VR support

[mcwaffles2003]

14 hours ago, Incarnation of Chaos said:

imagine the part you've spent the last few minutes carefully positioning suddenly snapping to a booster and clipping through the VAB along with your hand. 

How would this work? I have not experienced anything like this phenomenon on my quest 2 so far.

[Guest]

1 hour ago, mcwaffles2003 said:

How would this work? I have not experienced anything like this phenomenon on my quest 2 so far.

It works by playing VR in a room full of mirrors and other overly reflective surfaces, at least if you're playing on Valve's Lighthouse system (covering reflective surfaces is a small price to pay for the upsides of perfect tracking).

Edited May 23, 2021 by Master39

[Incarnation of Chaos]

5 hours ago, mcwaffles2003 said:

How would this work? I have not experienced anything like this phenomenon on my quest 2 so far.

 

3 hours ago, Master39 said:

It works by playing VR in a room full of mirrors and other overly reflective surfaces, at least if you're playing on Valve's Lighthouse system (covering reflective surfaces is a small price to pay for the upsides of perfect tracking).

Gents I'm talking stock KSPs behavior in VAB, and how parts will occasionally "snap" to literally the last thing you wanted. 

And how moving parts certain ways can send them several dozen meters through the VAB with no way to get them back in focus.

These are some of the primary reasons I always run hanger extension mods, even if I'm not making huge rockets. 

VR isn't going to change the snap system, just abstract it away from people. Which is fine, but my overall point basically was.

"they probably should make sure KSP2s implementation of snap and construction is way less quirky before even considering VR"

[Guest]

1 hour ago, Incarnation of Chaos said:

Gents I'm talking stock KSPs behavior in VAB, and how parts will occasionally "snap" to literally the last thing you wanted. 

And how moving parts certain ways can send them several dozen meters through the VAB with no way to get them back in focus.

These are some of the primary reasons I always run hanger extension mods, even if I'm not making huge rockets. 

VR isn't going to change the snap system, just abstract it away from people. Which is fine, but my overall point basically was.

"they probably should make sure KSP2s implementation of snap and construction is way less quirky before even considering VR"

In VR you can see 3D and parts are linked to your hands, they're not going to snap 200m away because you control the precise 3D location of the piece, not only its projection on a flat monitor.

The rawest implementation of a VR system would be completely scrapping how things work with the mouse and just replace it with a natural pick-up and move interaction system, which is basically one of the most documented things you can do in VR.

And, even before talking about VR, a 3D editor is not rocket science, I'm pretty sure it's not even on the list of hard things to implement, It may has problems in KSP1 for the same reason of all the other problems of KSP: it grew into place over the course of years with a bunch of inexperienced devs each one following different design principles.

[Incarnation of Chaos]

7 minutes ago, Master39 said:

In VR you can see 3D and parts are linked to your hands, they're not going to snap 200m away because you control the precise 3D location of the piece, not only its projection on a flat monitor.

The rawest implementation of a VR system would be completely scrapping how things work with the mouse and just replace it with a natural pick-up and move interaction system, which is basically one of the most documented things you can do in VR.

And, even before talking about VR, a 3D editor is not rocket science, I'm pretty sure it's not even on the list of hard things to implement, It may has problems in KSP1 for the same reason of all the other problems of KSP: it grew into place over the course of years with a bunch of inexperienced devs each one following different design principles.

It's still a projection on a monitor....

The monitor is just strapped to your face instead (and then is 2 monitors with slightly different depth to trick the eyes). Otherwise you wouldn't even have the performance issues that come with it, since the GPU wouldn't be doing any translation from the 3D to 2D. 

But I'm not going to get incredibly technical, since that's not the point.

You already have "the exact 3D coordinates" for these objects VR or not (the game is still 3D internally, always) So if that was truly the issue, you could solve it easily (you'd have to store them somewhere, and update it).

Either way, you're now suggesting having two completely different ways of handling input for parts. Which is kinda proving my point.

Most people here are asking if their potato laptop from 9 years ago can run KSP2. So spending time and resources to build essentially a game within a game just for the limited number of VR users just strikes me as absurd.

Plus the bugs and conflicts introduced would probably make KSP look tame lel.

I think we both agree that KSP2 should definitely have a much better system for parts and selection. 

I'm just of the opinion that VR should be literally the last thing added.

So I think we'll have to just agree to disagree on this one, besides intercept is the one that'll make the decision.

They could drop a dev blog tomorrow for VR, and I'd have a hat to eat (made of cookies)

[Guest]

why would KSP 2 use their resources for a mode that many people wouldn't be able to play?

