[0.24] If a Booster Has Enough Parachutes On It, Count it as "Recovered"

[Brian--Griffin]

Why not check around the corner DebRefund does exactly the recover thingy without altering physics range.

[Tuareg]

Yes, I did not consider that option. BUT... this would mean that every part with a control unit would constantly be inside an "active" physics bundle. The minute you say "well deactivate them after landing" you're just going to get other complaints by users who find that their plan of bomb-dropping probes all over the place isn't working.

Ok, so we keep 'm active*. Keep in mind that you're stuck in Physics Acceleration until all of those spawned off units have been resolved. And since you're running a handful of them (or more), Physics Acceleration at maximum speed probably means realtime. At best. More likely your simulation will be running at half-speed. Or quarter speed. "So what?" Well, you're going to be stuck with that for the full 500m those items are floating down on their chutes at a leisurely 7.5 m/s - That's about 1m gametime, and if you're running at quarter speed that's 4m of real world time. And that assumes you can keep track of them until they land and hit "recover" to speed things up. Let's hope your launch vehicle doesn't need attention in the meantime.

I'm not saying it's impossible. But it involves adding a lot of features to make it happen, possibly dramatically impacting framerate performance. It's a lot easier to say "design wisely. Boosters spent are boosters spent and won't be recovered. If you don't like that, start working on those SSTO skills"

What I do suspect in future versions is that some boosters will have some auto-recover flag. After all, the KD25K description already states "This super heavy booster is designed to be recovered after jettisoning. Once recovered, it is refurbished and refueled for another launch."

well seen you know nothing of development. the falling back could be simulated the same way as the rail for orbit with a simply check at the impact if the speed of the falling back part will destroy it or not. it doesnt need the unity physics just the primary school physics knowledge. you know, from weight and speed you can simply calculate the forces and check if its within the impact tolerance or not... you accept this simplified mechanism for the orbit why you want silly way over-complicated crap for the unsupervised reentry?

[cpast]

well seen you know nothing of development. the falling back could be simulated the same way as the rail for orbit with a simply check at the impact if the speed of the falling back part will destroy it or not. it doesnt need the unity physics just the primary school physics knowledge. you know, from weight and speed you can simply calculate the forces and check if its within the impact tolerance or not... you accept this simplified mechanism for the orbit why you want silly way over-complicated crap for the unsupervised reentry?

Because physics in orbit *are* substantially simpler. In orbit, with no thrusters of any sort firing and no reaction wheels moving, the only force is gravity. There are no other forces. With parachutes and with ground impact, there are multiple forces at play, and different parts have substantially different forces applying. There is literally no possible equivalent of rails; rails are based on a (reasonably accurate) patched conic Keplerian orbit, which is analytically solvable. With such an orbit, there is no force simulation required. Keplerian orbits do not require any considerations of force, except to initially derive orbital parameters; after that, there is an equation in the orbital parameters (which don't change during the orbit) and time that returns the position of the orbiting object. That means there is no need to go from time A to time B by means of many small steps - you can directly calculate where it will be at time B, without knowing where it will be at any other time.

No such thing exists for ground impact. First, the force is initially applied just to the bit that touches the ground; that means rotation angle is key (while it is irrelevant where there aren't collisions, such as in orbit). That has to be simulated. Likewise, you need to look at where precisely the part will land, what slope it has, what torque is applied on initial touchdown, how the part behaves from there... it is far more complicated than setting position equal to Pos(orbit, t).

[Kerbart]

well seen you know nothing of development. the falling back could be simulated the same way as the rail for orbit with a simply check at the impact if the speed of the falling back part will destroy it or not. it doesnt need the unity physics just the primary school physics knowledge. you know, from weight and speed you can simply calculate the forces and check if its within the impact tolerance or not... you accept this simplified mechanism for the orbit why you want silly way over-complicated crap for the unsupervised reentry?

Sure. Show me the math, if it's that simple.

Also: did you actually read the title of the thread? Unless I'm mistaken it's "remove local atmospheric despawn because of recoverability" and not "regard despawned jettisoned parts as recovered under certain conditions" but English is a second language for me, so please excuse me if I got that wrong.

Edited July 21, 2014 by Kerbart

[Kegereneku]

Keep in mind that you're stuck in Physics Acceleration until all of those spawned off units have been resolved. And since you're running a handful of them (or more), Physics Acceleration at maximum speed probably means realtime. At best. More likely your simulation will be running at half-speed. Or quarter speed. "So what?" Well, you're going to be stuck with that for the full 500m those items are floating down on their chutes at a leisurely 7.5 m/s - That's about 1m gametime, and if you're running at quarter speed that's 4m of real world time. And that assumes you can keep track of them until they land and hit "recover" to speed things up. Let's hope your launch vehicle doesn't need attention in the meantime.

