This is just a kneejerk reaction from your post, intended only as food for thought, but I wonder if your problem might partly be that you're actually trying to overanalyze the texture. Only, it sounds like you're trying to force yourself to observe just as much half way through a texture as you did at the beginning, then you cut corners when that turns out to be incredibly slow and frustrating. Maybe I'm doing it wrong and that is in fact correct, but the way I see it, a lot of the observation and decision making should be front loaded, and it's not only acceptable, but desirable to shift to focusing more on execution as you progress.

You need to keep looking at the reference so that you don't fall back on repetitive simplifications, but you can be very specific about what you're looking for. If you know what you're trying to add to your texture you can look for a specific example of that thing then immediately go back to your drawing to fit it in. It's never going to be as quick as just drawing what you think it should be, but you can still achieve a sense of momentum in your workflow. On the other hand, if you treat each pause to look at the reference as a break to reanalyze it, you're really just second guessing yourself and any progress will be painstaking. That doesn't mean it's bad to look at the bigger picture and reassess your approach. Just be clear when that is what you're doing or when you're just looking for something specific.

That is how I approach it at least. Maybe it's actually a terrible approach though.