Now that's a very intelligent question! It goes to the heart of epistemology.
Briefly, information can be transmitted either as symbols or as tokens. A symbol is any piece of transmitted information that is part of a statement that might be false. A token is any piece of information that must be true, but which might be misinterpreted. So all symbols are also tokens, but not all tokens are also symbols.
Examples: The word 'sentence' in the statment "This statement is false" is both a symbol and a token. The photons of light travelling from the computer screen you are reading onto the retinas of your eyes are tokens, but are not symbols.
Statements that are true by definition can only be made using symbols. Tokenized, non-symbolic information is always true, but never occurs in the form of a statement. Evidence is always transmitted using non-symbolic tokens, although it can be referenced using symbols.
The issue with statements that transmit information using symbols is whether or not the sender is lying--either intentionally or otherwise. The issue with empirical evidence transmitted as non-symbolic tokens is whether it is being interpreted correctly. Science is a methodolgy for making correct interpretations of tokenized information, and for improving its interpretive frameworks over time.
Finally, the interpretion of evidence transmitted as (non-symbolic) tokens requires the operation of an intelligent agent, who will necessarily use symbolic reasoning in order to intepret the evidence (which is why the intepretation can be wrong; were the interpretive process based solely on (non-symbolic) tokens, it could never be wrong.)
Bringing this back to the topic at hand, I consider all of reality to be the evidence that is relevant to the question of whether or not there is a deity.
The fundamental atom of reality is the distinction. In order for Nothing to exist, there must be a distinction between Something and Nothing. Once there is any distinction, then there is Something (the distinction.)
The point I'm trying to make here is this: the claim that evidence is the sine qua non of reality has at its heart a very large assumption: that sensory evidence (e.g. the "photons of light" in your example) actually exist and bear some relation to an external reality. In other words there is no way to demonstrate from evidence that sensory "evidence" means anything. For all you or I know, everything we see, hear etc. could all simply be part of a dream, or a hallucination, or some computer simulation. as far as the physical brain is concerned, signal = signal; the brain cells don't care if the signals they are receiving have their origin in some concrete external reality or not. of course, you and I believe that the things we experience via our senses ('evidence") correspond to Real things in the Real world, but we have no way of demonstrating that to be true. We believe in the Real World, but belief is an act of faith. Experience is a subjective process, since it relies on our subjective senses. Therefore, even the "objective" scientific method is ultimately a faith-based system of thought.
So what can one know? We can only know that which we apprehend without recourse to the subjective (and uncertain) physical senses. We do not need eyes to visualize something in our "mind's eye"; we no not need ears to hear ourselves think. Logically, therefore, only those things we apprehend directly can be known to be real. We do not "feel" ourselves to be living; we are ourselves. Placed in a sensory deprivation tank, a man might come to doubt the existence of the sensible world, but he will never doubt his own existence. Therefore, the only thing we can know with 100% certainty is that we ourselves exist. Cogito ergo sum — I think, therefore I am.
The scientific method might seem to be a foolproof way to absolute certainty — but rest assured, it only seems that way. At the heart of the scientific method lies an axiom very much unprovable — that the evidence of our senses corresponds to an external reality in some meaningful way. It may be easy to believe that statement, but by the very "rules" of the scientific method itself, its truth is not demonstrable from evidence, and therefore it is very much an article of faith.