Compositing Multiple Scene Layers

Okay, I’m not a complete Blender virgin, but I haven’t really tackled the compisiting aspects it has yet. This is what I have going.

I have a shot with 3 layers I am trying to composite together.

Layer 1: Background
Layer 2: Fourground Object
Layer 3: Lens Flare

I use nodes to color correct, adjust gamma, set a blur, etc. to get all three layers to blend, and so I render out the three parts into three frames each on a seperate render layer. When I composite the still frames together it looks great, everything is how I want it.

The problem comes whenever I try to animate the camera. As far as I know, you can only render from one camera a a time, so to get all three layers together in the same shot for the camera movement, the three pieces have to be placed on the same layer and the whole compisting node setup gets shot to hell.

I know I could render out the three layers seperatly, recombine them, and then render out the whole scene, but I know there has to be a a way to render all three at once, and keep their individual node setups unique to each one.

Any ideas?

Create new empty scenes and link them via the button highlighted on the output tab, then add new render layers for each scene and, on the highlighted arrows on the render layers themselves, select the scene from which to render. The compositor will render each scene separately then combine them and their associated node chains as a post render operation.


I’m a little confused as to why such a solution is needed.This sounds like fairly straight forward compositing?

Fairly sure I’m missing something, but I don’t see what.

I think RamboBaby may be unclear about your setup because I don’t think separate Scenes are needed. You should be able to do what (I think) you want with just Render Layers. Here’s a Render Layers/Compositor node setup I used to get some blur and color modulation effects for BG & FG subjects while leaving the midground unaffected, all in one Scene (“KataSet”):

I broke the scene into 4 Render Layers, each with its own set of Layers specified (see panes at upper right). These were then piped to various “effects” nodes like Blur and eventually recombined using AlphaOver for the final Composite node. One camera (static in this scene but in others using a similar setup it’s moving) will render all layers. You can use the Layer option for lights if you need different lighting in each Layer, just be sure to include the lighting layers somewhere in the Render Layers.

In this case the Layer containing the Camera is an active Layer only in the “Setting” and “Figure” Render Layers, but still functions for the other two as well. I think you’d have to have the Camera Layer as a member of at least one Render Layer, especially if it’s a moving camera.

BTW, as long as you have the Camera on an active (selected) Layer, camera motion will be followed – it doesn’t have to share a Layer with any other parts of the scene. I use this to set up “banks” of cameras that I can easily switch to and render different viewpoints of the same scene.

If you need two cameras active at the same time then you will need to go the multiple Scenes route RamboBaby describes. Also note that you can “share” cameras between Scenes by Linking them.

Here’s a short clip of the effect in motion – note that this was an earlier render, so her hair is in place, but not the costume top. The hair takes forever to render so I left it out of the demo still:

Render Layers Composite demo clip (Vimeo)

Sorry, I was tired and didn’t read your post very well.

If your BG is only a sky, then leave only that button enabled for that render layer. Do the same for the lens flare with the Halo button. If your BG has objects in it then you’ll need to set up camera clipping to occlude them. Then, following the technique I mentioned above with multiple scenes you’ll need to link your camerafrom the first scene to the second (select camera>CTRL+L key>to scene, goto the second scene, reselect the camera—it’s not active at this point—and U key>make single user>object). This results in a camera that shares an ipo and identical orientation with the first camera but you can now occlude the foreground objects via camera clipping.

Sometimes you’ll need to work like this to keep Blender from crashing due to memory constraints and render size. You build out the entire scene, link the scene as a set via the output tab and select renderlayers from multiple scenes linked this way to be composited in a final scene. What happens is Blender doesn’t have to include invisible objects per scene (those on layers not set as “visible” in the 3D window) then after the scene is rendered, all memory in use by that scene gets freed and the new scene starts fresh.