I think we are not on the same track here...

Adding other "physic bubble" will indeed ask more to the computer but it shouldn't in theory slow down anything else. (who know if we can't make the bubble simpler or smaller than needed for a player)

Considering how long is an efficient launch (with a good gravity turn) the booster would have more than enough time to go down the last 500m + the dive. Myself I rarely go over x4 to reach apoapsis unless I'm aiming high and circularizing can take me 2 minutes.

Lastly, even 4 full minutes (supposing you are not still busy, dropped the booster late and didn't used phys-warp since then) is nothing compared to the new possibility ! (I also see little reason dropping several probes at once wouldn't work under this hypothetical situation, even if you required an arbitrary limit to the number of physic-bubble)

well seen you know nothing of development. the falling back could be simulated the same way as the rail for orbit with a simply check at the impact if the speed of the falling back part will destroy it or not. it doesnt need the unity physics just the primary school physics knowledge. you know, from weight and speed you can simply calculate the forces and check if its within the impact tolerance or not... you accept this simplified mechanism for the orbit why you want silly way over-complicated crap for the unsupervised reentry?

He clearly understand it more than you.

Didn't you get we are discussing about a way for booster to land under the very same conditions than any other rockets, rather than asking the game to magically consider them "recovered" ?

In the game, object "on-rail" are little less than a point, data and a vector, they have no friction, no form, no acceleration, no potential energy to calculate, no ground to impact and thus definitively no impact tolerance because any of those would involve the physics-engines.

The only place this "on-rail" simplification work (barely) is in the void of space. That's WHY the game was made like that.

[Aggregat@13681]

Only more complicated/bigger parts with enough parachutes should have a chance of recovery. Some large boosters and jet engines, for an example.

[Tuareg]

Didn't you get we are discussing about a way for booster to land under the very same conditions than any other rockets, rather than asking the game to magically consider them "recovered" ?

what are YOU discussing and what was the original op are 2 completely different things but no worries.

there was nothing about very same conditions... they are left to their fortune... there is no any need of an other physics "bubble"

(however even that would be solvable as we are talking about the one single ship falling into pieces, if unity was able to calculate their physics together

with extra connections, it would be able to calculate them as separate crafts too, especially as they "live" only for minutes)

their fallback can easily be calculated. there is a suborbital rail its going till it reach the atmosphere.

now instead of erasing it there a node could be added (generally a burn against the orbit-vector as the slowing down effect of entering the atmosphere) and at 500m an other node an other negative burn as the slowing down effect of the parachutes. when it lands get the speed the mass and the collision tolerance of the weakest part. (yes, not the colliding part count, its the stupidity of KSP. try fixing a glass between 2 rocks and drop it. let see what will break...)

from these its easy to know if it survive or not. OP was asking to remove the stupid part erasing not to give a realistic landing for every unseen booster.

[Tuareg]

Powered landings are very successful...

With the new vernor rockets, this is more possible:

http://youtu.be/sSF81yjVbJE

http://youtu.be/sSF81yjVbJE

Currently KSP will destroy both those reusable stages if you attempt it in game...

[edit]

WAIT... What? I just tested it... KSP now lets you fire ships out on ballistic trajectories in the atmosphere, and keep them alive?!

I'm SOOOO making that grasshopper reusable rocket now... Although I still need to get perfect timing between craft, as it's going to drift quite some way without power (as it's outside sim range).

that video is nice but its just fantasy. there is no pod can hold enough fuel to touch down a pod like that from freefalling. in KSP you can solve the problem if you give a pod with crew to the stages and after separating you shoot the next stage further and while that reaches its AP, you manually land the dropped stage. at least thats what i do/did. i just gave up on these silly missions

[Kegereneku]

...I don't think you understood why we are talking of physic bubble and what mean "no physics" for object outside of it.

OP's suggestion was misinformed to begin with.

The game DO NOT and CANNOT calculate the physics or every object in the solar system in real-time. So once they get 2km away from the player (the bubble), the game stop considering them as object with a mass and apply a simplified trajectory that is only correct in space with no outside force like lift or drag.

This is the reason they erase object going through the atmosphere, because their trajectory wouldn't make any sense.

So now, your "solution" is basically to "apply physic but not too much" which is both : first, not preferable because it make a mockery of the effort of the game to simulate a space program (and not play pretend), and second it require to calculate the physic anyway.

