I am using Add "Image/Sequence" to load a set of JPGS files to the Sequencer. Next I run a "Render Animation" to output AVI file. That looks like just joining static 2D pictures in one movie clip.
Is that "Render Animation" call use internally the same/similar mechanism as e.g. 3D "BMW demo" rendering? Please explain differences.
Apologies for the delayed answer.
The Render Animation option is a single call that will execute one of several options.
Blender will first check if anything exists in the sequencer. If there is something there, it renders that, and nothing else. If not, it checks the compositor for nodes that don't require the 3D scene to be rendered, such as if you have a movie clip that has some effects applied there. Finally, if there is no non-3D data to render, Blender will render the 3D scene.
So the order of operations is this:
Under the Post Processing heading of the Output Properties tab, the sequencer and compositor can be overridden, so that Blender will skip one or both of them.
The sequencer is entirely 2D, as is the compositor. I believe all 3 are based on different underlying code. So the 3D view is entirely separate from the compositor and the sequencer. In order to get any information from the 3D view to the sequencer, you must render the scene to get a 2D image that gets stored internally or externally, then load that into the sequencer.
So the sequencer and compositor behave similar to other video editors or compositors, and aren't connected to the 3D scene any more than another software would be (that I am aware of).