It would take a lot of time for a small community

[LeHuH]

I think this is optional to add VR support. Most logic way to do it is add support in the update.

[Guest]

2 hours ago, Incarnation of Chaos said:

It's still a projection on a monitor....

The monitor is just strapped to your face instead (and then is 2 monitors with slightly different depth to trick the eyes). Otherwise you wouldn't even have the performance issues that come with it, since the GPU wouldn't be doing any translation from the 3D to 2D. 

But I'm not going to get incredibly technical, since that's not the point.

You already have "the exact 3D coordinates" for these objects VR or not (the game is still 3D internally, always) So if that was truly the issue, you could solve it easily (you'd have to store them somewhere, and update it).

Either way, you're now suggesting having two completely different ways of handling input for parts. Which is kinda proving my point.

You're not understading how it works.

When you control a part to attach it to a rocket you have 6 axis to control to position and rotate the part, the KSP editor has to do a lot of work to interface those 6 axis and make it possible to you to control all of them with a 2 axis mouse.

When you make the thing VR you don't add another layer on top of the mouse one, a VR controller already has 6 axis of control, you just remove the mouse layer and replace it with a single key for the grabbing interaction, at that point the 6 axis of the piece are exact same from the controller.

The HMD just does an illusion of 3D, you're right, but VR is not just a screen strapped to your head, you're forgetting about the most important part, the controllers and for those there's no such thing as an illusion of 3D, they are just trackers for your real hand, so basically all of the above just ends up meaning that you're just removing the mouse interface to be able to grab and place parts with your hands, with nothing more than the standard libraries for VR input no complex anything to implement at all.

 

Then you can argue all day if it make sense to use the developing resources on such a feature all you want and that's a matter of opinion, but the technical argument remains, it's not difficult at all to implement and it would be a simplification over the mouse interface because you remove the need to interact with a 3D object with a 2D periferal.

 

EDIT: Just to show how my opinion changed over time, you can see I originally bought my VR headset without VR controllers that's what I thought of VR back then. Now, if KSP2 would launch with VR support I would build in VR only and then switch back to 2D for most flights except maybe for some IVA ones.

Edited May 23, 2021 by Master39

[Incarnation of Chaos]

On 5/23/2021 at 4:17 PM, Master39 said:

You're not understading how it works.

When you control a part to attach it to a rocket you have 6 axis to control to position and rotate the part, the KSP editor has to do a lot of work to interface those 6 axis and make it possible to you to control all of them with a 2 axis mouse.

When you make the thing VR you don't add another layer on top of the mouse one, a VR controller already has 6 axis of control, you just remove the mouse layer and replace it with a single key for the grabbing interaction, at that point the 6 axis of the piece are exact same from the controller.

The HMD just does an illusion of 3D, you're right, but VR is not just a screen strapped to your head, you're forgetting about the most important part, the controllers and for those there's no such thing as an illusion of 3D, they are just trackers for your real hand, so basically all of the above just ends up meaning that you're just removing the mouse interface to be able to grab and place parts with your hands, with nothing more than the standard libraries for VR input no complex anything to implement at all.

 

Then you can argue all day if it make sense to use the developing resources on such a feature all you want and that's a matter of opinion, but the technical argument remains, it's not difficult at all to implement and it would be a simplification over the mouse interface because you remove the need to interact with a 3D object with a 2D periferal.

 

EDIT: Just to show how my opinion changed over time, you can see I originally bought my VR headset without VR controllers that's what I thought of VR back then. Now, if KSP2 would launch with VR support I would build in VR only and then switch back to 2D for most flights except maybe for some IVA ones.

All i was saying is those 3D coordinates exist, and can be mapped to "2D peripherals" fairly easily.

Far easier than making an entire system to handle VR.

[Guest]

6 hours ago, Incarnation of Chaos said:

All i was saying is those 3D coordinates exist, and can be mapped to "2D peripherals" fairly easily.

Far easier than making an entire system to handle VR.

You need a layer of interaction to handle a mouse that translates 2D inputs to 3D outputs, for VR you just need to implement a grab key and the object coordinates will be bound to your IRL hand by the VR libraries.

No risk of anything snapping anywhere unless it's your IRL hand that's snapping around.

Natural manipulation, no abstraction needed your hand and brain do all the work without any abstraction layer and 2D conversion in the middle.

 

There's UI design work to do elsewhere, a lot of it, to make menus and other things work, no doubts, and probably the game would need something smart to make the flight scene work even without IVA all the time, but the building itself is a non problem.

