When people comment on using a 45º perspective, they mean to draw the sprites from that perspective from scratch, not to alter the angle as you'd do with a 3d model.
Take a look at older RPG games, or shooters like Chaos Engine, even when the gameplay is Top-Down, the sprites are given perspective to differentiate what they are:
On the other hand, you can keep the pure top-down perspective, and rather than graphically, have players infer what objects are functionally
A tree, for example, would allow players to walk under the branches, so even when they might look like bushes from above, players would soon learn to recognize them.
An idea would be to compose trees from two sprites: The branches and the trunk.
When the player is below the branches, fade out (with transparency) the sprite (or just fade out a small section where your player is), but leave the tree trunk sprite below (with actual solid collision) visible.
That way not only will it be evident what the bushy thing actually is
, but it can lead to interesting gameplay effects.