I have quite a bit of a problem, and I wonder if anyone else has expirienced same symptoms. I have an H.323 lab environment which consists of one 3515 MCU (12 ports), one Tandberg 1700 MXP HD-video endpoint and one Polycom PVX software endpoint that runs on a PC equipped with Polycom ViaVideo USB camera. There also is a Cisco Unified Videoconferencing Manager 5.6 which has a gatekeeper embedded and runs on a virtual machine. I do NOT use any functions of CUVCM here - it only does the gatekeeper job. All components are plugged into Cisco Catalyst Express 500 24-port switch and operate in the same VLAN and same network addressing space. So practically my lab environment has a LAN speed between all of the components.
Cisco 3515 MCU has an h.323 service configured as HD/SD continuous presence, with top resolution of 720p and max. speed of 1920 Kbits/s. Under "advanced video" settings all codecs are selected with h.264 being the most preferred.
So, theoretically, a videoconference should run real smooth in these conditions.
However, whenever I start a conference by creating it on the MCU and then dialing out to the Tandberg terminal, I get it working but the picture is smooth only if the person is very calm, and doesn't move practically at all. Any movement causes a lot of artefacts to appear on the picture, and the quality drops below any reasonable value. I also noticed that I have average video jitter of about 60ms, and I can't get the idea of how can that be in a separate LAN environment.
Is there anything that can be done to fix this?
What type of screen are you using, and consider the refresh rates.
Two, if you are getting a jitter of 60ms than either check your CAT5e cables, and or your virtual machine can't handle the burst of packets. Sounds as though the gatekeeper is drowning.
I have the ability to choose to have my Gatekeeper be part of every video conference (settings), however I have chosen just to have it do it's job and verify connection instead of being part of the conference highway.
I used one LCD monitor, and Tandberg 1700 has integrated HD LCD Screen. I suppose refresh rates were ok, as there was no problem while using the built-in Tandberg MCU - yes, I also tried using that one, and for some reason it worked fine...
As for the Gatekeeper, all it does in my scheme is hadling signalling traffic, which is fairly small, whereas media streams flow directly between endpoints. I mean it doesn't seem to terminate any of the media streams and therefore shouldn't really be affected by traffic bursts.
I have the same problem you describe.
Whether I am using a Tandberg, Polycom, Mirial or Lifesize HD Endpoint, I experience artefacs in a LAN environment.
I didn't forced the ethernet speed at 100FD yet nor lowered the MCU at 1400. I'm running v5.7 on a MCU 3515. Does anyone know where the problem is?
While it may seem a bit different from the above conversation, but my problem was solved after manually setting all component links to 100FD and then back to Auto. After that I've run a test VC for about 3-4 hours and only had one or two glitches, which is quite ok for me.
Well, I put the interface at 100/FULL and it's a bit better but still some artefacts. I disabled Qualivision and I seem to get a far better quality.
I figured the MCU interface must be also forced. I checked on the switched and I got lots of collisions due to the fact the switch interface configured in auto was in 100/Half.
I forced all the 4 interfaces at 100FD (both MCU and switch) and now, everything is working fine!
Qualivision is not degrading the picture anymore.
This must be written in Cisco's Field notices.
Yes, I've had an issue with Qualivision as well, though it was another system that uses WAN channels for VC. There was some improvement after I switched it off, but that was on 5.5 software.