Tweet
Working in an architecture firm, we do not approach rendering the same way as a typical production house. Most of our users only have a cursory knowledge of hardware and software and also have a tough time coordinating with each other.
We do have a render farm, but most of the users either don't know of it or their own misconceptions prevent them from learning how to use it ("it seems so complicated"). Looking at the log, out of 30 or so vray users, only about 4-6 use our farm consistently. We've tried to handle this from the education side but most of our users simply prefer to watch the render develop on their local machine (often laptops...don't ask), no matter how good we make using the farm look.
90% of the time, we are doing still renders which makes Distributed Rendering extremely appealing in that it scales nearly linearly vs. a traditional render farm with one dedicated node per job. The only problem is right now the managing of the DR is done client side and is not dynamic. Whoever gets there first, gets it all.
I don't think we would be alone in the desire for a Distributed Rendering solution that is managed server side, with jobs split between available machines. Most importantly, it circumnavigates the user education issue. Our users rendering could be supplemented by our renderfarm with no change in the way they currently behave. They can still monitor it from their machines, except now it is supplemented by the entirety of our farm. A win-win for us.
I can understand the caveats with this solution; Instead of a first come, first served client managed setup, how would the manager decide what job goes where? Test renders are often canceled early into the job and as more users come online nodes would be diverted, thus making this whole process less than efficient. But needed, nonetheless, in my opinion.
Any thoughts?
We do have a render farm, but most of the users either don't know of it or their own misconceptions prevent them from learning how to use it ("it seems so complicated"). Looking at the log, out of 30 or so vray users, only about 4-6 use our farm consistently. We've tried to handle this from the education side but most of our users simply prefer to watch the render develop on their local machine (often laptops...don't ask), no matter how good we make using the farm look.
90% of the time, we are doing still renders which makes Distributed Rendering extremely appealing in that it scales nearly linearly vs. a traditional render farm with one dedicated node per job. The only problem is right now the managing of the DR is done client side and is not dynamic. Whoever gets there first, gets it all.
I don't think we would be alone in the desire for a Distributed Rendering solution that is managed server side, with jobs split between available machines. Most importantly, it circumnavigates the user education issue. Our users rendering could be supplemented by our renderfarm with no change in the way they currently behave. They can still monitor it from their machines, except now it is supplemented by the entirety of our farm. A win-win for us.
I can understand the caveats with this solution; Instead of a first come, first served client managed setup, how would the manager decide what job goes where? Test renders are often canceled early into the job and as more users come online nodes would be diverted, thus making this whole process less than efficient. But needed, nonetheless, in my opinion.
Any thoughts?
Comment