PDA

View Full Version : Genlock Issues



The Pope
05-24-2006, 07:04 PM
Hi, we are a church in SW Florida using VT4 for about two months. We are having problems genlocking the SX-84 to any device.

First, we tried using VT composite outputs to the camera’s genlock in as the reference signal and the camera’s composite out to the genlock in on the SX-84. The camera genlocks but the VT will not.

Secondly, we tried an external camera as the reference signal to the genlock in on the camera and looped that signal to the genlock in on the SX-84. Once again, the camera genlocks but VT does not.

Thirdly, we tried a reference signal from an external camera to the genlock in and used the genlock loop from the SX-84 to the genlock in on the camera. Once again, the camera genlocks but the VT will not.

Fourthly, we tried an external camera as the timing reference signal on a black burst generator. From the black burst generator, we took one output to the genlock in the camera and another output to the genlock in the SX-84. Once again, the camera genlocks but the VT does not.

Our camera is Canon XL-H1 and we have a horita BG-50 black burst generator.
Our VT version is 4.6(build 6016).

We have sent our equipment to Newtek and they replaced all our boards. So the equipment is not faulty.

The NewtekRTME sees the reference signal and endlessly acts as if it is trying to genlock but never finishes. We have attached pictures of the RTME information.

Jim Capillo
05-24-2006, 09:09 PM
Try this:

1. Send a composite out from the SX-84 to the Horita reference in.

2. Run genlock out from the Horita to the genlock of each camera.

That should lock the Horita to the VT and the cameras to the Horita. You can only adjust the VT SC and H phase if it is referenced to an external piece of equipment. I think what you are seeing here is a loop where the VT is trying to lock onto itself, thereby no reference signal is actually available.

You will have to adjust the SC and H phase of each camera to the timing of the VT. VT is the reference signal for the Horita, which acts as a basic BB generator Distribution Amp (multiple outputs) to the cameras. If your cameras don't have adjustments for SC/H phase, you can adjust through the VT proc amps, in which case you really just need a video out from the camera to the input of the VT. No genlock or Horita is necessary.

Cineman
05-24-2006, 10:05 PM
Hi, we are a church in SW Florida using VT4 for about two months. We are having problems genlocking the SX-84 to any device.
Why do you need to genlock the SX-84 to any device?


First, we tried using VT composite outputs to the camera’s genlock in as the reference signal and the camera’s composite out to the genlock in on the SX-84. The camera genlocks but the VT will not.
You have created a loop.

Rather than analyze scenarios 2, 3, and 4, if you will answer the question above and the one below, and you are trying to do what I believe you are, I think that I can have you good to go in a few words of instruction.


and we have a horita BG-50 black burst generator.
For what reason do you have need for the horita BG-50 black burst generator?

Nes Gurley

The Pope
05-25-2006, 04:00 PM
Jim, we tried your suggestion and we were unable to accomplish what we wanted to which was slice the delay from our present five frames to 2 to 2.5. The camera says it is genlocked and we tried adjusting the vertical and horizontal phase on the VT but it seemed to make no change.

The only reason we are doing this is to reduce the delay for IMAG.

Cineman
05-25-2006, 07:35 PM
Jim, we tried your suggestion and we were unable to accomplish what we wanted to which was slice the delay from our present five frames to 2 to 2.5. The camera says it is genlocked and we tried adjusting the vertical and horizontal phase on the VT but it seemed to make no change.
Yes. AFAIK, there is no reason ever to need to adjust VT or cameras coming in to VT with the procamp, to Genlock cameras and minimize delay for IMAG. And, as you found, the other things that you were told to do will not work for anything.


The only reason we are doing this is to reduce the delay for IMAG.
Starting with here in early 2002

http://www.newtek.com/products/vt/profiles/TrinityUMC/index.html

I have had good success in minimizing delay, latency, for doing IMAG in many churches. Depending on your answers to my questions above I should be able to have you through this problem in very short order.

Nes Gurley

Looking back at how I split your post, I wonder if the last is your answer to my questions. Perhaps I should rephrase my question(s). I am trying to find out why you need to genlock your VT to the Horito? What type of system do you have that requires doing that?

jcupp
05-26-2006, 07:21 AM
I think The Pope is over complicating things.

Run black burst to all of the cameras from the Horito using cables of equal length and your done. There's no reason to genlock the VT in this situation you gain only complexity. Genlocking the cameras together will give you a delay through the VT of exactly two frames, which is as good as it gets.

If, after doing the above, you see more than a two frame delay it's caused by something else. Possible suspects are projectors, line doublers, scalers etc. all of which will cause additional delays.

Cineman
05-26-2006, 07:40 AM
Jeff,

I also suspect / believe that Pope is over complicating things.

However, I totally disagree with everything else you say. What you propose will do nothing to reduce latency through the VT, IMHO.