We did understand what you are trying to defend, we (I in any case) just think this is a bad idea to "pretend" the whole booster/stage landed homogeneously on a flat surface.

[Clockwork_werewolf]

The big problem with the two or more physics bubbles are for when people are using large part number ships. If you are running just below the point where normal time slows (goes yellow or red) then adding a second or more bubble would cause things to slow a lot. If you are only using 50 to 100 part ships then you may very well be able to have three or four "bubbles" running at the same time without problems.

I did think of one possible solution of a sorts. Have a booster that has a built in parachute. Being on part it would have a default landing speed and as such not need to calculate that, it would also have a default drag and landing pattern. All the game would need to do is recalculate the landing zone with given and predictable drag (lets assume it is built to not tumble) then "get" the angle of the land at that location and if below a certain angle then recover based on distance form KSC.

One downside is there would be no mid air collisions but then they are rare for me and you could always have a chance of collision if more than one exits the normal physics bubble within a second of another.

The other great downside is people would just keep using that one booster because it is recoverable.

Really there is so much money about that recovering every part is just not an important project for the KSP team to work on. Modders are welcome to work on it (and already are) as we all have difference play styles.

[ScallopPotato]

I'm going to suggest that boosters that parachute down over land have higher recovery values than those landing in the ocean. Salt water corrosion was apparently a major issue with the shuttle's SRBs and made refurbishing them more expensive. The same would apply to liquid engines and all their delicate parts.

In addition, if the booster in question has fuel left over at separation and a probe core, the game could also calculate if it has enough delta-v to do a rocket-assisted landing.

Until that happens, I'm probably just going to strap multiple Marzia Boosters (my name for the LFB KR-1x2) and jumbo tanks together and use them to reach the ~3.3 km/s needed to reach orbit with Ferram.

[tntristan12]

I think people might be overstating how often this occurs. I know that when I personally play KSP the only time I get objects reentering any atmosphere (but particularly Kerbin's) is around launch when I drop lower stages, or when I have an upper stage designed to kiss the atmosphere and fall back to the surface (think pre-circularization or after a free-return trajectory from the Mun). All told I have never had more than 5 objects in the atmosphere at any one time. Furthermore, I'm not usually outside of Kerbin's SOI. For so few objects I think it's perfectly reasonable to run physics, even if we have to wait for multithreading first. Whackjobian nightmares are generally an outlier, and let's be honest, they're gonna have enough physics load on their own.

[Clockwork_werewolf]

I think people might be overstating how often this occurs.

It all depends on play style.

If you are like me and would much rather send Kerbals than probes to do science then the size and part count of ships increases.

If you can get parts back from dropping then people would have more stages and objects in the air.

I use a lot of boosters and even now with the decoupler costs I still use 2 sides of boosters. This means even without changing my play style to have more than two to three stages I would have at least 3 bubbles at once. With 2 fast burn and 2 slow burn booster stages and three main stages I could easily have 7 bubbles in the air at once.

In programming you can't just program as if people have powerful computers and resources parts sparingly for the stock game. It is fine to assume so for mods as they are optional but stock needs to work well for everyone. About half of the people that play KSP seem to use low part counts and half medium to high part counts, that's too many to program stock to use 2-100 bubble physics.

[Kerbart]

I think we are not on the same track here...

Adding other "physic bubble" will indeed ask more to the computer but it shouldn't in theory slow down anything else. (who know if we can't make the bubble simpler or smaller than needed for a player)

Considering how long is an efficient launch (with a good gravity turn) the booster would have more than enough time to go down the last 500m + the dive. Myself I rarely go over x4 to reach apoapsis unless I'm aiming high and circularizing can take me 2 minutes.

Lastly, even 4 full minutes (supposing you are not still busy, dropped the booster late and didn't used phys-warp since then) is nothing compared to the new possibility ! (I also see little reason dropping several probes at once wouldn't work under this hypothetical situation, even if you required an arbitrary limit to the number of physic-bubble)

Calculations inside the physics bubble are expensive since they take each part into consideration. Why not introduce a third physics mode, call it for argument sake "the simple bubble" where objects are still not on rails, but are now considered a single part with x mass. "Simple bubble" objects would only be spawned from objects under control and get destroyed (in an OOP sense) under the following circumstances:

• Hitting the surface above impact speed (determined by the weighted average maximum impact speed of all parts)
  
• Hitting the surface below impact speed -- object is respawned as the original object but is now static at surface like other debris/vessels on the surface
  
• Orbit: when the periapsis goes above zero, the object is respawned as an "on rails" object
  
• Destroyed: once outside twice the radius of the current SOI body the (it can be pretty large as the objects are lightweight to evaluate) and not meeting one of the above criteria, the object is taken out of the game
  
• Focus switch: any mini bubble objects are destroyed when the player switches to another craft.
  

