PDA

View Full Version : Video delays



sc73
06-01-2006, 11:23 AM
I guess this is one of the old discussion about video delay but I just need to understand more about this delay and what causes it.

I'm doing a project for a church and have recommended VT4 to them against all other video mixer such as Edirol and Datavideo. After installing the system I could see a significant delay (10+ frame) that just could not sync with the Audio that comes directly from the PA.

I'm using a dual Xeon 3.6 and 2GB of RAM. Now I need to find a solution or give them a resonable answer to why there is such delay.

I've done some testing and found out that if I output directly from camera to projector it is acceptable (even if there is delay but not noticable) but if I output from VT4 its obvious. I also notice that the VTVision is not as obvious compare to the output from BOB. (if only the output of the BOB is the same as what I get from VT Vision)

Question 1 : Is this a software problem or there is a problem in the output of BOB (hardware problem)?

Question 2 : Is there anything I can do to tweak my PC setting so that I can reduce the delay? I found that the VT4 preference does not do much in helping to reduce the delays.

Question 3 : I read somewhere that Canon XL1/2 or XM2 adds 1 frame delay when going thru VT4. Is this true and no other brand of Cameras has this problem?

Question 4 : When VT5 is released, it is going to help to reduce the current relays in VT4?

I really need an answer as I've only less than 6 days before the opening of this church and there will be almost 1500 people staring at the 18ft projector screen. There will be also representative from other church to see how the system works. If there is a miracle answer out there than I definately need one now.

Cineman
06-01-2006, 04:03 PM
After installing the system I could see a significant delay (10+ frame) that just could not sync with the Audio that comes directly from the PA.
I must confess that I have never, ever, found that much delay. How did you test, prove that much? I usually find, at worst, six. That may could indicate things beyond my knowledge. The only thing that I can think of is some awful scaler in the projector if you are converting to VGA from video at that point, or the scaler being used if you are sending the projectors VGA.

At any rate, any delay of five frames or more will prove totally unacceptable to the viewers minds eye.


I've done some testing and found out that if I output directly from camera to projector it is acceptable (even if there is delay but not noticable) but if I output from VT4 its obvious.
This challenges your "10+" frames above, and indicates to me that it was an estimate.


I also notice that the VTVision is not as obvious compare to the output from BOB.
That would be correct. There is delay caused by the Analog to Digital conversion on the camera signal coming in, and the D to A going out. The VT Vision is a digital display.


Question 1 : Is this a software problem or there is a problem in the output of BOB (hardware problem)?
Neither.


Question 2 : Is there anything I can do to tweak my PC setting so that I can reduce the delay? I found that the VT4 preference does not do much in helping to reduce the delays.
No, unless the VT is part of a system instead of being the system. You did not mention it being the former.


Question 3 : I read somewhere that Canon XL1/2 or XM2 adds 1 frame delay when going thru VT4. Is this true and no other brand of Cameras has this problem?
Well. all cameras will have some delay between the actual and what goes out as video. Consumer and Prosumer camcorders that expect their video out to be from tape playback have more delay because the video signal has been processed for the camcorders, that processing has been applied even if the video is just being passed from the camera. I know that both the Sony Pro camcorders that I work with mention the perceptible delay from the video out connector as compared to camera out.

I have not heard of extra delay from the cameras that you mention. I am familiar with Canon XL1 and 2, and have a XL1 at my disposal. It was used at the same place as the Pro Sony at first, and then again between that time and another Sony Pro. Of course the delay was much more, because it couldn't be genlocked to the VT. Your mention makes me think that it might have had extra delay but that could be perception.


Question 4 : When VT5 is released, it is going to help to reduce the current relays in VT4?
delays... Well now you have asked a question that only NewTek could answer, or perhaps, not even them at this point. My personal estimation is no. I think it has already been minimized as much as is possible.


If there is a miracle answer out there than I definately need one now.
No miracles. Just the standard stuff that has worked for years now, Connect an output from the VT to the "Genlock In" connector at the camera. For minimal latency, connect the camera's Composite Out to the camera input on the Bob.

If, in testing the delay, you find it breaking on a field instead of a frame, you may be able to gain a frame less delay total by going to the camera's Advanced Menus and advancing the vertical sync. Just keep going until the camera output glitches, and then back off one click.

Tell Ahmed Hello from his Friend and Fan,

Nes Gurley

Ahmed
06-03-2006, 01:21 AM
I must confess that I have never, ever, found that much delay. How did you test, prove that much? I usually find, at worst, six. That may could indicate things beyond my knowledge. The only thing that I can think of is some awful scaler in the projector if you are converting to VGA from video at that point, or the scaler being used if you are sending the projectors VGA.

At any rate, any delay of five frames or more will prove totally unacceptable to the viewers minds eye.


This challenges your "10+" frames above, and indicates to me that it was an estimate.


That would be correct. There is delay caused by the Analog to Digital conversion on the camera signal coming in, and the D to A going out. The VT Vision is a digital display.


