Well, here we go into the 0.16 update. As with last time, we'll post here not only the features, but also their status and confidence level for having them included in the update. However, we're not going to post any dates anymore, because quite simply, we can't promise that we'll meet them. Our deadlines are going to be internal from now on, so when the whoosh by, only we will know.

So, as I said a few days ago, we've decided to take a somewhat different direction on this update. We had been working too long in the same set of features, while there were other things to do that weren't even started. KSP isn't just a space flight simulator, and we want to get started on the other areas of the game now, not only because those will make for much more interesting releases, but also because we've been working on flight-related features for the last 6 months, and this is a very welcome change of pace here.

So, without further ado, this is the feature lineup for 0.16. Keep in mind that it's very likely several of these might not make it in the final update. We'll update our confidence rating for it as we go along:


Planned Core Features:

* Kerbal crewmembers in Extra-Vehicular Activities, both at the surface and in orbit.
Assigned to: HarvesteR, Daniel
Status: Early Development.
Confidence Level: Reasonably confident.

* Internal Cabin and Cockpit views for crew-accessible parts. (aka, I.V.As)
Assigned to: C7, Mu
Status: Assets Development
Confidence Level: Unknown

* New Particle Effects for Explosions and Parts
Assigned to: NovaSilisko
Status: 2/3rds complete
Confidence Level: High


Bonus Features (if we have the time):

* A much improved parts catalogue UI on the construction scenes.
Assigned to: Mu, HarvesteR
Status: Early designs
Confidence Level: Unknown


Unplanned Surprise Features:
* Wait, what?


Bug Fixes and Tweaks:
* Set all parts on the same vessel to ignore collisions with each other. Hopefully this will solve all part-to-part collision problems.

* Reduced the minimum collider intersection threshold to much less than before. Should increase collision accuracy and hopefully solve issues like landing gear going through the ground. (all thanks to above fix)

* [API Change] MemoryStream wrappers that were mistakenly not committed for 1.15.1 have been added to the code.


Frequently Asked Questions:


Q: What does IVA stand for?
A: Intra-Vehicular Activities, as opposed to Extra Vehicular Activities. Basically, you assume a 1st person perspective on the selected crew member, and you get to see the ship from the inside (where available).

Q: What about docking?
A: Docking is still being implemented. As we worked on it, it became more and more apparent that docking is really just the tip of a very large set of features and code changes, which include much more than just attaching ships to one another. Docking requires a pretty sizeable overhaul of several core systems in the game. Those changes however, will make it possible for us to have not only docking, but several related features will become possible to implement, like space station construction, surface bases, multiple vessel launches and so on. We want to make this work as well as possible, so we are taking our time to make sure we have a very solid base to build on. It's possible some parts of these changes will be present on 0.16, but there's no guarantee we'll be able to complete it.


Q: And what of Flight Planning then?
A: Flight planning is half-way implemented already with the patched conics trajectory display. The original plan for this update was to continue developing it, but we decided it was more important to focus on the least complete areas of the game first.


Q: Last time you said there wouldn't be any surprises, and you added a new moon. What can I expect for this time?
A: Minmus was a surprise even for me. In all seriousness, no one should expect anything that isn't written in bold and green on this list.

Time for a new update on EVA development I think. There are just so many questions about it on the dev board that I'm not even going to try to pick ones to answer right now. I'll just talk about our ideas for it in general:


So, currently EVAs are being developed on a little test scene, and so far the main focus has been on setting up the Kerbal character object in the game, so that it can be animated and moved around in the way we want it to. That's pretty much done I'd say.

The next step then is to set up the control logic for them. I'm working on a state machine system here, that will switch between the several behaviours based on your inputs and the current situation. It's not very simple, as you can imagine. Not only there are a few dozen animations already (and more are coming), Kerbals have a few overlapping states, which is somewhat difficult to manage, and several states will need to run their own logic as well, almost like sub-states.

