Some notes: Your calculation of the launch azimuth was correct -- but only because you were sitting on the equator when you made it. Due to the way in which the autopilot gizmo works, there is *no* constant heading you can select that will produce a particular inclination. The issue is that inclination is defined in an _inertial_ frame of reference, but the autopilot gizmo uses a non-inertial reference frame. A very visual way to see this is to make the autopilot gizmo visible while you are in orbit with a non-zero inclination and warch the green (prograde) arrow. It doesn't point at a constant heading, thus there is no constant heading during ascent that corresponds with a particular inclination. This is doable with Vizzy code, of course. As far as manually controlling inclination during ascent, the rule are: 1) If the craft's current heading (the direction it is thrusting in) is further away from due East (or West, for retrograde launches) than the *orbital* prograde velocity then inclination will increase (in proportion to how large the difference is between velocity prograde and the current attitude and the current TWR). 2) If the current attitude matches the velocity prograde direction, inclination shouldn't change at all, and _will_ change in small amounts either positive or negative. The problem is that the autopilot, while reasonably accurate, isn't 100% accurate, and the errors tend to stack up as time passes. 3) If the current attitude is closer to due East than the orbital prograde direction then inclination will be reduced as time passes. Finally, the only place to see inclination without using Vizzy is on the map screen. If you have a contact that specifies a required inclination, you can see it in the contracts window as you discovered in the video.
Thanks! That is quite a nice explanation! I kind of managed to get it right/better in the following launches by just waiting till the space port orbit intersects with the launch site and then launching at 60° heading - works quite nice.