Neither.


No, unless the VT is part of a system instead of being the system. You did not mention it being the former.


Well. all cameras will have some delay between the actual and what goes out as video. Consumer and Prosumer camcorders that expect their video out to be from tape playback have more delay because the video signal has been processed for the camcorders, that processing has been applied even if the video is just being passed from the camera. I know that both the Sony Pro camcorders that I work with mention the perceptible delay from the video out connector as compared to camera out.

I have not heard of extra delay from the cameras that you mention. I am familiar with Canon XL1 and 2, and have a XL1 at my disposal. It was used at the same place as the Pro Sony at first, and then again between that time and another Sony Pro. Of course the delay was much more, because it couldn't be genlocked to the VT. Your mention makes me think that it might have had extra delay but that could be perception.


delays... Well now you have asked a question that only NewTek could answer, or perhaps, not even them at this point. My personal estimation is no. I think it has already been minimized as much as is possible.


No miracles. Just the standard stuff that has worked for years now, Connect an output from the VT to the "Genlock In" connector at the camera. For minimal latency, connect the camera's Composite Out to the camera input on the Bob.

If, in testing the delay, you find it breaking on a field instead of a frame, you may be able to gain a frame less delay total by going to the camera's Advanced Menus and advancing the vertical sync. Just keep going until the camera output glitches, and then back off one click.

Tell Ahmed Hello from his Friend and Fan,

Nes Gurley

This is the most comperhensive answer... thanks Nes..

You should pay us a visit.

Cineman
06-03-2006, 10:21 PM
Good to see your sig on the NewTek Forum, Ahmed.


You should pay us a visit.
Man, I would love to get back over and visit you, spend some time back in Saudi Arabia, Turkey, India, basically make a laundry list of every place that you have an office. Guess of the hops that I used to make during my younger days in television, the only one where I wouldn't be as comfortable as then, would be Teheran, Iran.

Good to see you still Clicking, and close enough to drop a line,

Nes Gurley

Keith Nealy
06-04-2006, 12:47 PM
Aloha Nes,

When you recommend your genlocking system you mention taking the camera output back to the SX8.

Is the delay still the same with composite, Y/C and component?

What is your preferred camera signal to the switcher - and do you see an appreciable difference between the three?

If you use component, do you still see the interference bars that were so much of a problem a while back?

Thanks for your help and sharing your experience to everyone.

aloha,

Keith

Cineman
06-04-2006, 09:09 PM
Aloha Keith,

Before I say anything else, I must say that I have no background testing component cameras. If someone would just give me one to test, or, if I could get my hands on one, I would more than gladly do it, because I would like to know. I have some, I think, reasonable theories on component, but when I speak of that type camera below, please know that it is just theory.


Is the delay still the same with composite, Y/C and component?
Let me address composite and Y/C since I have done that conversion. I converted to composite based on hoping to gain a frame less latency. I never achieved that.

My recommendation, if IMAG delay is the only issue, and you find delay to a field, and you have some other processing in the line, such as a DA, you might be able to get back one vertical interval earlier by eliminating both. I had already eliminated DAs before I started converting.


What is your preferred camera signal to the switcher - and do you see an appreciable difference between the three?
Again, I should concentrate on what I've proven. Between composite and Y/C, it is definitely composite. That is easily understood from resolution where I moved from approximately 425 lines of resolution to, in one case 700, and in three others 800. I think nearly doubled fits nicely in the realm of "appreciable".

I theorize Keith, that Component, if by that is meant to a CCU in close proximity to the VT, it might be even better, especially for long cable runs.


If you use component, do you still see the interference bars that were so much of a problem a while back?
Again, I have never used component. But I have a theory! I wonder if it might be better to take composite from the CCU, located as mentioned above, and run that to the VT? It may seem illogical until you consider that the analog RGB from component might be enough different from the VT's digital RGB to be a problem. I really think it would be worth someone's trying.

FWIW,

Nes Gurley

Keith Nealy
06-04-2006, 11:22 PM
Thanks for your answer Nes, - one more question for clarification?

So, you are saying that you get higher resolution from composite than Y/C?

What length cable runs do you have? Is long runs the reason for composite?

aloha,

Keith

Cineman
06-05-2006, 08:27 AM
So, you are saying that you get higher resolution from composite than Y/C?
Yes, from cameras, and from the camera section of camcorders. Have found cameras with as much as 850 lines of horizontal resolution compared with 425 (450?) from Y/C. Beware of prosumer camcorders that only have composite after the VCR processing. They can be the opposite and have additional delay.


What length cable runs do you have?
Four cameras at Trinity UMC, all with different cable lengths. Longest is just over 300 feet, and two identical cameras are of much shorter runs. There is no perceptible or measurable difference in the picture quality.


Is long runs the reason for composite?
As I said above, my purpose in making the change was in hope of eliminating an additional frame of delay, which was not accomplished. The video from Y/C was not bad. Remember that Y/C is already more resolution than NTSC broadcast.

Nes Gurley