Genlocking watchout, possible? Is there an equivalent?

[Claude.Rivet]

Hello,

 

I am trying to find a way to genlock Watchout or any equivalent, here is the reason/problem:

 

-Everytime I use Watchout trough a system like Vista Spyder (X20 1608) or Folsom Encore (4VP system, background connections) I have issues with content not syncing from display to display computer, we are talking minute timing differences leading to either a wobble in the blend or a tearing.

 

The problem is very obvious on scaled live video that is passed trough Watchout (being an HDSDI card we do feed it either data content (powerpoint for example), IMAG or HD video playback from an external source.

 

However the problem also is present on any video playback, pre-split or not, going trough a blend. I suspect still images have the same issue but because of lack of motion it doesn't show in an obvious manner.

 

When we look at the Spyder or Encore input for the exact aquired scan rate we realize 1 display could be at 59.9997 rather than 60 and the other 59.9995 for example.

 

What we tried:

 

-using a lot of keyframe in videos, as much as one keyframe per frame, basically compromising compression which is not a main concern to us.

 

-encoding in a lot of format, h.264, wmv, mpeg, motionjpeg...

 

The problem remains or always find it's way back. The issue being the output timing not the video sync.

 

the problem does not exist or is less of an issue when Watchout is used alone and display computers are connected straight in the projectors, however it can still happen.

 

What we haven't tried:

 

-using one multiple output computer rather than several display computer, we will be limited in terms of processing power but we can find workaround to that, maybe it's the only way, maybe it doesn't change a thing

 

-genlocking, which seems the best solution as all component in the system will slave to one clock. I have no clue how I could genlock a watchout display computer however. I can see myself genlocking the video input card as most have genlock inputs but to genlock the entire display computer is not something I tried or even know if it is possible.

 

SO:

 

-Has anyone experienced the issues we are having?

 

-Have you found a solution? Details would be appreciated.

 

-Can Watchout displays be genlocked? How?

 

Sorry about the long post and

thanks to those who will answer!

 

regards

[ARTABASE]

YOU NEED TO HAVE THE SAME FRAME RATE FOR ALL YOUR DISPLAYS

SAME HARDWARE SAME DISK IMAGE (CLONE YOUR OS)

TRY THAT

 

BEST REGARDS

[Neil Stratton]

Do you have the sound card on any of the displays connected? Just connecting one can cause sync issues and if you look round the forum you will find that it is advisable to either run audio from another machine apart from the displays, or if you are using a display machine for audio, then to have all the display machine sound cards connected (though not actually being used).

Its a long shot!

Neil

[Fredrik Svahnberg]

Claude, before we explore this further, have you tried feeding your Spyder/Encore from a single WATCHOUT computer, equipped with a multi-head graphic card?

 

Best regards,

Fredrik S

[Jonas Dannert]

Not trying to complicate things, but I just want to point to this thread on the subject of WATCHOUT and Barco Encore.

It originates from the old Showroom forum.

 

http://dataton.com/forum/topic/91-wo-v50-with-barco-encore-display-outputs-sync-problems/

 

/jonas

[Brian]

What kind of projectors are you using? This may not be a WatchOUT, Sypder, or Encore problem at all...

 

 

 

And to answer your question to the best of my knowledge: you can't exactly genlock WatchOUT itself but you -can- genlock the graphics card if you have the right one.

 

I have ATI FirePro cards in the servers I use the most with an S400 module. This allows me to either genlock to an external source or use one of the GPUs as a lock source. I typically lock 3 heads of my main server to output 1 and then use the framelock cascade feature to lock the rest of my servers to the first output (4 heads per server, 4 servers).

 

This saved my carcass on a show where I had problems with pre-split video proxies not loading properly and had to run my blend screen as discreet files! Without the framelock active I had sync issues like crazy... put the cat5 cables in, slaved the sync of all my outputs to the first output on my first server and no more issues!

 

I don't recommend doing this and not using proxies... proxies are the way to go to improve sync across multiple Displays.

 

WatchOUT itself cannot be genlocked nor can it use a genlock signal coming in if you had a way to get it into the server, but you can lock the graphics cards and heads together!

[Claude.Rivet]

It took me quite some time to get back to this thread, I got pretty busy and kind of let it go but now things are slower so I can get back to this reccuring problem.

 

Thank you all for your answers.

 

Things I need to check:

 

-using one computer with several output, I tried but only pictures were used so it's hard to say if that was the solution but it deffinitelly need more testing, it does seem like a possible solution to our issues.

 

-Using FirePro cards and genlocking them. This to me seems like a good solution as the problem really seem to be the various cards not outputting exactly 60Hz or the exact same frequency. I do not care if the software itself is genlockable what I need genlocked is my vido output so this seems like a good solution.

 

- the sound card thing is pretty interesting, I know I stumbled upon the issue with not a single sound card connected but still I have a feeling the most severe case happened with an audio connection on display 1 indeed, need to test that.

