hjalfi

I really want to see a Gilly Space Program --- it'd be a completely different game, as you're effectively starting from an asteroid base. Build all the big, spidery, low-g ships you want! Then strap SRBs onto them and watch them collapse and implode.

I'll admit I had been trying for Eve, not Duna. I will give Duna a try. Thanks very much. (A really nice-to-have feature on the map is to show projections forwards in time. It already does it for intercepts *anyway*, it'd just be nice to have it more flexible...) For planets in a different orbital plane, I wonder whether boosting into roughly the same orbit as the target planet, doing a plane change manoeuvre, and *then* attempting a rendezvous would be at all feasible? I suspect that would raise the fuel consumption to silly levels as you're effectively doing two complete orbital transfers instead of one, but it's worth a shot. @Ernst: what you're talking about is a bielliptic transfer orbit, IIRC. If you want to get somewhere with very little fuel you raise your apoapsis hugely, do a tiny correction at apoapsis, and circularise again when you've got where you want to go. Wikipedia has an incomprehensible maths-filled page with a good picture: http://en.wikipedia.org/wiki/Bi-elliptic_transfer It's cheaper than a Hohmann but takes forever. I think changing inclinations is supposed to be done at the point when your ship's crossing the line when your orbital plane and the target plane meet. In other words, rotate the map view until you see the two orbits as lines, not ovals; and it's when your ship reaches the point where the two lines cross. Actually picking the optimum angle to do the burn is beyond my expertise but for minor changes a vector due north or south seems to be good enough, but you may need to correct your trajectory afterwards.

So I've got reasonably good at shuttling between Kerbin and Mün/Minmus, without doing any calculations and instead just pointing the nose and throttling up (using the wait-until-the-destination-comes-over-the-horizon technique). It's not very efficient but it works, and my current lander has absurd amounts of fuel so I don't really care about efficiency right now. However, interplanetary transfers are a lot harder. I've read up on it and how to do the standard Hoffman minimum-energy transfer orbit, and, basically, it involves doing the maths first. There's nothing wrong with that, but I feel that given the general seat-of-the-pants KSP style, there ought to be a more Jeb-like way. In particular, a fast high-energy transfer should be relatively easy: get into Kerbol orbit, raise/lower apoapsis until your orbit crosses your destination, tweak until you enter the destination's SOI, wait until you get there and burn like hell to circularise. It would use a huge amount of fuel but it should be doable. Unfortunately I have yet to make this actually work. None of the planets are in the same plane, they tend to move a lot over time so trying to find where a planet will be in the future is hard, and the SOIs are so small on an interplanetary scale that they're really easy to miss. Has anyone had any luck with this approach, and if so, any good tips? (Any suggestions involving protractors or trigonometry will be staked out on the launch site the next time there's an engine test.)

I've found that if I connect the parachute directly to the capsule it's much less likely to break off. Putting a ASAS between the capsule and the parachute generally didn't work. Many kerbals died to bring you this information.

Hmm. That's about where I do my burns, but I need a lot more than five seconds of burn time. My epic atomic interplanetary explorer lander pulls about 1/3 of a gee on full thrust. Is there a better way to do this? ...incidentally, Jeb got his wish: full instrument landing. Bob was flying because he's the only one who can see the radar altimeter and a window at the same time, but this turned out to be pointless as he's flat on his back and the only thing he could see was stars. I would video it and put it on Youtube but it would make for really dull watching. Next time I do this I need to take more care about the landing site. I ended up on a mountainside but was lucky and stayed upright. In the Apollo landers the astronauts stood upright, and could see out --- I should investigate the other capsules and see if any have better sightlines for use as landers.

To get back, get yourself into a roughly equatorial orbit around the Mün, wait until you're travelling in the opposite direction to the Mün's motion, and throttle up. Once you've left the Mün's sphere of influence you should be in a nice elliptical orbit taking you back down to Kerbin. You shouldn't have any trouble with that. I've gotten back from the Mün using nothing but RCS thrusters in the past (including the takeoff).

I had an 'aha!' moment the other day, and suddenly I can do landings. Hell, last night I managed to land on Minmus for the first time ever. *Twice.* The secret is, keep the retrograde marker in the middle of the navball. This means that you are thrusting in precisely the right direction to cancel out both your vertical and horizontal velocity. As your velocity decreases, you'll see this move towards the middle, which means that you're going straight down. Your goal is for your velocity to be as close to zero as you can manage just as you touch the ground, but a good lander design and the heavy duty lander legs can absorb quite a lot --- you just have to be careful not to bounce. If you apply too much thrust and start going up, the retrograde marker will go nuts and vanish. This just means that it's flipped to the other side of the navball because you're moving in the other direction. Don't follow it, just point the vehicle straight up, reduce thrust, and once you start going down again it'll come back. Note that this means that aborting landings is easy! If you start to topple over after landing, throttle up, get off the ground, get the nose pointed straight up and just try again. Also, quicksave is your friend. If you have a flight stick, use it; it's much, much easier that way than using the keyboard. (If you don't have one, you can get one off Amazon for less than the price of KSP, but of course I have no idea if those are any good.) My last lander had a huge RCS tank (this was the one I landed twice on Minmus with). I never used it.