[Incarnation of Chaos]

17 hours ago, Master39 said:

You need a layer of interaction to handle a mouse that translates 2D inputs to 3D outputs, for VR you just need to implement a grab key and the object coordinates will be bound to your IRL hand by the VR libraries.

No risk of anything snapping anywhere unless it's your IRL hand that's snapping around.

Natural manipulation, no abstraction needed your hand and brain do all the work without any abstraction layer and 2D conversion in the middle.

 

There's UI design work to do elsewhere, a lot of it, to make menus and other things work, no doubts, and probably the game would need something smart to make the flight scene work even without IVA all the time, but the building itself is a non problem.

There shouldn't be a risk of snapping with 2D perfs either lel

And there's going to be an abstraction layer no matter what, because KSP2 isn't going to be VR-only. So why not work on what the majority of people are going to use instead of this?

[K^2]

On 5/25/2021 at 6:01 PM, Incarnation of Chaos said:

Far easier than making an entire system to handle VR.

The whole point is that you don't really need anything to handle VR. Unity already handles VR. A 2D UI needs to interpret your intention when you're moving a part with a mouse, which is why snapping issues exist - misinterpretation of intent. VR UI doesn't need to interpret anything. So you simply inject the input one step further upstream and you're done.

The challenge is entirely in building up the menus and ability to select parts you want and get information you need about the rocket you're building. The actual building process in VR is the easy part. It's something a Unity-savy intern can set up in a day. Back when I tinkered with it, it took me a couple of hours to go from, "Hey, I have a working Unity project," to "Hey, I have a Unity project that works with VR." Granted, I had a lot of prior experience, with all relevant systems, but it doesn't look like Intercept is hiring people who learned C# yesterday.

Again, I want to stress that there's a lot more to making an existing game into a VR game, but the points you're focusing on are the ones that are simply a non-issue. Grabbing, moving, and attaching objects is by far the easiest thing to do in VR, and especially, if you have an engine/platform that's built around it, such as Unity.

[Incarnation of Chaos]

18 hours ago, K^2 said:

The whole point is that you don't really need anything to handle VR. Unity already handles VR. A 2D UI needs to interpret your intention when you're moving a part with a mouse, which is why snapping issues exist - misinterpretation of intent. VR UI doesn't need to interpret anything. So you simply inject the input one step further upstream and you're done.

The challenge is entirely in building up the menus and ability to select parts you want and get information you need about the rocket you're building. The actual building process in VR is the easy part. It's something a Unity-savy intern can set up in a day. Back when I tinkered with it, it took me a couple of hours to go from, "Hey, I have a working Unity project," to "Hey, I have a Unity project that works with VR." Granted, I had a lot of prior experience, with all relevant systems, but it doesn't look like Intercept is hiring people who learned C# yesterday.

Again, I want to stress that there's a lot more to making an existing game into a VR game, but the points you're focusing on are the ones that are simply a non-issue. Grabbing, moving, and attaching objects is by far the easiest thing to do in VR, and especially, if you have an engine/platform that's built around it, such as Unity.

oof, I'll concede.

[HotVector]

I would like to add to this discussion, with my experience as a Unity dev and a VR player.

One of the biggest challenges in making a *good* VR experience(not just a functional VR experience) is implementing input that supports all the controllers currently on the market and is extremely intuitive. A good example of this is the input in Half Life Alyx - it's not challenging to remember VR actions in Alyx and the actions are repeatable and you can perform those actions without thinking about them. This came after years of research by Valve, so it won't be easy to implement input that's as intuitive as Alyx in Kerbal Space Program, and it's most certainly not as straightforward as others on this thread are making it out to be. As for actually displaying KSP on a VR headset, the only constraint is performance, and optimization is not an easy topic in computer science by any means, especially when working with a game engine such as Unity which still has a lot of it's optimization features in beta(such as the Unity DOTS system). Even when it comes to a problem as simple as moving parts around in the VAB, it's not easy to translate an experience with a mouse/keyboard to a VR controller. How would you handle easily changing between all the various different modes, such as snap and mirror, for example? There's a reason that not every CAD designer uses a VR headset even though they have become affordable. And the challenges only get harder when you start working on other interfaces, like how would you operate a ship from IVA, or how do you position the exterior camera in a way that doesn't give every player severe nausea? There are even challenges with things like seamless switching between VR and normal monitor, which could make or break the VR experience in a game like KSP.

