Georgina,
I know a woman whom all called 'Georgi' since high school, and her husband was an acquaintance also. Both are unpretentious, self-reliant, hard working and conscientious. Back then there was a Pop song 'Georgi Girl' that was light, lively and encouraging of young women to become all that they were capable being and wanting. So If I slip and address you in a presumed familiarity, treat it as a form of endearment.
My thanks for the references are in an aside to Akinbo, but beyond that as to your criteria which I noted has an IFF quality to it, and as might be relevant to your ponderings of SR; complex analysis is called *complex* because it employs imaginary (only by definition in mathspeak) numbers but can be applied to your illustration of the observer dependence in relation to the aspect of the cup. I prefer 'illustration' to analogy because correctly an analogy means a 1 to 1 correspondence, an analog. Here goes;
Say you cut an orange in half, three times at mutual right angles. Now you have 8 equal segments of a sphere. Stuck back together, if you want to measure from one octant into another octant, you have to correlate the point in each octant to its respective x,y,z axes. But there is nothing upside-down and backwards to the temporal spatial relationship between those two points like you would find if the line of sight went through a lens at midpoint. So the non-commutative algebra of Quaternions was devised to say if you designate your start of measure in octant 'A' then in the other octant what would be displaced from the x axis is transformed to a displacement off one of the others depending on which direction you rotate your measurement scheme, and if its in an octant that your scheme designates as in the opposite hemisphere, then it operates as a negative (imaginary number) and that prescribed axis that would otherwise be x is now z but = 'i'. Yeah, I know. There's got to be a better way. Especially if your proportion of measurement is a quotient because while you can multiply by zero and only need 4 transforms, you can't divide by zero and so need 8 transforms known as Octonions or 'division rings'. But! it's okay! because all you have to do is pick your intial point and that sets which ring, or 4x4 matrix chart, you use throughout all computations. Those axis transforms correlate to an * if and only if * criteria from the initial point of observation.
But so do your definitions of qualitative designated reality. Light intersecting light suffers no interference and doesn't slow it down, but the refractive index of a medium through which it may or may not pass, does. What emerges from your criteria is (at first blush) consistent with the physics of 'just because we don't observe the light bouncing off it due to our distance or position, doesn't mean it's not physically part of the visible universe'.
Glass is silicon and is transparent to light, but the spark plug wires in your car are actually not wire at all, but long strands of silicon insulated by spongy silicon. And so opague that the high voltage electromagnetic pulse carried by the strand is shielded sufficiently by the molecular arrangement in the insulation sheath that in all likelihood the low millivoltage from the engine control computer that operates the 'step' motor of the idle air control valve is carried in a wiring harness that rubs up against the distributor or plug wires. IFF. :-) jrc