Latency between outputs

Hi All,

i´ve done a lot of testing during the last days.
And now another question pops up.

I´ve done some testings with one of our Systems with the following configuration:

8Core 3.2 GHz / 4GB Ram/ Nvidia 8800 running Leopard

connected to 2 Panasonic PTD 10000 with a resolution of 1400px by 1050px.

In that resolution there is a latency between output 1 and output 2 of the NVidia Card which makes it impossible to playback a movie to the hard edge alligned projectors, because the latency is that big that the movie is looking crap especially when there are fast moving elements in it.

I KNOW,THIS IS NOT A CATALYST PROBLEM
and this Problem appears also when i´m moving windows in the finder between the 2 Outputs,
but maybe anyone got an idea to solve it.
Maybe I should check the 5600 Quadro FX for that kind of Problem?
Did anyone tried the Quadro FX yet, and has the Quadro FX better syncronised outputs?

Cheers,

Holger

This might sound unconventional but I do it with hard edges and blending all the time. Use a DualHead2Go or TripleHead2Go (digital) in 2 output mode.

Don't use the second graphics card output, just the primary.

I am not sure about doing non-standard resolutions like 1400x1050 with the Matrox boxes but when I need to edge blend two HD projectors 1920x1080, this is the solution I use.

Quote:

---

Originally Posted by SourceChild

This might sound unconventional but I do it with hard edges and blending all the time. Use a DualHead2Go or TripleHead2Go (digital) in 2 output mode.

Don't use the second graphics card output, just the primary.

I am not sure about doing non-standard resolutions like 1400x1050 with the Matrox boxes but when I need to edge blend two HD projectors 1920x1080, this is the solution I use.

---

Hi Todd,

the problem with the TH2Go is that you cannot choose the resolution of 1400 by 1050 Pixel. Maybe Matrox will change that in the future. Because this is the native Resolution of almost all DLP Projectors (Aside from the HD ones).
For Example Barco R22+ has this Resolution.

Even with a Softedge Area of more than 20 Percent you can see this little latency between the outputs, causing some blurry things during fast movements. A Hard Edge Setup is almost impossible.

Thanks for your response!

Seantava

The best way to do what you want is to use a DH2G Digital (or TH2G) at 3200x1200 (2x1600x1200). The take each output and run it through an image pro to down convert from 1600x1200 to 1400x1050.

Something else you can try is to use a Mac Pro with an X1900XT in it. In many of the observations, the X1900 seemed to match outputs better in a blend.

I was having this same problem recently when I was feeding a spyder with two HD outputs (1920 x 1200) for the left and right of a blended widescreen and we observed latency between the halves as well. In this scenario, we butted the two hard edge signals up side by side in the spyder which looked great other than the latency on fast moving video. Our servers are the new MAC pros with 2GB RAM, 128GB SSD, and X1900XT Video.

In considering future projects, I think that the Tripple head is an interesting potential solution but I am concerned that we'll be a bit limited in resolution as well.

Scott Riley

Quote:

---

Originally Posted by SvenCRen

I was having this same problem recently when I was feeding a spyder with two HD outputs (1920 x 1200) for the left and right of a blended widescreen and we observed latency between the halves as well. In this scenario, we butted the two hard edge signals up side by side in the spyder which looked great other than the latency on fast moving video. Our servers are the new MAC pros with 2GB RAM, 128GB SSD, and X1900XT Video.

In considering future projects, I think that the Tripple head is an interesting potential solution but I am concerned that we'll be a bit limited in resolution as well.

Scott Riley

---

Hi Scott,

i´m really happy to hear from you. I hope that all the things are running fine for you as well.
Regarding the spyder issue, we just did another job with this system, and our operator noticed a little delay in between the outputs of the triple head to go as well. Less than a frame but noticable in fast movements.
I know that the MacPro (not the catalyst) has problems with latency between output number one and two. But from TH2Go this should work.
I also know that the spyder has a framesync option for the inputs, that option was switched off in that case (they noticed that after the show). We will do a full test setup next week to check these things out.
It´s strange to see things like these happens. We just bought a spyder system and we will use Catalyst as the background machine everytime.

And yes, for us here in Europe the limitation in resolutions of the TH2Go is a bad thing. Because we are playing almost all of our movies (when there is not enough time to do custom ones) in 25 Frames per Second. That means that all HD Resolutions of the TH2Go are not useable, because they are all 60 Hz and requiring 30 Frames.

One last Question: Did you play the 2 HD Feeds out of one system (so, did you use output one and two) ?

In the moment im waiting for a test version of the GForce Quadro FX5600, just to check out if the latency between the outputs could be solved that way. There is a Syncboard Option on the Nvidia Webpage for the FX5600 as well. To Genlock multiple systems. But the Driver is only :eek:WINDOWS:eek:!
But i know this is an expansive option. Quadro FX is around 3000$ and Syncboard around 1200$ i think. So you have to pay more bucks for your GPU than for a MacPro.

Scott, I will answer to your private mail later, because i will keep all the things regarding the latency problems public here.

Send some greetings to Andrea as well! ;)

Holger

Hello Holger,

We will definitely be doing more and more productions like this and hopefully we'll be able to come up with a solution to it. We were using both the outputs on the same machine running in widescreen mode so it was one large file across both outputs. That would eliminate the idea of a frame sync issue between layers but I find it challenging to figure out where the problem actually resides. I do suspect that it would be latency between the two outputs on the video card but I am also thinking that it could be the inputs to the Spyder.

What we observed was not consistent, sometimes the playback was just fine however the problem did happen on the backup server as well so it was not just isolated to one machine. It would be good to find out a way to test this a bit further but I am not exactly sure how because two monitors side buy side would not be a good reference in this case.

I look forward to your email regarding the other issue that I sent on to you. I have to put my video map together for that one ASAP.

Best wishes,
-scott

I have had the same problem directly off machines as well as through Spyder.

The GPU renders each output separately by the nature of how it works. Dynamics in the rendering of each side would very likely create delay artifacts.