We could just focus on science to find a good starting point, at least thats what I do.
Regarding localisation of things, there is interaural time difference ITD, interaural level difference ILD, there is Blauerts bands to suggest up down/front behind.
Level differences can of course be frequency dependent, lets call that interaural spectral difference ISD.
Then there is Cremers law of the first wavefront…
You can utilize all that, you can create phantom sources, you can freely move them around in stereo field by altering certain parameters.
Starting point would be: Two identical channels of content x. Hard pan one left, one right. Equal levels. Start altering levels, delay (<2 ms) to each other, filter cutoff. (ILD, ITD, ISD). Use a lowcut to suggest distance, try blauerts bands, combine all that in one way or another.
That is pretty much what there is to do with two ears/two sound sources. Stereo is stereo, thats about it, two ears. You can start playing with phase and stereo separation of course…there is lots of things you can do, but getting the basics down might be a good thing.
Headphones are different anyway because of in head localisation.
BUT if you do things knowing what youre doing, you can maybe achieve the effect you like.
I played around with stuff like that in this track:
Of course there is stereo delay itself, but there are several channels with the same content being delayed and altered in frequency spectrum via eq and cutoff to broaden stereo image and depth of things, even stereodelay is doubled and altered in its delay from channel x to y and their respective frequency spectrum.
I tried to achieve depth and some huuuuge comfy space to get lost in if you will.
IIRC I might even have had valhalla reverb moving around as a phantom source…
tl;dr: To begin with, create your own phantom sources and play around with them, do not use the real source alone, combine them maybe.
It’s not just about where to place in the stereo field, its also about the room you create itselft…