UrToast Oct 3, 2012 @ 10:32pm
Do you agree Valve should create a render farm for SFM or to have the renderer support disturbuted computing?
I really want a scene to render fast but still have quite a lot of samples of DOF, Motion Blur, AO, etc. and a way to do this is to get several computers to render it at the same time.
Showing 1-8 of 8 comments
< >
Crazy Pie Oct 10, 2012 @ 12:59pm 
What you are suggesting would consume insane ammounts of bandwidth. If you mean you want some sort of distributed computing solution that could be used with multiple computers on a local network, I completely support the idea.
UrToast Oct 11, 2012 @ 7:26am 
Originally posted by Happyfeet:
If you mean you want some sort of distributed computing solution that could be used with multiple computers on a local network, I completely support the idea.

Yeah, sorta like Unreal Engine's "Lightmass" global illumination lightmapper, which can have rendering of a single map's lightmap done on several computers in order to speed up the process.
Bumping this over-a-year-old discussion due to interest. (Please don't hate me)

I don't really agree with having a central render farm (due to the massive bandwidth/cost), but I totally agree with Post #1's idea.

A distributed computing solution would not only be cool as hell, but it will also be a huge benefit in how we render SFM footage. It kind of makes me sad that I have at least 3 computers running Intel Core i7, and I can't just combine all three to make the rendering much faster (and I prefer to stay away from overclocking). Having a distributed computing system like BOINC (or whatever equivalent for graphics/video rendering, for example) would be a gift from heaven, especially in a situation where a team is making an SFM project.

I wonder how you send feedback/feature request to the SFM team?
Last edited by Dr. Cooldude ᕦ( ͡° ͜ʖ ͡°)ᕤ; Nov 3, 2013 @ 4:12am
The Resonte! Nov 3, 2013 @ 4:23am 
That makes sense.
Last edited by The Resonte!; Nov 3, 2013 @ 4:24am
raptornx01 Nov 3, 2013 @ 10:57am 
what you are talking about is what they original planned to do with the xbone. it was their excuse why you needed an always on connection. several games coming out (biggest being titanfall) will have a portion of their rendering done on the server. this makes up for hardware limitations. ofcourse in the end it was just an excuse to force you to have an AOC to prevent pirating and the use of used games, as well as enable their kinect data mining.

the render farm idea COULD work without eating too much bandwidth, if you set up a system where you send out your dmx save file and they send you the fully rendered scene.

as for just using multiple computers, why not send the save file to each computer, have each one render out a different clip, then combine them with a video editor?
UrToast Nov 4, 2013 @ 6:57am 
Originally posted by raptornx01:
as for just using multiple computers, why not send the save file to each computer, have each one render out a different clip, then combine them with a video editor?
Actually that's not a bad idea.
Dr. Cooldude ᕦ( ͡° ͜ʖ ͡°)ᕤ Nov 4, 2013 @ 10:40am 
Originally posted by UrToast:
Originally posted by raptornx01:
as for just using multiple computers, why not send the save file to each computer, have each one render out a different clip, then combine them with a video editor?
Actually that's not a bad idea.
This. Distributed computing would still be a neat "Nice To Have" feature though, but of course I don't expect Valve to make it right away (or to make it at all).

As for the Xbone argument, of course I don't want Valve to force us into being connected 24/7 to their servers, but more like having a few clients on a network that could be used to render in parallel (as an additional feature).
PeterPeterson Nov 5, 2013 @ 10:20am 
So who's building the CLUSTER.dll
Showing 1-8 of 8 comments
< >
Per page: 15 30 50
Date Posted: Oct 3, 2012 @ 10:32pm
Posts: 8