Evening Max, Well, perhaps there is some sort of unresolved new issue with the latest version of Red5 with regard to how it handles seeking through a FLV file. Although, I'd still believe that the placement of keyframes in the encoded video are what is causing the lack of smoothness when it comes to seeking.
Anyone else have something to contribute to this conversation? We'd have to test the same FLV file between the Red5 versions (rc2 and the latest trunk) to determine whether or not it is actually finding the closest keyframe or not. There have certainly been changes made lately - to account for larger (300+ MB FLVs) files, and faster cataloging of the meta data on keyframes. I'm not at a developer level, so perhaps there is in fact something differnet that you're noticing. Again, we'd have to test this with some kind of baseline example to find out what's actually happening. Nathan On Wed, 28 Mar 2007, Max Medvetsky wrote: > Hi Nathan, > > Thank you again for your reply. The point I'm stressing is that the > only variable in my setup is a version of a red5 server. For the sake > of argument, let's disregard for a second what is the origin of a > produced video (bitrate, codec etc) -- the main thing here is that I'm > using the same video and only change the version of a red5. > > 1. flv video stays the same > 2. client side player stays the same ( well, actually it relies on a > server part of the oflaDemo for getting a list of videos. This part > also stays intact, meaning -- I didn't touch it.) > 3. the only thing that changes is a version of red5 ( rc2 vs latest > revision from the trunk) > > Unfortunately I am out of town at this very moment, but I'm going to > provide you with an example of the same video deployed on different > versions of red5 upon my return, so you'll be able to see the > difference for yourself. _______________________________________________ Red5 mailing list [email protected] http://osflash.org/mailman/listinfo/red5_osflash.org