What we have done so far though, is this: Kerbals are already able to walk and run around the surface of a spherical planet. They are handled by the flight model as a single-part vessel, so that means whatever parts do in the game, Kerbals can do also. They can be put into orbit, they have a heat exchange system, they can (in theory) attach to things physically, and most importantly, they're persistent. The Part and Vessel system lend themselves pretty nicely to an astronaut in EVA. Their spacesuits really are small, self-contained spacecraft.

Kerbals are also able to become ragdolls here. Not just when they die, they can seamlessly transition from animation-driven control to a physics-driven ragdoll, and once they tumble to a stop, they transition back to animation control and become controllable again. If you've played GTA IV, it works in the same way as when you crashed a motorbike and sent Niko tumbling down the road. If your Kerbal doesn't die from the tumbling, he can pick himself up and walk it off.

Kerbals will become ragdolls from several situations. If they are hit by another object fast enough, if they themselves slam into something, or if they're walking and lose their footing, which can happen if you decide you want to run down steep terrain, or if they start falling without you having commanded them to jump first. Those last two, of course, only apply if you're at the surface.

The ragdoll system is already implemented and working here. I've been having a lot of fun making them run around the test scene and stumble down. As was mentioned in that thread about how they should die, ragdolls are just the right blend between realism and cartoony deaths. We're just missing the 'pick yourself up' animations to call that part done.


What needs to be done next now is to continue implementing the movement logic for surface EVAs, and after that the plan is to work on the orbiting EVA controls. When Kerbals come off the ground, they will go into 'floating' mode, and use their EVA pack's maneuvering thrusters to move around. This mechanic is the primary means of moving around in orbit, but we want them to still be able to use the ladders as hand-holds there, as I can imagine moving outside a space station without being able to grab hold of something would be quite difficult.


There's a huge amount of work to do on EVAs still, but we don't plan on having it all in for this update. We'll add as much as we can of it, and continue developing this part of the game on later updates. Right now, we just want to have the essentials bits of it in the game.

Another update here:

Yesterday I spent the entire day completely focused on the EVA control system, and it was a really good day. It's been a while since I was able to dedicate an entire day to doing just one thing.

Anyhow, I've made some progress with the control system. It's nowhere near complete yet, but it's starting to look like something.

I'm getting the hang of my new FSM system I concocted here, and the more I get used to it, the better I like working with it. It's very easy to add new states and events, and because it won't change states without a proper event, it's pretty easy to see if something is going wrong. You always know what was the cause of the last state change.

Most of the work on it so far was just to rewrite the old prototype code into the new format, but yesterday, I was finally able to start adding more stuff to it, and I figured out some conceptual issues I was struggling with.

For instance, how to make sure Kerbals walk properly under the different gravities of each world, was one question that was really bugging me. My original plan was to somehow do a controlled blend of the different walk cycles, based on the current gravity, but that approach had flaws. That idea worked ok to transition gradually from Earth-like gravity to Mun gravity, but what of the extreme cases, like Minmus? Animation cycles just aren't flexible enough to make that work across the entire range of situations.

What I decided on then, was to set up a threshold for the G force, below which walking normally is simply impossible (as it probably is IRL). Then, if the current body G force at surface is below that, we switch to a low-gravity surface walk system that isn't an animation that looks like jumping around, it IS actually jumping around. Kerbals are perfectly able to jump in all directions, and when they jump, they go into 'floating' state, where they basically wait for the ground to come up again, and then they land (or stumble and ragdoll).

With this system, if the gravity is low enough, each "step" is a sub-orbital trajectory, and you can use your EVA pack to help yourself along with its thrusters. Also, this system can work at any level of surface gravity, even at near-zero G. Of course, on such a situation, your first step would probably throw you into orbit.

I want to post a video the new EVA system soon, I think there is only so much I can convey by writing, and it's always fun to watch Kerbals tumbling out of control.

Time for another update:

Work continues on the EVA control system, and at least for basic surface movement, I would say we have reached a point where it could hold its own if it were to be released as is.

