Albatroon, this is a cool and very interesting project. Some thoughts:
The speed with which the solenoid can move is only one factor in how fast the water flow will start or stop. At these fast speeds every little thing counts.
What pressure are you using for the water? What solenoid valves are you using? Are they designed for fast response? Because there is only so much you can do with the electronics and you need to look also at the electromechanical and fluids aspects. I see there are "fast" valves that have a response time of about 20 mS and you can get super-fast valves that will respond in 2 mS.
You should try to avoid long tubes that add inertia to the fluid. The best design would be to have the valve right next to the jet orifice and connected, not to a long pipe but to a common manifold water header for all the valves.
The switching transistor should be oversized in current and voltage specs. While the solenoid armature is moving the inductance decreases a lot and the current increases a lot. And the switching voltage transients should be limited with zener diodes or other means.
Also, heat dissipation and operating temperature should be taken into account.
Increasing the voltage and therefore the current to the solenoid might make it switch faster, because it will apply more force to the valve.
Putting the solenoids, as near to the jets as possible is good advice. It will also help if the tube after the solenoids is as rigid as possible.
I'll try this software, Looks simple!
What Zener should I use ? Any one would work ? Should I get a big one, Like 1w zener ? or anything will work ?
The power rating of the zener depends on the operating frequency and the inductance of the solenoid.
Thinking about this circuit in more detail, how are you driving it? I know you mentioned the 595: is that the old TTL type i.e. the 74(LS)595 or 74HC(T)595? If you're using the old TTL type, then it won't have enough drive to Q1 and Q2. I'd recommend the 74AC(T)595, which should be able to drive the circuit directly.
Other possibilities are: Darlington pairs for Q1 and Q2, additional driver transistors, or MOSFETs which would require an additional driver IC.
One thing I didn't consider before is the power dissipation in Q3, which could be too high if it's switched frequently enough. Dissipating some of the power in a series resistor would help.