Shrad wrote ...

breakout point + aluminium tape over the toroid to flatten the surface, for what I can see

Blitzorn wrote ...

That is a real nice-looking first coil - similar to mine, a bit tidier of a build though (mine was quite rushed). Congrats, and beautiful work! I'm relatively new at this so don't take me on my word - but since nobody else is responding, I'll throw in what comes to mind...

1. It may help your output a little to add a power factor correction capacitor in parallel with your NST's primary winding (on the mains side). This rectifies some of the power losses due to the desync of voltage and current phase from high inductance in the NST. Make sure to use a run-type capacitor. I can't tell from the pictures if you have already included one in your design.

2. Moving away from saltwater caps is probably one of the bigger changes to improve power and performance - an appropriate MMC will simply perform better than the beer bottles. You already seem to know that, but it's worth stressing the lossiness of glass. You'll thank yourself later for cutting your capacitor's mass down by a factor of ten or twenty as well. My first coil used a gargantuan saltwater cap and it suffered from low output.

3. Check your wiring. Looks like you used good quality wires, so any upgrade at this point is probably not going to make a tangible difference, but HV grade wire is another possible source of losses. Junctions should be soldered to maximize contact and keep the risk of arcing to a minimum. Looks like you did a pretty good job with the junctions too but just saying, bad joins may be lossy. Make sure connections are as short as possible, and that there aren't too many kinks or sharp corners. Minimize interfaces between different kinds of conductors in your wiring (tiny effect, compared to other stuff, but if you're min-maxing...).

4. Like shrad said, smooth surface and/or a specific breakout point will help to get all your uber-high-voltage charge running down the same streamer - smoothing reduces corona discharge since it reduces the amount of bleed-off, and specific breakout point is essentially the opposite, creating a region where charge may leave the surface more easily. It might sound like these would cancel each other out, but they don't have to. Imagine a bucket full of little holes. When you throw water in it little squirts go out in all directions. Now imagine plugging up all the little holes but one, and making it wider. Throw water in it, and you get a pronounced, directed jet of water. There's your power.

5. Tune it some more! If you were very thorough with your tuning then of course, this isn't a big issue. I don't know if you were able to test the capacitance of your primary array with a good multimeter, but those saltwater buggers are hard to estimate, and sometimes when you run them a little bit their performance changes. Software is great, but ultimately the models are only as good as their fidelity to phenomena, and every real-world situation is unique!

6. Connect your other NST, silly! ;D

7. The rotary spark gap, or even better quenching on your current gap. Sounds to me like your current fan is totally inadequate for proper ventilation, even just duct taping a shopvac hose on it and letting that puppy rip would be a step in the right direction. You really want it to quench as fast as it possibly can.

Hope I was able to help - anyone with more experience or knowledge feel free to overwrite my suggestions but I think they're mostly pertinent!