:lol "You do not need to be a C programmer".

You need to know the C syntax, you need to know how to compile the thing, you need to know how to understand errors and fix them.

I appreciate the effort, but so far it looks like "I cobbled some C code and eventually made it to work", and doesn't explain why or how things work. The only thing I learned from your example is that there is potentially a way to have one input and redirect it to several node inputs (on same node or on other nodes).

But looking at implementing this effect as a GEGL op, I'm still left with many questions: how do I apply vivid-light? How do I "scale" an input signal before I give it to a node input, and more generally how do I algebraically combine several inputs to make one input? how do I compile the code separately from the whole of GEGL (and other folks would also wonder where they put the compiled result without polluting the "official" GEGL code(*))?

(*) On Linux it's in ~/.local/share/gegl-0.4/plug-ins/ it seems

I just don't know how to make a proper guide. It all had to do with my luck of learning GEGL Syntax so I could use the graph to automate things back in mid 2021 to mid 2022. After that I started studying GEGL's source code and got lucky when it came to rearranging the nodes to make third party filters. I could record myself making a third party filter just to show everyone what I do.