An ambitious goal! If you create an accurate model of the flight of a bird, it will be a scientific study worthy of the Nobel Prize:)
Again, the current version of flight simulation is very good. I said it would be great to improve it. But in reality, it's hard for me to even identify what needs to be improved.
The first thing that catches my eye is the bird's equal ability to fly up and down. Without changing the mode of flapping wings, the bird retains speed when the angle of inclination of the trajectory changes. It is not difficult to bring this feature into line with real physics. There are at least three possible solutions. The bird can vary the intensity of the wings, maintaining a constant speed at any slope of the trajectory. This option is not suitable for too large and too small angle values. The bird can operate its wings in a constant mode, but with the implementation of different steady-state speeds at different angles. The third option is a combination of the first two.
But it seems to me that from the point of view of the most ordinary player, other subtleties will be more noticeable. The behavior of a real bird is characterized by a large number of modes of operation of the wing. It's not just the frequency and amplitude of the strokes. It seems to me that a very noticeable feature of the wing of a real bird is the variable shape of the flaps. But this problem is much more complex than the influence of the gravity factor.