That means the next goal is to get the flight controls working now. That means new states to the control system FSM, possibly even new FSMs as well. Right now, I'm planning how to best implement the EVAPack™ logic.

The plan is that the EVAPack™ will contain all the basic necessities to keep a Kerbal astronaut alive outside his spacecraft and move around. Right now, what we want to do is implement this as an all-in-one EVA solution, which we can add in the limited time we have for the update, and later make it more customizable and hopefully also moddable. For now though, we just want to have it in the game working. Even if it means no customization options for EVAs on 0.16, we'll at least have the EVAs.


So, what I'm planning on doing for the EVA systems programming is to split each bit of functionality into separate module controllers. You'd have a main one that handles character controls, like walking, jumping, ragdolling and swimming, and other controllers for things that could be attached to him, like RCS thrusters for MMU-like maneuvering, headlamps for night operations, reentry survival systems (think MOOSE) and so on.

This idea ties in nicely with the modular parts system we have implemented already actually. Each one of those systems would be basically a part module, that adds another bit of functionality to the EVA Kerbals. This is another indication for me that the decision to implement EVA Kerbals as actual vessels was a very good one. The spacecrafty things just get done, without us having to reinvent the wheel.


I did run into a bit of a snag with the ragdolls here today tho. Because the ragdoll uses other rigidbodies than the main one used during animation control, the ragdoll colliders send their collision messages to those rigidbodies, and the EVA Vessel is unaware of whether the ragdoll is landed or not, because its own collider is disabled during ragdoll.

To fix that, I assigned each ragdoll rigidbody object a collision event handler component, that receives the collision events on each of the ragdoll colliders, and forwards them to the Part. The part then uses its already-implemented collision handling system to let the vessel know that it's landed. The solution worked, so now we can rely on the vessel situation detection system. That takes care of a lot of potentially complicated issues.


I guess that's about it for now. More news later (and hopefully a video as well).

This here was just a test to see how well EVA objects can be instanced in the main flight scene. Other than a couple of manual tweaks I had to do here, it works just as expected.

The guy on the screenshots was spawned manually here. There isn't any code yet for actually taking a crewmember out of the ship on EVA. It should be relatively straightforward to do though, once the EVA control system is done.

Well, new update today, I've been poking around at having Kerbals spawning in the main flight scene, and found an issue that needed to be taken care of.

When parts unpack, they reset their rigidbodies in a pretty agressive way. That causes the ragdoll on the Kerbal object to lose references to the root rigidbody, which means the ragdoll joints lose the attached body link, and attach themselves to world space. I don't need to tell you just how much that disrupts normal functioning of the whole system. It's pretty bad.

Fortunately, it was a simple fix. We just needed to reset the ragdoll links that get broken when the part unpacks, and everything worked just fine. Trouble was, the EVA controller didn't know when the part was unpacking, until it was too late. So I set up the vessel and part components to fire off messages when packing and unpacking is being done, to give other attached bits a chance to do stuff.

That worked out pretty well, and I just needed to tweak a few details to get the kerbals packing and unpacking in the game, just like any other vessel. Some cheating later, we had a Kerbal walking on the Mun:



That same jump wouldn't get you more than a foot off the ground on Kerbin. Things are a lot more bouncy and fun at 1/6th of
a gee.

Well, more progress news now:

I've been tinkering with the part physics and how the Kerbals interact with that, and I think I've been able to simplify the part packing/unpacking system to a point where it is not only more reliable and stable, but also much less laggy on the transitions.

This is a change that's right on par with the same-vessel part collision ignore thing, as far as how deeply-rooted it is, and how fundamental the changes are, so it needs to be under observation here, to make sure it doesn't cause any unwanted side effects.


So far though, it looks good, and some more cheating later, I got a Kerbal on Minmus. Here's me jetpacking around over a frozen methane lake:



This reminds me, I need to get the EVAPack thruster effects in still... Well, one thing at a time.