i have found a free render farm that allow you easily upload your blender file and then the users render your scene. There can use cpu and gpu. When you help to render other projects the priority of your projects are more and more important.
So you can checkout the website and mabye you liked it and you will join in the community. Then a biger community render your projects faster And when you really like the website then you can donate for the website to alive it.
I’ve rendered 1 project on it to test, and have volunteered about 30 hours for it, seems quite decent.
One thing however, because of the credits (which I like) your better off putting some time in BEFORE you need to render, otherwise you may have to wait a bit to get your animation rendered.
What else I like, is that you render your animation first, so I can upload an animation, render it on my 3 PC’s, and have other people render it as well.
Edit: Currently testing, I’ve launched 10 Amazon EC2 instances, using the Sheep-it command line render farm program.
This enables me to add cloud compute nodes that will render my own uploaded projects first, (earn extra credits) and easily combine my PC with extra instances to render my own stuff quicly. Basically using the sheep-it system to manage the distribution and collection of my animation for me.
Good value given 70c/hour AUD gives me 10 c1.xlarge Amazon instances, each having 2x Quad Core E5410 Xeon Processors, for 80 Cores at 2.33 GHz
I want to do something like this in a future. Can you give me some URLs on how to use Amazon EC2 with Sheep-it? A kind of FAQ or something… I’m not familiar with this but it’s interesting. And is it feature full? Can it “suck in” cache files also?
I’m not sure exactly about how well it works with caches, but I do know that it supports uploading .zip files with relative files, so I suppose a cache could work (will test later) except that the overall file limit is 500MB.
Have you much experience with Linux or command lines in general?
The first step is to launch some Amazon instances (if you sign up, the c1.xlarge are 7c/hour spot pricing, with 8 cores) and ssh into them (I use Putty from Windows)
as this downloads the render farm manager (that auto downloads scenes and blender).
The second command is
“java -cp farm-client-3.2.1288.jar com.renderfarm.worker.standalone.FermeDeRendu https://www.sheepit-renderfarm.com MYLOGIN MYPASSWORD”
as this logs in with your username and password and auto renders your own projects first. You can use “nohup <command here> &” to stop the terminal connection killing the render if there is a network disconnection.
As a speed comparison, I’m using the cc2.8xlarge instance, which has 8 cores/16 threads, 60GB ram and over 500GB storage space.
At ~$0.27/hour, it takes 4min 20sec to render what takes my Quad Core i5 750 @ 3GHz ~41min to render
OK, that’s a good new about this limit and zip acceptance!
I’m not familiar with Linux yet.
Don’t want to bother you a lot. Are these Amazon instances well described on Amazon.com? I’ve heard about them 1st time I’ve seen some video from this year Blender Conference.
These command lines looks easy. Thanks for sharing, BTW!
Is this 8c/16 threads speed comparable to something like new NVidia 4** series? Approx.
About 2/3 down the way of this page is a table showing all the instance types, otherwise there is a text description of each instace type on the page as well:
Hmm, I shall need to try see if I can the Cycles Cornell GPU test on one, and see what speeds I get.
The closest comparison I have is that my it 750 Quad Core takes 41 mins on a scene, the 8 (2 x quad core 7c/hour) core c1.xlarge takes 18 mins on that scene, and the cc2.8xlarge (2 x 8 cores x 2 threads = 32 threads) takes about 4 minutes 20s to render the same. Thats at 27c/hour.
Hmm, I just tried it again today, seems to be working fine for me
If you upload your own project first, and then start some EC2 slaves rendering, you should be good to go (and render your animation fast :P)
Unfortunately, if your rendering to get points (to use later) there are some issues (with the standalone in browser client) where some projects (some of those left) are incorrectly configured, and it takes a few minutes (around 3) each time to move onto another project [I think it blacklists projects that dont render for your machine, so after a bit ti should find some good ones. Thats why if you upload your own project first, you’ll start rendering that first
I tried it with the “Amazon Linux AMI” flavour of the “c1.xlarge” instance. I used the command lines (updated versions) and everything went fine until I got the “error render have generated no output file” messages that you mentioned. Except that it never actually gets around to rendering anything. I left the “Amazon Linux AMI” instance on for two hours and nothing rendered, i just constantly got those errors - very frustrating.
Trying now with the “Ubuntu 13.10” flavour of “c1.xlarge” but still no luck. Which flavour did you have sucess with?
edit: this was just an experiment to get some points at this stage, i haven’t tried my own projects yet.
You did a good job explaining It because I managed to get as far as I did lol. On a seperate note I uploaded a small project and started my own computer rendering assuming it would render my projects first but it started rendering someone else’s. That can’t be right can it?
I did it … but when I launch the client there seems to me are rendered frames, but when I check status/connected clients -in my row there is still zero frames and GPU is only in my case “null x4”.
I understood there should be an option to specify GPU card when I launch it.
Hey vklidu, try to wrote the devloper a email again or look at the offical facebook page and ask him. The site is pretty safe, these java warnings are on many sits to. So sorry that i can’t help really because i don’t use OS X.