I am playing around with the Player (Ex9.Texture Timebased) Node
Finally I want to play 12 HD streams with about 1200 images each, and switch (hard or alpha blended) between these streams. Is this possible to do on a striped SSD raid with a AMD Firepro W7000?
which compression I should use? dxt5, filetype dds? or is this absolutely not possible to play the streams parallel?
if you want to switch between the streams then you would need to play only one at a time? or do i misunderstand your setup?
your right, my fallback solution is to play and show only one image at the same time. Perhaps this could be done with normal filetexture and getslice?
or a simple video and jump inside the clip? I did’t this in quicktime animation with OF but not in vvvv? is there a prefered codec to do this.
the fine solution should use the alpha blend. I tried it on a slightly slower machine today with spreaded player, the result didn’t meet my expectations… it worked with 6 spreaded folders… with 7 or 8 folders the framerate went terrible down.
alpha blend would need two streams which should be no problem for a recent machine. DXT compression in a DDS sequence has the best performance. DXT5 is only needed if you have alpha channel in the video… so DXT1 should work for normal videos.
ok, how should I play the two streams, with a simple Filetexture and getslice to switch the position, or with the Player node?
ok, I’ll test it tomorrow on a fast machine.
should i put alle images in one directory and jump forward or backward or swtich between directores? I think in both cases I should wait a few frames to get in sync…
I programmed it with two different player nodes. when one player is changing the directory with the images I wait for the loaded spread. then I blend it. nice.
for my understanding, does it matter whether I’m using 2 Player nodes or one spreaded Player?
this is something that @elias would know.
Well each additional player node adds a little overhead in terms of memory as they do not share memory/resources between them. So you should see a higher memory usage when using two player nodes in comparison to one spreaded player node. What they do share though is the thread pool, so in terms of performance it shouldn’t be that bad.
thanxs for the explanation.
Now I’m using the boygrouping. I’m doing the framerate and framepos calculation on the server side to get the player on the client side in sync. or is that too much traffic?
This highly depends on the network, the bandwidth is no problem, but if you have different delays to the clients they will not be in sync or play jerky. But the sync and clock nodes can help you to get a consistent time base between the machines.
@eno and @milo have some more experience, maybe they have some hints.