I'm gonna re-post something I did in your previous thread
@Marchesk, but will add to it.
This is one of those philosophical issues that I don't see the point in, but I don't think this is because of quietism, rather because I don't really see what's at stake in the issue.
What is a theory of perception? Presumably it's a way of assigning a description to the following kind of event: X perceives Y and set of properties and relations P(Y) influenced or deriving from the set of properties and relations P(X,Y). As an example.
I perceive a cup on my table, it is plain white and filled with coffee.
I (X) perceive a cup ( Y ) on my table ('on my table' is a relation between the cup and the table, a member of P(Y) ), it is plain white (a property of the cup, a member of P(Y)) and filled with coffee (being filled with coffee is another member of P(Y)).
I think any direct realist and any indirect realist would agree that indeed I do see a cup on my table, and that it is plain white and filled with coffee. What matters between them is how to analyse 'I see' in terms of the subject: me, X; the object: Y, the cup. Specifically, what matters are the properties of the relation 'sees' between X and Y. How does it arise? What does it mean for me to see X? What are the relations between the seen object and the object? (representational sense data or identity for indirect/direct examples). Answering these questions gives elements of P(X,Y).
What does a theory that uses sense data or identity as fundamental entities in P(X,Y) achieve? At best a generalized description of what it means to be a sensory object - an element of our perceptual world. Whether it is constituted by sense data or populated by the objects themselves doesn't gives us any information about why the relation between the seen object and the object obtains. We 'see' sense data, we 'see' objects, so what? How can someone learn anything about vision or perception in general - how it works - just by attempting to describe the conditions of access to the sensory object?
Let's take a couple of, very abridged examples, of how to learn something about perception philosophically. In the transcendental aesthetic in Kant's Critique of Pure Reason. It proceeds, abridging a lot, by attempting to found the perception of objects in terms of a mental application of necessary qualities of a sensory manifold. Stuff has spatial extent, stuff persists in time. So then we're left, even if you do not agree with the specific conclusions Kant has, that a perceiving subject conditions the observed objects in some way. Hurrah, we've learned something. How do we condition the objects? Through the application of these constraints to the sensory manifold. What about the 'real' relationship with the object? ... Well, whether Kant's noumenon is given a positive (there's really real stuff underneath our perception) or negative (the noumenon is the name of a conceptual delimiter between the intelligible and the unintelligible), no longer tells us anything about perception, rather about how perception relates to knowing. The latter is still debated within Kant scholarship, the former is well established science at this point.
Another is Husserl, with his idea of 'bracketing','reduction' or 'epoché'. This means, roughly, forgetting the objectivity or veridicality of our experiences and instead attempt to deal with their internal structures and webs of meaning. One way he proceeds is by using his imagination to vary perceived objects in order to filter out their non-necessary properties for being those objects, and thus attempts to derive internal structures to perceptual acts. Great, we can learn something through these descriptions about how we intuit objects and ascertain what they count as or are identified as. Whether the object is 'really there' or 'just a sense object' doesn't matter for the purposes of (transcendental phenomenology) description of perceptual events. If you asked Husserl whether his phenomenology cared about the real existence of objects vs their status as perceptual ideals, he'd probably say something like 'no, I don't want to repeat the errors my method was meant to avoid'.
The debate between direct and indirect realism(s) proceeds after granting people a perceptual world. The next step is for some reason thinking 'how perception works' can be answered through analysis of our condition of access to the already granted perceptual world. How perception works is a question on the level of the manifestation of the perceptual world, not on its conditions of possibility. Is it then surprising that absent from this kind of analysis is any analysis of the performativity in the perceptual event, and this changes the kind of questions that would be asked of a perceptual theory. A contrastive question between direct and indirect realism, of specific sorts, might be 'do I see the cup of coffee or do I see a representational sense datum of the object?', an analysis inspired by the performativity of the perceptual act (it's a verb, c'mooooon) might ask "how is it that I see the coffee cup? what perceptual structures allow me to see the coffee cup?". It changes debates from, ultimately, a semantic theory of perceptual verbs or their conditions of possibility to 'what makes us perceive how we perceive and how do we perceive?'
In terms of the original formulation, the debate between indirect and direct realism does not attempt to flesh out P(X,Y), it instead attempts to look at the conditions for the possibility of P(X,Y)
while forgetting that it does this. Is it any wonder that this thread and the previous one are full of unsubstantial semantic dispute, and that any 'evidence' for direct or indirect realism based on the real properties of perception can be interpreted favorably or explained away...
If we already grant the 'world of perception' to a person, what remains is to give an account of its formation and stability rather than our conditions of access to it.