Minibubble objects would have very simple behavior: they're immutable in the state they were at creation (so you better make sure those chutes activate when staging). Above criteria would allow you to do a "bomb run" on an atmospheric body, drop parachute probes over a large area to cover multiple biomes and get science, without having to make a full descend. It would also allow recovery of booster stages (downside of that would be that everyone would apply parachutes to anything to recover it).

I have no clue to what degree it could be implemented though (and I doubt it ever will)

[Kegereneku]

This is no different from all other "pretend my booster had enough parachute and landed perfectly on a flat surface" suggestion. I think we should either simulate entirely the landing/recovery process or don't bother with it.

I would not suggest we give a pseudo-physic-bubble to any object dropped, only to sub-assembly with a specific part made specifically to make the distinction between debris and reusable-stage (let's say a control part).

The core idea is to only recover stages you worked hard to make recoverable, harder than just adding a chute.

[Monger]

I think it is amazing how many players regularly demand overly complicated solutions just so that the game becomes easier.

Let's ignore the atmosphere for a second. So what you basically want is that all debris parts that can land in any way actually do that automatically, if you are not in direct control?

Even if you can find some algorithm that can predict that at least partially (needs a certain amount of chutes per metric ton, needs a control unit, a battery with enough power, or an energy source that will not break when entering the atmosphere...), where is the fun in that?

The current rule is nice, because it doesn't only require you to separate your stages somehow, but to land them properly as well. It is a nice bonus task for those who want to maximize their recovery value.

By the way: in real life, just because a booster splashed somewhere in the ocean doesn't mean that it is recoverable.

[LethalDose]

To contribute to the discussion: I feel like this is an over-complicated solution to a problem that just doesn't exist. There are way bigger issues in the game than dealing with ballistic debris.

Seriously, how is this topic not on the WNTS list already. It keeps coming up, the discussions go nowhwere, there's apparently a mod to deal with it.

[tntristan12]

Calculations inside the physics bubble are expensive since they take each part into consideration. Why not introduce a third physics mode, call it for argument sake "the simple bubble" where objects are still not on rails, but are now considered a single part with x mass. "Simple bubble" objects would only be spawned from objects under control and get destroyed (in an OOP sense) under the following circumstances:

• Hitting the surface above impact speed (determined by the weighted average maximum impact speed of all parts)
  
• Hitting the surface below impact speed -- object is respawned as the original object but is now static at surface like other debris/vessels on the surface
  
• Orbit: when the periapsis goes above zero, the object is respawned as an "on rails" object
  
• Destroyed: once outside twice the radius of the current SOI body the (it can be pretty large as the objects are lightweight to evaluate) and not meeting one of the above criteria, the object is taken out of the game
  
• Focus switch: any mini bubble objects are destroyed when the player switches to another craft.
  

Minibubble objects would have very simple behavior: they're immutable in the state they were at creation (so you better make sure those chutes activate when staging). Above criteria would allow you to do a "bomb run" on an atmospheric body, drop parachute probes over a large area to cover multiple biomes and get science, without having to make a full descend. It would also allow recovery of booster stages (downside of that would be that everyone would apply parachutes to anything to recover it).

I have no clue to what degree it could be implemented though (and I doubt it ever will)

For the record: this "simple bubble" approach is exactly how I would advocate doing it. Some loss of accuracy can be expected but the player is ostensibly not looking too closely at the exact physics of the problem. It strikes me as a more elegant solution than simply assuming anything that hits the atmosphere just blows up.

[cpast]

And yet again, I have to ask: what gameplay justification is there for letting all stages be recovered if you slap parachutes on it, rather than making you work and plan harder to lose as little as possible? If all parts are recoverable, it undermines the whole point of a funds system -- pretty much all it restricts then is missions happening at once, not total missions. Recovery of spent stages should be fairly hard to do, so that you actually are rewarded for the more challenging task of making it to orbit with as few dropped things as you can.

[tntristan12]

And yet again, I have to ask: what gameplay justification is there for letting all stages be recovered if you slap parachutes on it, rather than making you work and plan harder to lose as little as possible? If all parts are recoverable, it undermines the whole point of a funds system -- pretty much all it restricts then is missions happening at once, not total missions. Recovery of spent stages should be fairly hard to do, so that you actually are rewarded for the more challenging task of making it to orbit with as few dropped things as you can.

