I've never done this before, so I don't know the exact actions you'd use, but in terms of logic I'd go about this way:
1. Calculate the distance between the two players every frame, output to some variable. I think you would use a raycast for this.
2. Use that variable to control the zoom level of the camera (and position if needed) - you may need some math actions to get it working right. You'll need to experiment with numbers to get the zoom level just right. Maybe if the players are 15 units apart, you find that the camera looks best zoomed 150 units away. This would mean you take the variable, multiply it by 10 every fame, store it in a new variable, and then set the zoom distance of the camera to the new multiplied variable every frame.
3. If the camera moves horizontally as well, you would set its position to the center of the raycast line drawn between the two two players ever frame.
Sorry I don't know exactly what to do, but hopefully that points you in the right direction!