In the article What is Compositing? we explored the idea of marrying pixels to produce magic. Compositing applications are divided into two ways of working:
Ultimately, both of these do the exact same thing. The difference is somewhat akin to the difference between taking a car to your destination and taking a plane to your destination, where both car and plane travel at the same speed!
Ever made lasagna? You basically place layers of different ingredients before it goes into an oven. Let’s say you’ve done twenty layers, and you realize your recipe calls for one more ingredient – which you’ve inconveniently forgotten.
To make matters worse, you have to put this ingredient exactly between the 9th and 10th pasta layer. It’s not a situation even superheroes want to be in.
On the other hand, the ability to put ingredients into layers and throw them into an oven is as easy as cooking gets.
Believe it or not, this is the fundamental difference between layer-based and node-based compositing.
Layer-based compositing places layers of video on top of each other (the composite, or comp). The key advantage of doing things this way is that your software begins to look like your NLE (Editing software). There’s a timeline, with each video on its own track, and so on.
If you haven’t already, please read the article on What is a Digital Image Channel and the Alpha Channel? first. In it, we learnt that we can reach and isolate specific properties of an image, and put that data in its own channel.
You can’t understand node-based compositing without understanding channels.
What is a node? Look at this:
A Node is anything that changes a signal. If a video enters a node, and you tweak a value (color, resolution, whatever), the video signal that leaves it will be a whole new person.
The cool thing about nodes is that you can bring in or take out as many signals as you want. The only limitation to this is the algorithm of the specific node itself. E.g., if I write a small program to split a 1080p video into two 960×1080 parts, the node will look like the diagram shown above.
Whereas layer-based compositing can be like playing Jenga (naughtiness and fun not included), node-based compositing allows you to see all the effects (nodes) added to a video as a graph, called the node graph:
Don’t get scared of the node-graph. It’s actually fun, and quite easy. If you have hundreds of ‘layers’ (actually nodes), it’s easy to find the one you want and make changes to it.
This is the power of a node-based workflow. You forego the NLE-type workflow (timeline and layer-based) for greater control over your nodes and signals.
This is where the power of channels come in. Imagine each arrow carrying as many channels as you want from one node to the next. The second node can choose which channel it wants to manipulate and only perform that operation. Then, it chooses which channels will proceed to the next node, and so on.
I can’t begin to tell you how much power that puts into your hands.
Which is better?
is the ultimate layer-based compositing application. If it didn’t exit, nobody would be talking about layer-based compositing!
Most of the other industry heavyweights, like Nuke, Fusion, Flame, etc., are node-based. When you are working on a shot-by-shot basis, with each shot taking hundreds of nodes to get that perfect finish, nothing beats a node-based workflow.
On the other hand, when you need only a few effects, and you need to see your effects, cuts and all, across many clips, playing back exactly like an NLE, nothing beats a layer-based application like After Effects.
The decision is not whether to use layer-based of node-based applications. The decision is: when to use what.
You have to have both! They are both equally that good.