View Full Version : Video Delay

09-19-2007, 05:20 PM
I am the Visual Arts director at my church. We started our 3 camera live production 2 weeks ago. We have a huge video delay issue with our BRAND NEW VT4. The computer running it is very very well powered. We by-passed the VT4 and are using an Edirol LVS-400 for our switching. Each camera has an input and the VT4 is the 4th input on it. It is so much better. Very very little delay. If anyone can explain this please let me know.

09-20-2007, 12:05 AM
Itīs just the design of VT4 (5). Itīs a software TBC and it causes an internal 3 Frames Delay. Nothing to change, you have to live with it.
Spoken with an technician at IBC, he sad Vt5 is the same design. Maybe(MAYBE) Sdi could (COULD) decrease it for a few milliseconds.


09-20-2007, 06:42 AM
To minimize image latency in IMAG situations:

1. Genlock your cameras
2. Try not to have any additional processing between the VT and Projector. ie. Scalers or Switchers like Folsoms, Extrons, Analog Way etc.
3. If your projector has some sort of image processor then turn it off. (Barcos are terrible with the image proc on)
4. Use the Component video out for best signal path to the projector. Set the projector for Y, Y-r, Y-b 480i input
5. If you HAVE to use RGB to the projector then send a signal that matches the native resolution of the projector (most common is XGA at 1024x768) If you send something other than the native resolution then the image processor has to scale it and that will add more latency.

The VT can have only 2-3 frames of latency if your cameras are genlocked. 2-3 frames is hard to notice even to a skilled eye. If you are getting 5-8 frames it can be noticed and you should look hard in your signal flow as mentioned above.

What kind of cameras are you using? I have noticed for whatever reason that Canon XL's and GL's seem to make the VT have more latency. Just my experience.

BTW the Edirol does have some latency too. It may not have as much as the VT but there is Latency since it has to time the video coming into it.

09-22-2007, 04:50 PM
I posted something last week with a similar problem. No responses yet. I'm getting 7-8 frame delay from the VT directly to a monitor (I'm still in testing phase.) I gave a lot of information about my set up. Please see if you can help.

09-23-2007, 05:57 PM
There has been SIGNIFICANT discussion about this issue many times on this forum. There are lots of experts on the subject... but my guess is that they grow weary of saying the same things over.

Search for "latency" and "delay" and you'll find numerous long threads that detail some of the issues.

Know this - it's a common problem for those using VT for IMAG (which happens to be alot of churches!) You CAN get it down to less depending on your cameras and other gear.


09-24-2007, 08:31 AM
One thing that muddies the discussion is that some who raise the question are talking about 1) delay between live action on stage and the a/v program seen at the final output device, whereas others use similar terms to refer to simple 2) audio/video sync errors. Sometimes a single thread switches back and forth between topics, making it quite confusing for all.

#1 above is generally the cumulative result of all delays from devices in the chain on the way to the output device - VT adds its (several frame) load to the larger problem. Numerous suggestions, including Jef's above and others discussing screen and speaker placement, etc., etc, can be found in these forums.

The latter (#2) is a different matter, and often occurs in layouts where video is routed through the VT but audio is not.

09-26-2007, 08:06 AM
I'll weigh in here also.

The VT has a worse case insertion delay of 3 frames. If your cameras are genlocked this drops to exactly 2 frames.

In my discussions with engineers at NewTek they claim that the VT hardware doesn't have sufficient buffer to delay more then that so a delay of seven or eight frames is impossible. So if you are seeing more than three frames of delay the additional delay is coming from some other device.

Some cautions on testing delay:

Some cameras have a built in delay.

LCD monitors have a built in delay so use a CRT for testing.

Digital processing of any kind, in any device adds delay.

Lip sync is subject to a waterfall effect - one frame you can't tell, 2 frames you can't tell, three frames you can't tell, four frames looks ok to many people, five frames looks like a Godzilla movie.

With IMAG the speed of sound is an issue.

Cable length and analog DAs do not introduce significant delays so don't sweat them beyond normal genlocking issues. 100m coax = 500ns delay or 1 frame = 80 km coax

Someone will argue with me about these points

09-26-2007, 08:22 AM
Someone will argue with me about these points

I definitely will not argue with you Jeff. You are right on with everything you said. Well done.

Those Jeff guys are pretty smart fellers.

09-26-2007, 08:33 AM
Someone will argue with me about these pointsBut not me.

Someone should comb through the relevant threads, do a little editing, then post it as a sticky. Or is this stuff, and a lot of similar 'frequently asked questions' already in an FAQ somewhere highly visible? If not, should be ...

09-26-2007, 01:03 PM
Those Jeff guys are pretty smart fellers.No matter how many 'f's they have.

09-26-2007, 03:14 PM
Everyone's so friendly now, I've seen "discussions" on this subject almost come to blows! :)

10-01-2007, 03:18 PM
Someone will argue with me about these points

I won't argue, but add, Jeff.

Audio delay is 1.13 milliseconds per foot of travel at sea level and (I think) 72° F., and such-and-so % humidity.

So, if your speakers are 100 feet from the only listener, the listener will experience a 113 millisecond delay, or .113 seconds. Just a smidge over a 1/10th of a second.


10-01-2007, 04:46 PM
Or put another way, audio is delayed by one NTSC frame for every 37 feet the listener is from the source. Technically there's delay on the light too, but for these kinds of distances it's practically instantaneous.

10-04-2007, 02:05 AM
Or put another way, audio is delayed by one NTSC frame for every 37 feet the listener is from the source...

Yes, and thats the reason when doing IMAG with VT the first row of the listeners should start at 100 feet. The only disadvantage is, that the speakers are always complaining about the long distance to the stage....

"Someone will argue with me about these points"

Not argueing, just to mention that this kind of delay is more flashy in PAL-Land, as you know we have only 25Frames per second = 40 ms. In almost 95% of IMAG situations you can`t go under 4 Frame delay (160ms) (VT + Projector) and yes, this is godzilla...

10-04-2007, 06:22 AM
Yes, and thats the reason when doing IMAG with VT the first row of the listeners should start at 100 feet. The only disadvantage is, that the speakers are always complaining about the long distance to the stage....

I'm not sure I would agree with that. If your speakers are at the front of the room, I would shoot for the delays being optimized to make the audio and video in synch at the middle of the room, not the front row.

That will have audio running ahead of video for viewers in the front row, and behind it for viewers in the back row.

If you arrange so that the front row gets perfect synch, your back row will have synch problems that are twice as bad as if the center row were the synch point.