You know how awesome Compound Blur is? I want that power everywhere.
That is, it would be extremely powerful to be able to switch a parameter to be "driven by" another layer.
You would set your minimum and maximum values and pick the layer it's pointed to. Just like how Compound Blur blurs more or less depending on the luminance, this would drive whatever the parameter was by the luminance of the referenced layer. (You could also include the ability to posterize it and soften it.) Oh, and the min and max values would be keyframeable too, of course.
Some use cases:
- Drive the complexity of a Fractal Noise layer by a depth map. So you have big fog up close and denser, more complex fog the further away it is.
- Some very cool, stylized effects by driving the size of the Venetian Blinds effect by footage.
- The size/density of CC Ball Action balls
- Mosiac sizes
- Glow size and intensity being driven by a different layer would be neat
- Stylized directional blurs where the blur angle changes based on the luminance of another layer (or the same one)
- Soooooo many more
There are some third party tools that do things like this and it opens up a ton of creative options. Red Giant's Mir lets you displace a 3D mesh based on another layer's luminance and Form lets you drive all kinds of particle parameters with another layer's values.
I've been thinking about a feature like this for a very long time, but now that Red Giant has released Depth Generator and I've been playing with depth maps, I keep wanting to drive things besides the opacity, blur, or color of a layer with these maps. The more I think about this, the more useful it could be. I'm aware it would likely require some significant refactoring (and some parameters wouldn't make sense for it), but sooooooo many things could benefit from this.
I know you'd have to have the ability to restrict it. For example, particle tools shouldn't change depending on the pixel values of another layer - I mean just think of changing the particle count based on another layer's luminance: it'd be simulating 255 times in an 8-bit project instead of 1!
But this could unlock so much procedural coolness.