Uh, since when is working harder to make your spacecraft more reusable a bad thing? You slap parachutes onto it yes, and parachutes cost money! Furthermore you're still not recovering the full cost so something is better than nothing. It doesn't render funds unnecessary in the slightest. One could argue that this kind of mechanic is exactly the point of funds. It makes you think about your expenditure and try to optimize returns.

Furthermore, what about manned pods or scientific payloads? You might want to drop them into an atmosphere on eve but keep focus elsewhere. You might not be recovering it but being able to land something outside of focus has tangible benefits as well.

[cpast]

Uh, since when is working harder to make your spacecraft more reusable a bad thing? You slap parachutes onto it yes, and parachutes cost money! Furthermore you're still not recovering the full cost so something is better than nothing. It doesn't render funds unnecessary in the slightest. One could argue that this kind of mechanic is exactly the point of funds. It makes you think about your expenditure and try to optimize returns.

That's my point. This *doesn't* mean "work hard to make spacecraft reusable". It means "slap parachutes on to make spacecraft reusable". Parachutes cost money, but not that much, especially relative to how much boosters cost. You say "something is better than nothing", and I agree. That's why actual effort should be required to recover something, and adding parachutes to an existing design (which is rather easy) should return nothing. As to why this undermines funds: For funds to provide a meaningful restriction, launches should be fairly expensive. Letting you recover the majority of the price of your whole rocket means launches are *not* expensive; the main function of funds then is just to restrict simultaneous missions.

As an example of parachute price: Calculations based on the values in the wiki (and confirmation ingame) suggest that one radial parachute is sufficient to recover a Kerbodyne LFB. That's $700 for a $13,000 part (LFB minus fuel). At that point, there is no reason *not* to use parachutes, which will return the majority of the cost of the booster, meaning you have much less incentive to do the much harder task of minimizing dropped stages.

Furthermore, what about manned pods or scientific payloads? You might want to drop them into an atmosphere on eve but keep focus elsewhere. You might not be recovering it but being able to land something outside of focus has tangible benefits as well.

Landing maybe (for crew and science), but funds should be no recovery.

[Technical Ben]

that video is nice but its just fantasy. there is no pod can hold enough fuel to touch down a pod like that from freefalling. in KSP you can solve the problem if you give a pod with crew to the stages and after separating you shoot the next stage further and while that reaches its AP, you manually land the dropped stage. at least thats what i do/did. i just gave up on these silly missions

Why is it fantasy? They are already testing the first stage and the crew capsule. It's just the middle stage they have not yet tested... so why is it still fantasy when they have such resources and flights already?

AFAIK it does use more fuel, but they've taken that into account with a larger tank/engine in the first stage and it turns out it's not much more than a normal crew capsule launch, as the recovered parts more than enough make up for the cost of a few ton of fuel.

[Monger]

You should get some funds back so long as your booster has enough parachutes to land safely and doesn't experience reentry heat or land on a mountain. Those parts wouldn't just vanish normally (excluding reentry) and you do deserve a bit if you put in money to recover them.

This is just wishful thinking. In reality, an uncontrolled landing usually means you have no chance at all to recover your stage. It starts with the proper landing site: not all parts are buoyant, so any kind of water surface (> 70%) is out of the question. Also all forests, jungles and mountains are out of the question. Recovering from arctic or antarctic climate would be way too expensive, landing near populated areas is too risky.

KSP makes this a lot easier, you can land basically anywhere. The only requirement is that it is a controlled landing, which is a perfectly reasonable demand, I think.

[Maelstrom Vortex]

So why not use chutes to control the landing? I've literally built craft with chutes designed to have it fall one particular way and sharp edged structural supports to keep it from rolling once it did gently touch down. Just because it hasn't been thought of or done doesn't mean it cannot be innovated.

[Monger]

When discussing about gameplay, it is almost always useful to ask yourself: does this create an interesting decision?

Before funds and recovery were introduced, there was no reason to keep parts and land them in a controlled manner, other than crew or scientific equipment. So once funds and recovery were introduced, this created an interesting choice: should I make my rocket as simple as possible and non-recoverable, or invest a bit more and make it recoverable? Both strategies are valid, both make sense in specific cases.

Now, what would I gain if I make every part auto-recoverable that has enough parachutes attached? What choice does that create? This could be an interesting choice if there was any significant downside to a parachute: significantly increased drag while packed, extremely high costs...

But since this is not a case, it is just an exploit to automatically reduce part costs by attaching another part. No interesting gameplay mechanics are gained, but it introduces a bunch of very intransparent game rules. If this is what you want, SQUAD would be better off to trash the whole recovery system and just lower the part prices.