Implementing even the most seemingly basic features in VR needs a lot of thought put into it, so I wouldn't be surprised to find out that the KSP 2 developers didn't want to go through that effort, at least for the initial launch version. However, I am most certainly looking forward to flying crazy spaceships in VR

[Guest]

12 minutes ago, HotVector said:

One of the biggest challenges in making a *good* VR experience(not just a functional VR experience) is implementing input that supports all the controllers currently on the market and is extremely intuitive. A good example of this is the input in Half Life Alyx - it's not challenging to remember VR actions in Alyx and the actions are repeatable and you can perform those actions without thinking about them. This came after years of research by Valve, so it won't be easy to implement input that's as intuitive as Alyx in Kerbal Space Program, and it's most certainly not as straightforward as others on this thread are making it out to be.

This is beyond misleading. Years of research to make a good VR shooter, which is quite a bit more than manipulating objects, thing that Valve got right basically at the first try with The Lab, a 2016 demo.

Here's a one-man developed demo, Moondust, Valve gave out to devs alongside their "Knuckles" prototypes: 

Yes, they spent months working on door interactions, on what level of precision you need when reloading guns, on trying to fit melee combat even though they ultimately didn't, on the balance of a completely physical environment vs smart animations, on how much you can stretch the difference in position between the real and the virtual hand before breaking immersion and on the flick movement of the Grabbity gloves (how the Russels were called during the prototyping phase) and a ton of other such tiny but immensely important details that makes Alyx the best VR game out there, but all of that is completely irrelevant when talking about the very basic idea of picking up a piece from a table and put it on a model rocket you're building, even in its best possible implementation KSP VR would be simpler than the very first scene in Alyx.

 

22 minutes ago, HotVector said:

Even when it comes to a problem as simple as moving parts around in the VAB, it's not easy to translate an experience with a mouse/keyboard to a VR controller.

You don't have to translate the experience, the mouse/keyboard experience in the VAB is about using a 2 axis controller to move multiple 6DOF objects, in VR you just remove all the mouse/keyboard layer and enable the player to use his own 6DOF controllers (also known as "hands") to move the objects in the 3D environment. 

 

25 minutes ago, HotVector said:

How would you handle easily changing between all the various different modes, such as snap and mirror, for example?

How they do that on consoles? If you're lazy you can do the same, if you're creative you can use holograms, gestures and a ton of other VR-specific controls. Probably you don't need to go farther than SteamVR Home to find at least a dozen viable ways that a competent dev team could implement in without any problems.

 

27 minutes ago, HotVector said:

There's a reason that not every CAD designer uses a VR headset even though they have become affordable.

A friend of mine does that for a job, half of the time their company is fixing the work of some other company still working in 2D on hardware from 2 decades ago, and that's even without acknowledging the fact that affordability is not a problem in that environment, comfort is. I've done a couple of 8 to 10 hours runs in my Valve Index, but if I wouldn't work like that every day for 8 hours a day for  whole work week and the Index is as comfortable as current VR headsets can be.

I'm not expecting them to make a VR version at all, and I'm pretty sure they said a couple of times they don't plan for one, but please don't treat the thing as an impossible or incredibly hard goal because it is not one. Even disregarding the whole "VR controllers brings your hands 1:1 in the game" part of VR if KSP works on a PS5 controller it can work on a VR one without much changes.

[cubinator]

I would only want VR for realistic IVA flight sim. If anything, there could even be a custom modded capsule or lander pod which makes it more workable to sit in there pushing buttons/have a custom control panel. Basically a 'lite' version of any spacecraft trainer.

[Brofessional]

I wouldn't count on VR at release but it's certainly something that could be added post-release. It would probably be limited to IVAs though, the rest of the game would probably be sub-optimal in VR due to the complexity of the interface.

[Guest]

26 minutes ago, Brofessional said:

It would probably be limited to IVAs though, the rest of the game would probably be sub-optimal in VR due to the complexity of the interface.

The VAB is the best candidate for VR and it's where the VR controls would shine.

IVA will without doubts fall in the category of not enough effort put in for a full interaction model like VTOL VR and will end up being just another case of using it as a 3D monitor.

[K^2]

4 hours ago, Master39 said:

IVA will without doubts fall in the category of not enough effort put in for a full interaction model like VTOL VR and will end up being just another case of using it as a 3D monitor.

Idk, the effort bar to get the main controls, at least, interactable is super low. I don't know if it'd be anything more than a gimmick even then, but making the whole mission flyable entirely in VR is not a lot of work.