I am aware that this is a rather broad question, but after searching around for a distributed rendering solution, the only real viable pieces of software that I have come across are:
OpenMosix
DrQueue
and Farmerjoe: The Render Farmer
A friend of mine played with OpenMosix (with ClusterKnoppix) and although he said that it worked great and decreased render times, it only runs on linux (at least, not on windows). So, that solution is no longer a choice as I use windows as my primary OS.
I tried installing DrQueue and it seemed as if no matter what tutorial I followed (the main article I followed is here), DrQueue just didn’t want to play friendly with any of the other machines on the farm. Sometimes they would finally connect, until a job was distributed, in which case it all fell apart.
And, farmerjoe is the only solution that I found that works reliably and well, except for the fact that it doesnt support true animation rendering. It will render each frame, but unlike rendering it on a single machine, it won’t output to a quicktime (or whatever codec your heart desires) file for immediate viewing.
So my question to all of you is, is there a distributed rendering application that:
Works with Blender and YafRay
Supports Multiple CPU’s (per Machine)
Runs on Multiple Operating Systems
Is Fast (obviously)
Although not required: Supports Bucket Rendering (Distributing a Single Frame to have multiple nodes work on it)
Supports True Animation Distribution (Where it will generate the actual movie file instead of 1000’s of JPEG’s)
And, of course, if it was relatively easy to setup that would be a nice plus. Does anyone know of software that will do this (like Backburner for 3DS Max)?
If DrQueue will do this, can anyone point me to a tutorial that will leave me with a fully running, ready to process, renderfarm?
expecting a movie instead of single frames is totally insane.
no one wants a compressed movie as the result of days of rendering, neither Pixar or the boy experimenting in his own pc.
you need image sequences (not jpeg, but lossless png or even better openexr) to make postprocess compo or nle or just to re-encode them in the different formats you need (hd, pal, web, etc…)
I use FarmerJoe - very easy to setup and very good. Yes it creates a sequence of images but it neatly puts them all in a folder and you can use a free program like virtualdub to open up the files as a sequence and view the animation at anytime during or after the render.
You can also open them in the blender sequence editor: add images, go to the folder where 10000’s of tga/png/exr are, and put in the name file box Frame* for example, if you have Frame0000.png, Frame0001.png, etc. They appear as a single chunk, as any other video strip.
Yeah, I see your point Jazzroy. The main reason I was asking for a solution that would package the frames up for me was that I couldn’t seem to find a free solution that would do it for me (I didn’t realize that VirtualDub or even Blender would do it for me and ImageReady just doesn’t cut it!)
So, it sounds like FarmerJoe is the way to go. Thanks for your help.
Apple’s Xgrid on a local subnet works wonders…provided all the machines are macs or at least Unix based. Using the internet is a tad tricky, especially if your behind a firewall and having to use port forewarding. Although the Java clients allow others to connect to an Xgrid controller…actually getting a windows unit to render is a different story.
I use Xgrid at home with 4 computers and it does a fair enough job. Now, getting it to where others can connect and use it has been a bit tricky. Again Mac users don’t seem to have any problems, everyone else does…
We provide render farms with DrQueue pre-installed on a linux cluster, drqueue will install on windows and you can submit the jobs from windows to your cluster. It can fulfull all your requirements and more, check out the official website drqueue.org for a discussion of the windows port.
An alternative is you could use a virtualisation program like VmWare to run Linux under windows use a samba share to share your files and run drqueue on linux, then you could use ffmpeg to put the animations together from the individual images.