[theora] mixing effects when joining videos

[Ondrej Certik]

On Thu, Jun 18, 2009 at 9:18 AM, Gregory Maxwell<gmaxwell at gmail.com> wrote:
> On Thu, Jun 18, 2009 at 10:52 AM, Ondrej Certik<ondrej at certik.cz> wrote:
>>> The recommended technique is to initially encode to lossless (if
>>> possible; usually for shorter material) or near lossless (i.e iframe
>>> dirac at zillion mbit/sec) and only do low quality transcodes at the
>>> final rate when everything is done.
>>>
>>> Failing that you don't have many other options. You can try to do some
>>> fancy stream splicing so that the unmixxed parts take no loss.
>>
>> Ah I see, thanks. Btw, something like flac for video would be handy
>> here. I am saving my screencast to png images and it's very big. For
>> this particular purpose, I think just taking the difference of
>> successive frames and encoding that as png would save lots of space,
>> I'll try it.
>
> Ogg/Dirac in lossless mode, although the cpu complexity is may be more
> than you would like.
>
> MNG is probably the best pre-existing fit for this simple application.
> Too bad no one implements it! :)

MNG is exactly what I need! I just install the advancecomp Ubuntu package and:

advmng -v -a 15 a.mng /tmp/tmpeiQlYj/screen*.png

it compresses the 300MB worth of png images into 3.2MB. It's true that
even mplayer can't play it, but if I want to get the images back, I
just do:

advmng -x a.mng

and it extracts the png images back, so that's all I want. Encoding is
very fast (it compresses 5 min screen cast in just couple seconds),
but decoding to png images is surprisingly slow (about 2 png images
per second, so you can imagine how long a 5 min screencast at 15 fps
takes) So I think that for screencast programs, it's best to save the
video into mng and audio into flac and I just archive those two files
and I can recreate any video later on.

Thanks for the tip.


Ondrej