Compositor UI Logic?

I’m asking this question to help me better understand the way things work in the Blender Compositor.
Am I wrong or if one element is to be put over another element, then it actually get’s plugged into the bottom input of the node, and the element that is supposed to be under gets plugged into the top input of the node?
Is this for real?
Maybe there’s a good reason, and if someone can explain it, then it might help me or other’s new the Blender Compositor out.
What might be nice is a flip switch. So instead of having to disconnect and reconnect, we can just flip the switch the order gets flipped.
Thanks in advance.

Don’t think of it like layering, you aren’t layering. :wink:

The mix/math/alpha-over nodes all follow a convention of top input = starting image, bottom input = image that operates on starting image. For some other modes, like subtract, this is pretty plain and simple. You are subtracting the second input (bottom) from the starting input (top). When you want a normal blend mode, you set the mode to mix (aka over), it’s second input over starting input. Multiply is second input * starting input.

Hmmmmm… any UI experts out there?
This kinda smacks of the difference between programmer think and artist think. I guess it’s a conflict all software developers need to contend with. Where what goes on under the hood vs. what the user holds in their hands.
Or maybe it’s because of my Houdini past where over visually is over and under visually is under so it makes sense in the brain. I know no one who when composting is thinking about the math behind the node. Not unless they really need to.

I think that it may be a math metaphor, imagine a vertical add function where you start with the base value then below you begin adding extra values.


Or like me just remember that Blender is backwards. Top goes on bottom. (mantra)

Hahaha actually thats what I do too…

(Shrug …) It’s an arbitrary decision, and the programmers had to do something that at-least smacked of being consistent. The convention has to be applied across all sorts of nodes, not simply those who have a Z-depth front/back human analogy.

Maybe someday there could be more visual cues – pop-up notes when you hover, different shapes (some of us don’t have great color vision), and so on. But that could be a fair bit of work with moderate payoff.