 

-As for Encore and Spyder issues I read the linked post and all that was there I already knew, I sure know about Encore being the pickiest piece of hardware not only with cable lenght but litterally anything, been working for over 6 years with Encore, most often with watchout as Background and/or foreground unit. Source link cables are all copper, short, exact same lenght and you have issues, swap source link 1 cable betwen VP1 and VP2 with a new one and the PiP on VP4 finally shows content... funny, but not really, Encore is VERY tricky. As for the Spyder x20 it basically uses internal routing so it's not a cable thing. Of course those machine might be at fault but it would have shown with other pixel pushers (playback pro, a computer with a matrox card...) but it's not the case, then again I haven't tried with other media servers, just Watchout. Know that all component are being tested everytime we use them but we do not have the time to sit down and create dedicated test and this really adds up to the troubleshooting issue.

 

Thank you all for your answers guys, awesome community we got there. I'll keep you informed if there is any devellopement. Sorry again for the delay.

[Brian]

You're using a HDSDI capture card... Are you de-interlacing your input feed?

 

If you animate the tween position/scale/rotation/etc of a still image across a blend region do you see the same problem?

 

Is the tearing only in the blend region?

 

When the problem occurs is it across all the blend regions or in any predictable manner or is it random or just a single blend region (assuming you have seen this on 3+ projector blends...)? Do you use the same projectors every time? What kind of projectors?

 

Have you tried any other scan rates or resolutions?

[Claude.Rivet]

You're using a HDSDI capture card... Are you de-interlacing your input feed?

 

If you animate the tween position/scale/rotation/etc of a still image across a blend region do you see the same problem?

 

Is the tearing only in the blend region?

 

When the problem occurs is it across all the blend regions or in any predictable manner or is it random or just a single blend region (assuming you have seen this on 3+ projector blends...)? Do you use the same projectors every time? What kind of projectors?

 

Have you tried any other scan rates or resolutions?

 

The problem is in the blend region and it affects everything, not just live feed, or animations, all of it (pre-splits, discrete files, scaled movies...). The problem shows mostly during rapidly changing content, but I suspect it is also present during still images but because of the nature of a still it doesn't show as much. HDSDI is not deinterlaced as the card is fed with a progressive signal (although we did use interlaced signal from time to time, not much different appart from the slight delay induced by the de-interlacing). We used Barcos and Christies, various models.

 

When watchout is used alone the problem is not as visible, mostly perceptible on live content but even then not much, I could genlock the input card I guess. When fed trough Encore or Spyder the added delay makes the problem worse and then it becomes an issue.

 

Since those units read the outputs of the various watchout display computer at very slightly different scan rate (59.99996Hz, 59.5554Hz) when set at 60Hz I suspect the problem could be solved but either feeding Encore and Spyder from a multiple out watchout display computer or genlocking the outputs (the firepro solution). Yes I did try 1080p 30Hz, 1280x720 75Hz and a whole lot of other resolutions and scan rates but the problem remain, the read scan frequency slightly differ on each display computer.

 

We cannot get firepro cards right now but I will deffinitely try the multiple out solution soon and get back with the result.

 

If you have any other solution to suggest go for it, this problem has been plaguing us for years but until now we worked around it, it is becoming more and more troublesome to do so since the new trend here is to use Watchout to augment pixel density not enlarge the projection surface, therefore avoiding blend areas is less and less feasible.

[Claude.Rivet]

Ok I tried feeding the Spyder with a multiple output display computer and the issue was less apparent, the frequency was read properly in the sense that the 3 outputs were read at the same frequency. Although I got some strange issue where output #3 was seen as 1280x960 even though it was set at 1280x720 in both the display computer and Watchout. Catalyst control center was confirming the output at 1280x720 and the EDID comming from the Spyder was set at 1280x720 The content was not scaled but rather letterboxed inside the 1280x960. In the end we just cropped the 1280x960 to 1280x720 but the problem remained. The problem came from the Spyder because connecting the output 3 to another monitor was giving us 1280x720.

 

Anyways, important part is we know that multiple output display computer have less issues as far as timing goes when connecting into a Spyder system. I still think genlocking GPUs and input cards from the Spyder would greatly help so I have to try this. In the meanwhile we will use multiple out display computers.

 

Fun stuff, thanks for your help so far!

[jfk]

Ok I tried feeding the Spyder with a multiple output display computer and the issue was less apparent, the frequency was read properly in the sense that the 3 outputs were read at the same frequency. Although I got some strange issue where output #3 was seen as 1280x960 even though it was set at 1280x720 in both the display computer and Watchout. Catalyst control center was confirming the output at 1280x720 and the EDID comming from the Spyder was set at 1280x720 The content was not scaled but rather letterboxed inside the 1280x960. In the end we just cropped the 1280x960 to 1280x720 but the problem remained. The problem came from the Spyder because connecting the output 3 to another monitor was giving us 1280x720. ...

x2

 

I experienced the same phenomena feeding a Christie Digital Systems Vista Spyder.

Fed three outputs to a Spyder from both a FirePro v7800 and a Radeon HD5770.

 

Catalyst Control Center and WO displays set to 1920x1080.

Spyder operator states his EDID is set to 1920x1080.

The ATi GPUs did not think so, they would output 1920x1200

with the 1920x1080 image letterboxed in the overall raster.

Initially the Spyder operator cropped the input back to the original 1920x1080 and we moved on.

Next session, we added Lightware DA2DVI-HDCP-Pro2 EDID managers to all the outputs.

The Spyder then properly received a normal 1920x1080 raster.

But of course, then all the Spyder settings were now wrong,

requiring the Spyder operator to undo the earlier workaround

(remove the crop, and reposition the WO inputs).