Nes Gurley

appsyscons
05-26-2006, 09:46 AM
I volunteer along side of Mark P (screen name “THE POPE”) at the church.

For some time, we have disconnected everything from the SK-84, except 1 Canon XL-H1
ginlockable camera.

We consistently measure 5 frames of delay in VT, no projector, no scaler, no line doubler.
The delay is measured at the SX-84 video out port. Additionally, when capturing both
video and audio into VT, we see that the video is 3 frames delayed from the audio.

The camera genlocks every time, regardless of the genlock methodology. We tried
sending signals to the SX-84 genlock in and also tried having nothing attached to the
SX-84 genlock in. In all cases and combinations that we have tried, we measure 5 frames
of delay in VT. The only reason we even own a black burst generator, is that this seemed
like the way to do the genlock, and minimize the delay to 2 frames.

We sent the VT boards, cables and SX-84 back to Newtek for diagnosis, several weeks
ago. They replaced the boards, so we know they are good, and we still have the same
problem.

It seems like VT does not know the genlock in in effect, and adds 3 frames during the
processing. We know we must have something simple wrong, since many of you have
this working. We desperately need to get to the 2 frames of delay in the VT.

At this point we are willing to pay someone to come in and fix it. We are 45 minutes
south of Sarasota and 45 minutes north of Ft. Myers.

jcupp
05-26-2006, 10:01 AM
Genlocking the cameras will reduce the latency from between two and three frames to exactly two frames. In other words genlocking the cameras can reduce latency by as much as one frame.

So, once the cameras are genlocked together any observed delay above and beyond those two frames is caused by something other than the VT. Line doublers, scalers and LCD projectors, by there very nature, introduce additional delays.

jcupp
05-26-2006, 10:08 AM
Appsyscons our posts crossed in the ether!

I have a few questions.

Have you tried a different camera? I have seen complants that some cameras have an internal delay. Maybe the XL-H1 does, I don't know.

What is your methodology for measuring the delay?

appsyscons
05-26-2006, 10:34 AM
The only genlockable camera we have is the Canon XL-H1, connected to SX-84, video
#1.. We have 2 other non genlockable cameras, a Canon XL-2 and a Panosonic. These 2
are not presently connected to the SX-84.

The way we measure frame delay is to video a mechanical piano metronome, set at 60
beats per minute. Run this to ‘program out’ of the VT. Connect SX-84 video out to a
monitor and set the monitor next to the metronome. Record both the live metronome and
the VT output image on the monitor, into another camcorder, not connected to the SX-84.
Play the recorded image into VT, and look at each frame in VT-edit. Count the frames
between a fixed reference point in the live image and the same fixed reference point on
the VT image. Both images are in each frame.

The monitor image is visbly behind the live opject, and is visbly perfectly in sync with the VT-vision screen and the previow out screen.

jcupp
05-26-2006, 10:48 AM
For this test you don't need a genlockable camera. If you only have one camera connected the latency will (should) be exactly two frames, so any other camera would be worth trying. Your testing methodology looks good to me - as long as you are using a CRT monitor; an LCD will introduce it's own delay.

Edit (I hit post too soon), So I'm stumped.

appsyscons
05-26-2006, 10:53 AM
Yes, we are using a CRT monitor.

Are you saying that if only one camera is connected to the SX-84, even if it is an older non genlockable camera, the VT delay is only 2 frames?

jcupp
05-26-2006, 10:58 AM
Yes;

Cineman
05-26-2006, 11:53 AM
Yes, we are using a CRT monitor.

Are you saying that if only one camera is connected to the SX-84, even if it is an older non genlockable camera, the VT delay is only 2 frames?

Yes;
First of all Frank, let me assure you that there is absolutely nothing wrong with your testing above, or your findings. Jeff Cupp is simply wrong in his assertions.

I test exactly as you describe, except that I use a strobe light where you use the metronome. That does allow me to test to a field. Every VT system, and TriCaster that I have tested since NewTek reduced the latency has showed either five or four and a half frames of delay, video in to video out, CRT. Unfortunately 4.5 frames is functionally the same as five.

Nes Gurley

Cineman
05-26-2006, 12:58 PM
The only reason we even own a black burst generator, is that this seemed like the way to do the genlock, and minimize the delay to 2 frames.
Finally, the information I needed to tell you what to do without any fear that I could be leading you down the merry path to no change or not working, once again.

Let me tell you right up front Frank, that I in no way expect you, after this, to only have 2 frames of delay through the VT. I do expect that you will have three, or possibly 2.5. AFIK NewTek has never said anything below "2 or 3 frames of delay". I have never achieved just two frames, nor has anyone that I have worked with. Some dealers have made undemonstrated claims of two. I think it must be mathematically possible, but not functionally.