A standard Windows process can only address 2GB of RAM (because the 4GB address space is divided into 2GB for the user process and 2GB for the kernel). So more than about 3GB of physical RAM is wasted. *But* there's a magical Windows option to change the address space split, to give 3GB to the user process and 1GB to the kernel. I've never tried this. Does anyone know if KSP is smart enough to use the extra RAM if it's available or if it just assumes that 2GB is the maximum it can allocate?

I learnt how to fly (and get to the Mün) from watching Winterowl's youtube series, starting here:

Yep. For control surfaces of any kind to be effective, they need to be as far away from the centre of gravity as possible. I typically put an ASAS right at the top, just under the capsule's decoupler, and a bunch of canards right at the bottom. Putting them on the SRBs isn't terribly effective because you're going to jettison them after twenty seconds, so you'll want them on your fuel tanks, but remember that they will rotate and bashing into your fuel tanks is unlikely to make for a good ascent. With configurations like yours I've had good luck placing them on the side of the tanks; this lets me put them further down than if I placed them on the outside. And it makes the rockets much more stylish.

So I dug up my old flight stick. Works very nicely, and makes flying aircraft so much easier (like, feasible). I can land aircraft now! It's a shame that KSP only recognises the X and Y axes, so I don't get the rudder or throttle, but I can live with that. (I gather this is a Unity bug.) Unfortunately when the stick is set to pitch and roll, which is what you want for aircraft flying, it's now pretty much useless for orbital manoeuvring. For that, I want pitch and yaw. I also want bigger dead zones, because as there's no self-righting impulse in space, if the stick is offset slightly my spacecraft start to spin. But in flight, I need them to be as small as possible. And while it's technically possible to fly a plane with rudder and elevators I don't actually know how. And then there's rendezvous, EVA and münwalking, all of which require different joystick maps. I know it's possible to change the joystick mapping by quicksaving, quitting, reloading the game, fiddling with the settings, and then quickloading again; but that's an incredible PITA. Is there any way to change the joystick map on-the-fly? Ideally I like to have several set up and switch between them --- hey, I have buttons on the flight stick to do that! --- but I'll live with what I can get.

That nightclub is totally epic. Were you landing that behemoth of the crew bus on the Mün by hand? How were you manoeuvering it so smoothly? I have terrible trouble landing on the Mün, cancelling out the residual lateral velocity, and have never dared try a targeted landing yet... I'm also interested to see you using the nostril engines. I always thought they were too weak and heavy to be useful.

It's Jebediäa of the Valley of the Wind! ...there's something very odd about the aerodynamics here; you're to be gaining height with no engine. I know that KSP's aerodynamics model is pretty approximate --- what's happening?

Bielliptic transfer orbits FTW! It also occurred to me that one feasible way to get Hally home would have been to have her get out and push --- EVA jetpacks refuel in the command pod, IIRC, so you get unlimited dV that way. But it would be very long and very tedious and it's unquestionably cheating.

The grapple is all stock parts --- just the lightweight landing legs arranged in a circle. And, facepalm about the tanks. As the R-series started life as a lander I wanted the CoG low, which is of course totally not what you want for orbital manoeuvring. (In fact, on reentry the R-series had 1/3 full in the external tanks and the internal tank was untouched. Still, better fuel you have but don't need that fuel you need but don't have, eh?)

Okay, now that's done (it took about four days RL), a few questions: The grapple didn't work. When the legs closed, the capsule would jitter for a few moments and then something would explode and it would go flying. (I absolutely spammed the quicksave doing this.) I've seen people build grapples on Youtube; what am I doing wrong? The alternative rescue mission was to send a capsule and EVA Hally over to it. Unfortunately all the capsules are full. Is there any better way to get a spare seat on a capsule than to EVA on the launchpad and push a kerbal off? Fine scale rendezvous is hellishly hard, not least because I had no idea which way up the vehicle was, and so didn't know which thrust axis did what... the forward hull cam really helped: I spent ages aligning it to match the nav ball. Unfortunately the R-series shook like a collie every time I used the RCS, and the SAS didn't do a great job keeping it steady. It would tend to swing back and forth. Any suggestions on how to improve handling in space? Is it just a matter of more SAS units and more thruster blocks? And if you look at that last picture of the R-series grapple, something very odd is going on...

Hally watches Kerbin approach. Splashdown! ...and while waiting pickup, she watches the dawn she was afraid she might never see again. Now it's Matvis' turn. The faithful R-series, job done, begins to break up in the upper atmosphere. Matvis, quietly pleased, poses with his capsule. By sheer chance he lands a mere 90km from Hally's capsule. He thinks he'll go and visit: hell, Jeb walks that far every morning before breakfast.

