I think that sort-last rendering solve my problem. i understand that
in sort-last the scene is decomposed across all machines in cluster.
Then, there is a one computer which composes all decomposed parts into
the final image and displays it on its own monitor. But how to
implement sort-last in OpenSG? Any suggestions?
Thank You in advence
Aleksander Stroinski
AS> I looking for example, where one computer render the scene,
AS> which is done with 2 machines.
AS> The are some clients. Clients are running the application (modifying the scenegraph)
AS> and one server will only render.
AS> it is possible to do this kind of parallel computing with OpenSG?
CN>> Hello Aleksander,
CN>> Aleksander Stroiński wrote:
>>> Hello,
>>>
>>> I am beginner at OpenSG and i found this statement in tutorial:
>>>
>>> http://opensg.vrsource.org/trac/wiki/Tutorial/OpenSG1/Clustering
>>> "...Of course this example was just one out of several possibilities on
>>> how to use a cluster with OpenSG. Another possibility is to drive
>>> a wall of screens (I mentioned the HEyeWall earlier) or to render
>>> just an output image, but by doing it with 48 machines for example."
>>>
>>> I want to do the last thing, render high-resolution image, by doing
>>> it with many machines(clients).
CN>> hm, to be honest, this has not come up before, so there currently is no
CN>> code to do this for resolutions greater than the screen resolution.
CN>> Why do you want to interactively render hi-res images ? Or do you just
CN>> need hi-res screenshots ?
CN>> In the latter case there is some example code by Marcus Lindblom on the
CN>> wiki
CN>> (<http://opensg.vrsource.org/trac/wiki/PotentialContributions>, see
CN>> the attachments at the bottom osgutil.zip it contains a HiResScreenshot
CN>> class).
>>> In tutorial, there is only example
>>> with one client and many render-servers. My question is: how to do
>>> many client running on cluster, generating a scene and saving result in
>>> file or display on one screen(server).
CN>> I'm not quite sure I understand that last part; what do you mean by
CN>> "...many clients running on a cluster, generating a scene..."?
CN>> If the scene is so complex that creating it requires a cluster, how
CN>> could only one machine handle the rendering ? And how would the pieces
CN>> created on each machine in the cluster be put together ?
CN>> As I said, I probably misunderstood something, could you please give
CN>> some more details on what you want to do, that might allow us to give
CN>> you better advice on how to get it running.
CN>> Thanks,
CN>> Carsten
|