Elliott Balsley wrote:Within a node, there are many adjustments possible (primaries, log grade, all the types of curves, blur, etc.), and the order of these things can make a difference. When in doubt, I can add another node, but sometimes it's easier to do 2 things in one node. The only thing mentioned in the manual is that node LUTs are applied last. I'm guessing all other operations happen in parallel, but I'd like to know for sure.
I think it's a mistake to try to do too much within one Node. The main reason is, you can get lost very quickly in a complex correction -- say, with 6 or more simultaneous nodes. (And I've had rare situations with 15-20 nodes within a lot of cuts, but I'd say 3 or 4 modes is a more normal average.)
I think using the first node for an overall correction is reasonable, then secondaries in the next node, windows & keys in the one after that, and so on, works pretty well. If you have to use a parallel / layer / mix node, you'd generally attach that to the first node. If I had to do a reposition, I'd generally do that in the last node. Opinions differ on whether to use project LUTs or node LUTs and where to position them.
Warren Eagles and Alexis Von Hurkmann have very good strategies on how and why to use specific node structures in their respective tutorials, and I tend to agree with most of what they say. One important point is that if somebody else winds up working on the project, they need to understand your basic node structure; by the same token, if the project ends, then somehow comes back 6 months later for fixes, you might find yourself at a loss to remember exactly what you did within each node. If each step is broken down in one adjustment per node, it's a lot easier to figure out what went on so you can trim what needs to be fixed.