<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
<br>
Ralph Giles wrote:
<blockquote cite="mid20061004194603.GA15954@ghostscript.com" type="cite">
<pre wrap="">On Wed, Oct 04, 2006 at 12:55:58PM +0200, Mat wrote:
</pre>
<pre wrap="">What optimization we have should work with both x86 and x86_64 cpus. You
do need a fast machine for SD res, but CIF shouldn't be a problem. By "4
BNC" do you mean capturing 4 separate CIF feeds (4x2.5 Mpixel/second)
simultanteously?
</pre>
</blockquote>
<br>
4 video analogic inputs... yes.<br>
<br>
4 x 2.5 Mp... mmm. Ah ok... you get this value with a 25 fps.<br>
For now we use 12.5 fps... so it's about 4 x 1.2 Mpixel / sec<br>
( but with motion-detection so it's not continusly with this rate ).<br>
<br>
CPU... we are using AMD Athlons XP actually.<br>
But we will pass very soon to Athlon64, or another CPU if I can find
something better on the same price level ( but I don't think )<br>
So I hope that this change will improve a lot the situation...<br>
<br>
<blockquote cite="mid20061004194603.GA15954@ghostscript.com" type="cite">
<blockquote type="cite">
<pre wrap="">- CPU optimisations... they are enabled by default, right?
</pre>
</blockquote>
<pre wrap=""><!---->Should be, if you're using the latest release. alpha6 had a bug where
they were not. If you 'make debug' from svn it should tell you what it's
using.
</pre>
</blockquote>
<br>
I'm using the alpha7 release actually...<br>
<br>
I made our theora module about 1 year ago with alpha5... and now I'm
back to check the project again.<br>
- Video quality: perfect<br>
- Size: very good<br>
- valgrind check: no mem leaks<br>
<br>
But I have to use a statement: <br>
_ogg_free( tOGGpacket.packet );<br>
in the function where I create the clip, to free a small one.<br>
If I look in the example I don't see this line and a valgrind-chechink
of the example gives me 0 leaks...<br>
So I'll check this thing later eventually.<br>
<br>
I have some problems with timestamps I think...<br>
The clips I produce go on playing for some seconds after the clip time
is passed.<br>
<br>
<blockquote cite="mid20061004194603.GA15954@ghostscript.com" type="cite">
<blockquote type="cite">
<pre wrap="">- Encoding parameters... now I use:
</pre>
</blockquote>
<pre wrap=""><!---->These looks like they're appropriate. Reducing the framerate will help a
lot, but that may not be acceptable for a DVR.
</pre>
</blockquote>
<br>
It's for video surveillance so 25 fps it's not a must... but I have to
keep at least 12.5 fps.<br>
I was thinking to other parameters I could reduce... mmm.<br>
<br>
<blockquote cite="mid20061004194603.GA15954@ghostscript.com" type="cite">
<blockquote type="cite">
<pre wrap="">I'm studying the Theora's examples and ffmpeg2theora to see if I can
optimize my code... but I don't know libogg / libtheora so I can't
understand well some functions. Is there an API reference also for
libogg... ?
</pre>
</blockquote>
<pre wrap=""><!----><a class="moz-txt-link-freetext" href="http://xiph.org/ogg/doc/libogg/">http://xiph.org/ogg/doc/libogg/</a>
It is unlikely to be a bottleneck, however.
</pre>
</blockquote>
<br>
Mmmm... ok.<br>
I'll post some code later to see if someone can help me to correct /
improve it.<br>
It would be nice to have a V4L + Theora example ( in the Theora source
package ) IMO...<br>
<br>
Greetings,<br>
-Mat-<br>
<br>
</body>
</html>