Efficient workflow in composite?

Hi folks,

I’m really new to compositing in Blender so it could be I’m about to say some stuff that makes no sense whatsover!:

As i understand it, if you want to bring an object into it’s own render layer in composite, you need to do the following steps.

Firstly, put all object that don’t need to have their own render layers into the same collection.

Next, for every object that needs to be in it’s own render layer:

Create a collection and put that object inside it.

Create view layer.
Right click on all collections that you DON’T want to see in a render layer, then: > view layer > disable from view layer.

Now you can create render layers in composite, then choose whichever view layer you want to see in that render layer.

SURELY it cant be like this? There must be a better or more efficient way? The more objects you need to isolate in a render layer, it becomes exponentially more work it is to set this up.

PLEASE, any words of wisdom would be very welcome, and i’m totally open to the possibility that i’ve got this all wrong!

You rarely need a unique view layer for per object adjustment. Instead, use cryptomatte.

View layers are most suited to large separation like foreground/background, isolating ‘hero’ objects, or special situations where you need otherwise obscured pixels.

1 Like


Crytomatte?! Never heard of it, will look into that.

Just in case, for any young players stumbling into this thread:

There’s a quicker way to get objects into their own view layer:

First you need to put an object inside it’s own collection, then:
Click the “add view layer” button (top right in the outliner) and choose :blank.
Now you have a view layer with everything disabled. Now you can select the object you want isolated and press alt + E …
… note that even if the camera is disabled, the render will still take place from the camera’s viewpoint. Not true with lighting, lights need to be enabled or won’t render. And another little time saver is: to name the new layer the same as the object, just drag the object onto the view layer name slot!

Yes, Criptomattes are really a good alternative if you only need to do tweaks to colors on specific objects.
I tend to try to render everything in one layer to avoid the needs to recombine everything in comp. There are always some issues , lets say you want to isolate each characters, then some characters gives an object to another, in which layer will you put the object ?
Having everything in layer / passes allow to do some edits on heavy renders, but blender isn’t really suited for that unless you export everything to some multilayer exrs.

My advice is try to avoid doing renderlayers unless you can’t do otherwise or if you have very specific needs. But all that depends a lot on the kind of projects you’re working on.

Now, folks, "here’s where I think someone needs to put together a video tutorial. (See which existing Blender tutorial-site would help you with it and then publish it.) Because, while you are talking about issues that I happen to understand, a whole lot of people who are reading “just this text” would have no idea what you’re talking about – they cannot “see it.” It should be a straightforward “tute” that shows what you’re shooting for, how you need to do it, and various ways that you can get it wrong. (Because, there sure are a lot of ways to get it wrong.)

1 Like