I’m confused again (or should I say still). I’ve tried backburner ones before without much success a long time ago. It seems a lot of people are using BB instead of just DR. What are the pros and cons of both?
I don’t know why you would want to use BB if you can use DR (bear in mind I know squat about BB). OK so sometimes when I’m using DR, the renderslaves kick in a bit late, when my workstation already almost rendered half of the image by itself! I experienced this with high res stills as well as 768X576 animation frames. Sometimes my pc renders the whole animation frame before the slaves even think of starting to render. (slaves is same spec as workstation more or less) and (various render times from 5 minutes to 20 minutes a frame). Will BB fix this problem for me?
What is the best way to render animation? 10 PC’s each sharing one frame or each PC doing its own frame? (if BB can do such a thing)
The reason to use BB is for queue management. If you are rendering an image overnight, and one or two others on your team are doing the same, how do you queue them otherwise? Before 1.5, you had to use BB render strips if you wanted to use all your nodes.
The reason DR is faster than BB render strips is because render strips always overlap by about 50 % regardless of specified overlap.
We can now send DR via BB, so there is no reason to use render strips anymore.
Usually your nodes kick in late because they are still rendering their LC pass. LC is not yet distributed.
@Clifton
Thanks that makes sense, and I geuss it will make more sense if I actually try it with the next animation job I get. Is sending DR to BB usefull for 1 still also or mainly for animation?
Like RErender said. DR is faster for stills, but for animations its not important. Animations are naturally distributed because of the nature of the exercise. The only time it might seem helpful is if you have really slow nodes and at the end of the sequence you are waiting too long for you slow nodes to finish. But if thats the case, you should just retire those tired nodes…
It seems like a tough decision to make, but I retire a node after 3 years (and donate it to IT for a server or whatever), because the cost/speed ratio just doesn’t make sense.
In BB, each machine will do a frame, unless you specify otherwise. If you use strip rendering or DR via BB, then the nodes share a frame, which is good for still renderings.
Okay. You can do DR by itself. You can do BB by itself, either with each node rendering a frame, or all nodes contributing to a frame using render strips. You can now, in 1.5 sp1, do DR with BB.
Unless you have only one computer, you want to use BB or DR so that all your nodes can contribute to your job. DR is faster than BB render strips for still renderings. If you have more than one person in your office doing DR, it is helpful to queue your DR jobs in BB, so there is no conflict.
For animations, you don’t need to use DR. Just BB.
You should probably practice using both DR and BB and these things will become very obvious.
thanks I usually use dr for stills and animations. Never tried bb.
So once I’ve tried bb, what is best to do animations, strip rendering with bb or each pc doing a whole frame by itself with bb?
How do you queue a job without BB? Do you use a different render manager?
Render strips are for stills. It allows each computer to contribute to a frame. It has about 40% overlap, so it will always be slower than DR. There is no need to use render strips any more, so don’t worry about it.
For animations, just send to BB. Do not use render strips or DR.
How do you do animations with DR? its never worked for me, the dr machines will do 10 frames and then die off..
Look, its simple, stills just use DR, first calc your LC, with DR off, stop render when LC is done, turn on DR with the LC rendering from file and the nodes should appear very quick, depending on your scene size and network speeds.
Animations,
Calc your LC and Irr map on one machine, when these are complete send your job to net render rendering from the saved LC and IRR map and then all your machines will render all your frames from the saved maps.
Or you could split your IRR maps over the network as ReRender mentioned.
so if you have 1000 frames and 10 machines and you are rendering every 10th frame you can send the first 10 frames to machine one (MAKE SURE YOU HAVE TICKED DO NOT RENDER FINAL IMAGE in the globals) and tell it to save IRR map and call it machine one, and just keep doing that for all parts of the irr map, once the machines have finished, use vray irradiance viewer to merge all the 10 parts of the irr map, once it is merged re save it and call it what you like (final map). Then send the job for rendering to BB rendering from the final map and LC.