I’ve been working with — and loving — Deadline for years now. And ever since version 5, when they revamped the whole workflows and extended it to multiple platforms, I’ve loved it even more. Now, with version 6.1, the new features make it even more attractive.
The largest step in the latest, at least for me, is the job dependency view. Well, let’s back up for a second to talk about dependencies. Deadline will now wait to start jobs if something it needs is missing. So, it will hold off until a cache is finished writing, or a texture is created or rendered, or if a Python script returns true. Now, the job dependency view shows all of these connections in a node-based window – just like Nuke, Houdini, Thinking Particles, Magma or any other powerful piece of software. It only makes sense to view your render submissions in this way. And, it’s not just viewing. You are able to connect, disconnect and manipulate the dependencies within the graph, making management much easier to follow … especially if you have multiple jobs waiting for similar assets.
Additionally, if you are working with 3ds Max, you can take advantage of advanced multiregion rendering and auto-assembly of single images rendered across the farm. Not `only do you have arbitrary, savable areas, which can be manually set, or shrink wrapped to a object, but if you want to render across an animation, that region can be set to track the object you wish to isolate. Pretty darn handy — and I hear Maya will be coming soon for this. But in the meantime, Deadline still supports distributed bucket rendering for Maya V-Ray as will as Arnold region rendering.
For those of you who like to get under the hood, you can use the Native Python API to talk to Deadline directly without having to call the Deadline command app — seriously beneficial for those who need to tailor Deadline to their own pipeline.
As always, it’s a pretty UI with an intuitive approach, supported by a robust foundation. Some hardcore Linux and OSX guys may take umbrage with the bits of Windows NET legacy methodology but for the majority of us out there it just works.