Here is what to do. Connect a VT composite output to the Canon XL-H1 Genlock In connector. I am not intimate with the XL-H1, but if there is a Camera Out composite connector, that would be the best feed back to the VT for minimizing latency.

If the only composite output from the camera is after the recorder section, that could cause latency, and decreased video quality. (If there is only one such output on the camera, you can know that it is after the recorder section, since otherwise, you could not play tapes out of the camera's recorder.)

Just toss the Horita BG-50 black burst generator. It is for other things and could actually cause latency.


At this point we are willing to pay someone to come in and fix it. We are 45 minutes south of Sarasota and 45 minutes north of Ft. Myers.
Now that is a real tempting invitation, especially since your Avator appears that it might be on a golf course. But to here, it would be grossly unfair to the church, as you guys are totally capable of making the connections and testing. I expect that you guys can make BNC connections in your sleep by now.

If after doing this, you don't find, at most, only three frames of delay, or if that is not good enough for you (note that at three frames, the audio out of VT will no longer be perceptibly out of sync with the video), there are additional steps that can be taken to reduce it more (but never to less than two). For doing that I would need to become intimately familiar with your camera, possibly consult with Canon.

Please post your results.

Nes Gurley

appsyscons
05-26-2006, 01:43 PM
We would be very happy with 2-3 frames of delay.

We have already tried connecting the VT composite output to the camera genlock, and
camera output to VT video input. The camera indicates genlock, but still 5 frames of
delay.

We also tried connecting the camera composite output to the SK-84 genlock in. This
caused NEWTEKrtme to see the reference signal, but it loops forever. See pics attached
to #1, this thread. Measured delay just for fun anyway, still 5 frames.

It seems like VT would need to know about the genlock so as not to do additional
processing, which causes the added frames of delay. If this hypothesis is true, then
connecting something to the SX-84 genlock in, seems like a must.

I have tried any and all suggestions given to us, and will try another camera at our first
opportunity, and post the results. Since we are volunteers, we may not do the test until
early next week.

The picture in my Avator is my back yard, along the Myakka river.

jcupp
05-26-2006, 02:14 PM
I did some searching of the archives and came up with this:


pfrench (NewTek R & D):
The minimum possible latency to output on VT is 2.0 frames. If all your inputs and the VT are genlocked together, it will be 2.0 frames, forever.

If a video input is not genlocked to VT/house sync, it will slowly "drift" between 2.0 to 3.0 frames of latency to output. You will experience this as the occasional frame-drop or frame-double by the software TBC (just like a hardware TBC.)

I've heard Dr. Cross say the same thing so I'm not a "dealer making undemonstrated claims"; I'm a dealer quoting the developers of the software. But I may have been wrong about the need to genlock the cameras to the VT.

I just re-performed this test using a Canon GL1 and an electronic flash and the delay through my VT, with nothing genlocked, is three frames. Look at the attached RTV file (it's only four frames long); Flash-one frame-second frame-Flash.

But otherwise it looks like Nes and I agree that a five frame delay is not correct and that it should be between two and three frames.

Cineman
05-26-2006, 05:31 PM
Interesting way to test Jeff. Actually shooting a light and the monitor together, would be better and more of a proof. It is also likely to show to the nearest field that way too.

The biggest problem is that you cut it so short that it is no test at all. You must be able to count either from first on to second on, or from first off to second off. You provided neither.

Looks like a dealer test to me. ;)

Nes Gurley

Cineman
05-26-2006, 05:40 PM
Jeff,

Do you have something like a battery charger that can provide a consistent blinking light? Try shooting that in a frame with the VT output of it on the monitor at the same time. You should be able to send a couple of iterations of that.

Nes

jcupp
05-27-2006, 07:16 AM
Interesting way to test Jeff. Actually shooting a light and the monitor together, would be better and more of a proof. It is also likely to show to the nearest field that way too.

The biggest problem is that you cut it so short that it is no test at all. You must be able to count either from first on to second on, or from first off to second off. You provided neither.

Looks like a dealer test to me. ;)

Nes Gurley

Attached is a longer version of the same test clip, I don't know what else its going to tell anyone but here it is.

WISE TV
10-11-2006, 11:26 PM
I have been a toaster user for 15 years. Recently went to the new PC VT
with SX breakout box. Now all of my audio and vidio is out of sync by about
a split second. I love the VT system but I need help getting the audio
in sync. I have tried everything.

j.

ssdcinc
10-12-2006, 08:25 AM
j.

Hello, generally "Out of SYNC Video / Audio" is a thing of the past
(ie: VT1) where it can happen with type 1 DV, if this is what your
using then you should try type 2. If your using RTV this
shouldnt be an issue.

If you post your system spec's and what format your recording in,
some of the more knowledgeable members can chime in with usefull
suggestions.

so again details details details......