Thread-topic: VOD via darwin/RTSP is actually worser then via HTTP?
Hi.
>>Movie does not break/pixalates during seeks
>That would be a client QoI issue.
There is plenty of bandwidth for the client. The feeling actually is
more like the DSS is jumping according to the hints, rather the
keyframes in files, hence producing the distortions.
>>Fast-forward/slow-down is supported
>Again, client issue.
It never worked for me via RTSP - what player/DSS version it works for
you?
>>Moreover, the RTSP behaved quite unstable for me, the playback hanged
<>several times after seek, and only resumed after I did a seek again.
<>Also, the seeking slider sometimes jumped to points completely
unrelated
<>to where it was actually dragged.
>Client issue. Did this happened for _both_ QuickTime and VLC?
Yes.
>Since you contexualized this as video-on-demand, then all I can say is
that
>perhaps you should just use HTTP.
>Reasons to use RTSP: Live streaming; delivery to 90% of the cellphone
market
>which only support RTSP; video modifications on-the-fly; arguably
better
>performance on poorer networks (i.e., if you prefer pixelation rather
than
>pausing & skipping).
This exactly why I'm interesting in RTSP - the loading times of HTTP
could be sometimes very significant, reaching 10-15 seconds, even on
fast networks. Problem is, the pixilation matter on RTSP happens for me
even on 100 Mbit LAN, so I presume the problem is within something else
- perhaps the encoding?
What encoding settings you use for low-bandwidth (200 Kbps - 400 Kbps)
video streams?
Regards,
Stas.
-----Original Message-----
From: William Ahern [mailto:email@hidden]
Sent: Tuesday, November 18, 2008 23:07 PM
To: Stas Oskin
Cc: email@hidden;
email@hidden
Subject: Re: VOD via darwin/RTSP is actually worser then via HTTP?
On Tue, Nov 18, 2008 at 10:48:06PM +0200, Stas Oskin wrote:
> Hi.
>
> I posted it to the 2 lists, as I'm not sure whether it goes to users
> with configuration options, or developers with the possible bugs.
>
> Anyhow, I compared the overall experience of playing mpeg4 media both
> over RTSP (using Darwin 6.0.3), and over regular HTTP (using Apache),
> with players QuickTime and VLC.
>
> >From my own impression, playing over HTTP produces significant
> advantages:
>
> * Movie does not break/pixalates during seeks
That would be a client QoI issue.
> * Fast-forward/slow-down is supported
Again, client issue.
> * No need to hint the files
True, but the files still need to be post-processed to shift the moov
atom--which contains the sample tables--from the end to the beginning of
the
file. Although, this isn't necessarily needed technically, AFAIK not all
common clients have the smarts to do this correclty (i.e. stream parse
the
MOV file, and use HTTP/1.1 byte-range requests to skip the mdat section
when
loading the meta-data).
> Moreover, the RTSP behaved quite unstable for me, the playback hanged
> several times after seek, and only resumed after I did a seek again.
> Also, the seeking slider sometimes jumped to points completely
unrelated
> to where it was actually dragged.
Client issue. Did this happened for _both_ QuickTime and VLC?
> Can someone share some of his/her thoughts on the matter? What VOD
> approach do you use?
I use all of them.
Since you contexualized this as video-on-demand, then all I can say is
that
perhaps you should just use HTTP.
Reasons to use RTSP: Live streaming; delivery to 90% of the cellphone
market
which only support RTSP; video modifications on-the-fly; arguably better
performance on poorer networks (i.e., if you prefer pixelation rather
than
pausing & skipping).
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Streaming-server-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden
