I once startled a cab driver overseas because he didn't expect me to be American. I had a horrible accent in the foreign language, but what little I spoke to him were words I used often and so were the least objectionable. I had dark hair and it was late in the evening. He gave me no more than a glance.
When I accidentally revealed myself as an American (unaware I had been hiding it), he jumped.
A decade later, in the US, I have a similar experience.
Now, I look extremely white. I have no illusions about being able to blend in or some nonsense like that. That isn't what happened in these cases. Rather, I accidentally 'passed' because the drivers in both cases were running on autopilot and made assumptions based on the fact I was overseas or in a predominately non-white neighborhood. Skin tones vary enough and my hair, even with the growing gray, is dark.
It isn't that I blend in, but rather that I didn't immediately ping any sense of wrong. I didn't overly betray expectations and so they saw what they expected rather than what was actually there.
And it makes me wonder how to use that. When crafting a disguise or trying to blend in, what might matter most isn't looking right, but rather not looking wrong. A wrong detail doesn't immediately say someone is out of place, but it does draw a second glance and a second glance would reveal the ruse.
We don't remember everything about the people we see and we build sketches of who we expect to see, but those aren't super detailed either. Based on a few details, location, context, actions, etc, we see what we expect more than we see what is real.
If you see someone dressed a bit smartly standing in front of a lecture hall looking through a pile of papers or a laptop, you assume that person is a professor.
If you know your professor is a light-haired white man with a beard and the guy at the front of the room is a light-haired white man with a beard, you may even assume that the man in the front of the room is your professor until you give him a second glance. The general appearance and context match, you make an assumption, and move on. It is only if you look more closely that you see the truth.
(Others, people who pay attention at the start--possibly because they need to ask a question or they're worried about something related to the class--may notice sooner, but may not do anything because no one else is doing anything and because they have no reason to assume that the man who isn't their professor is doing anything untoward).
It is like the uncanny valley. The closer you try to match what's actual, the more the wrong details scream out. However, if you just approximate, adopt the right attitude, and use context to your advantage, no one has any reason to give a closer look.
What snares attention is something being wrong.
If the guy at the front of the room is wearing sandals that you've never seen your professor in? You're going to give another glance. That snags your attention. But if the guy is wearing generic shoes that aren't completely wrong even if they aren't even close to what your professor actually wears? You'll likely not even notice.
It's like, when you meet a stranger, you don't remember what they look like. You remember enough to distinguish them from the people you already know. Let's say that growing up you've somehow never met a man with red hair and then, in early adulthood, you meet two. One is a bartender and the other is a professor. Neither have beards or immediately noticeable tattoos. They both dress a bit, but not quite like hipsters. The styles are different, but not extremely so. Their body types are also similar. You don't know either man very well. In your head, both are described as 'red haired man.' One is 'red haired male bartender' and the other is 'red haired male professor.' You don't need more to recognize either one.
I will guarantee you that if you met one of the two men in a bookshop, you'd need a moment to determine which red haired man it was. You'd have to put the face into a mental context (bar or classroom) and see which matched.
Once you got to know them better, you'd add more detail to your mental image of them to better recognize them outside their context and to distinguish them from each other and any other red haired men you might meet.
I'm going off on tangents.
The main points are these:
People see what they expect to see.
Something being very wrong draws attention in ways things being not-quite-right do not.
Context shapes what we see.
You can use that to make people think they see something they don't.
This kind of illusion falls apart though once someone has a reason to give a second glance.
But! Even if the illusion falls, as long as they don't have reason to think something is very wrong, they'll likely assume that what's going on is all right.