you are still losing a lot of quality from the
studio version...
> 2. Instead of this buisness of predicting file size, could there be an
> option where you first deside file size? Is it possible to scan the
> video and audio and to somehow register an "ease of compression" value
> for bunches of audio samples and video frames? If yes, then these
> values could then be used to determine exactly how many bits allowed
> for audio and video while matching file size and still not using the
> same bitrate for the entire file. (a 0-1-2 audio-video control like in
> previous question should apply to this too).
This is commonly called "2-pass VBR". It has not been done with vp3 to
date, and it is a very uncommon thing with audio. The main problem with
2-pass VBR is that it generally takes about twice as long to encode any
media file, as the encoder essentially has to encode it twice. That
said, 2 pass VBR is very cool, and if it could be added, it would
totally rock.
> 3. Would it be possible to work into the theora framework the
> possibility that a portion of a video file could be played from an
> external file? Example: I want my Star Trek TNG episodes from DVD to
> drive. The opening and titles are the same in each episode (almost)
> and therefore I want to save space by forking them in like I said. Is
> this possibly not the job of theora but rather some higher level
> mechanism, or am I just being stupid? Perhaps something like those
> strange text files that hide in front of wmv streams are what should
> do this, sort of a playlist only one clip is forked into another. If
> the ogg had 2 logical streams there wouldn't be a problem putting
> something in between, would there?
I am not familiar with the Ogg format, but I believe that they have
tried to maintain it to be relatively simple. It focuses on audio and
video streams, and nothing else. Something like this can be done
relatively easily with a format like the quicktime format, but that
format is much more complicated than the ogg format... complexity has
side effects like low adoption rate... if the ogg format includes too
many cool, but relatively unused features that require a lot of time to
implement, then people won't support it very widely...
of course, I could be completely mistaken on this point.
> 4. Could there a standard "aspect ratio" (like DVDs and mpegs have)
> tag in the ogg file that would allow for any future players to
> prescale the video (otherwise, you must resize video prior to
> encoding, hence losing data or creating more data).
I have no idea about this. If there are aspect ratios, I request that
they are made more flexible than MPEG-2 aspect ratios.
Blibbler
--- >8 ----
List archives: http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'theora-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body. No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.
More information about the Theora-dev
mailing list