There's only one thing for it. Matvis decides to push Hally home. Balancing the L-series on the top of the grapple is hard, but Matvis thinks of Jeb and it's easy. Well, not easy. Still hard. But he does it. They ascend out of Mün orbit into a transfer orbit to Kerbin. Once at Kerbin apoapsis it's time to do the rendevous all over again. This time it's even harder, as the dark L-series capsule is almost invisible against the blackness of space. Who painted them black, anyway? Matvis swears he will find the kerbal responsible and punch them. (The view through the forward camera. That's not the moon, it's the bottom of Hally's capsule. Matvis looking imperturbable, as usual.) They boost again, slowing the L-series capsule into a descent trajectory to Kerbin. Success! Hally is going home. Continued...

Is, as you can see, totally 100% successful. Hally there is looking quietly pleased. Or possibly concerned. It's hard to tell. As the consummate professional that she is, of course, she sees this as an opportunity rather than a problem. In particular, she notes that despite the ascent engine being some way off, she still has an RCS tank and thrusters. Some adept piloting later... And she's safely in Münar orbit. With a periapsis of 4500 metres, she gets a stunning view of the 4000m high mountains go by below. Meanwhile, back home on Kerbin, a major R&D effort is underway. It's time for a rescue mission! And so her fellow kerbonaut Matvis Kerman flies to the rescue... ...in a custom built and totally untested vehicle, the Behemoth 5D R-series. And it's totally not a quick hack of the L-series lander with the legs stuck on top. A very fiddly few hours of exquisitely careful piloting later and they rendezvous, scarily close to the Mün. There's only one problem. The grapple on the top of the R-series doesn't work. Continued...

Alas, I'm a Linux user, and the ATI drivers for Linux are dreck, so NVidia it has to be. But lots of good afvice here --- thanks, everyone!

Can anyone suggest a reasonable budget PC graphics card that works well with KSP? Preferably fanless, if possible... Right now I have a heavily customised GT9500 which is pretty good --- it runs Portal 2 seamlessly at high settings --- but it doesn't get on with KSP: to get a decent framerate I have to run at 1024x768 with all the various low-quality hacks. As the card is now several years old it would seem like a good excuse to upgrade. Plus, having shadows that worked would make landing on Mün loads easier. I'm totally out of touch with modern graphics cards. What's the current sweet spot in price:power?

I had tonnes of trouble with my Behemoth lifter (still trying to live up to its name; so far, 15 tonnes to a 100km orbit, which is pants) with this. As soon as you start the gravity roll you begin tumbling, yes? I believe that it's a combination of lack of control authority and the vehicle weathercocking in the upper atmosphere at high speeds. Once you start turning, the air's coming from the side of the vehicle, the upper tanks are empty and therefore it's tail-heavy, and so the nose gets pushed round so that you're going tail-first. Which never helps. If your vehicle is asymmetrical it'll probably make things even worse. I think the only solution is more control authority. I did a bunch of experimentation with SAS modules and RCS and control services and I found that what worked for me were winglets, placed as low down as I could manage for maximum leverage. (I tried canards at the top but they were much less effective. *shrug*)

Do you want newtonian mechanics, or proper relativity? Relativity would be a lot of fun, if potentially hideously confusing and hard to implement --- your view of the universe would have to change depending on what vehicle you were focused on. Plus, of course, there's the whole FTL = time travel thing, which would probably cause problems with the physics engine.

So Kerbin has a surface gravity of 9.8 m/s3 (a very pretty coincidence there) and a diameter of 600km. This gives it a density of 58000 kg/m^3. This is hugely, weirdly, scarily high. Earth's is about 5500 kg/m3. The densest known material (under STP) is osmium, which comes in at about 23000 kg/m3; I see a few references that the synthetic element hassium (atomic number 108) will have a density of about 41000 kg/m3 if anyone ever makes enough of it to measure. But that's still not enough. Plus it has a half life of 12 minutes so even if you did have enough of it, you wouldn't have it for very long. Therefore the only plausible explanation for Kerbin's ludicrous density is that it has a core of some kind of degenerate matter. Electron degenerate matter is the stuff you normally get in the cores of stars. It's ludicrously dense... for a gas: it goes up to about 10000 kg/m3. The next step is neutron degenerate matter, a.k.a. neutronium; with a density starting at about 109 kg/m3 you'd only need a sphere of the stuff 47 km across. (Minus some to account for the rock padding needed to form the planet we all know and love and crash into.) The only problem is that AFAIK you can't have neutronium bodies smaller than 1.4 solar masses, or they explode. (There is also proton degenerate matter which is somewhere in between, but it seems to be rare and I can't find any references as to how dense it is. And there's quark degenerate matter, but that's even denser that neutronium and is scary.) So, unless some ancient god-like being put a blob of neutronium at the center of Kirbin inside a very strong box so it wouldn't explode, there is only one candidate left: Kirbin's core is a black hole. Just a small one, and the rest of the planet is otherwise ordinary rock; a black hole with a mass that of Kirbin's would have a radius of about 10 microns, so while it would be sucking in mass, it's not doing it very fast. And while small block holes do evaporate over time, converting their mass to energy increasingly more quickly as they shrink until they finally explode in a planet-destroying blaze of energy, this one will take about 1e52 seconds to do so. But it's still something to look forward to.

Fair enough. Thought it worth asking. Looks like my astronauts will just have to live with it. Er... well, sort of.