Install Steam
login
|
language
简体中文 (Simplified Chinese)
繁體中文 (Traditional Chinese)
日本語 (Japanese)
한국어 (Korean)
ไทย (Thai)
Български (Bulgarian)
Čeština (Czech)
Dansk (Danish)
Deutsch (German)
Español - España (Spanish - Spain)
Español - Latinoamérica (Spanish - Latin America)
Ελληνικά (Greek)
Français (French)
Italiano (Italian)
Bahasa Indonesia (Indonesian)
Magyar (Hungarian)
Nederlands (Dutch)
Norsk (Norwegian)
Polski (Polish)
Português (Portuguese - Portugal)
Português - Brasil (Portuguese - Brazil)
Română (Romanian)
Русский (Russian)
Suomi (Finnish)
Svenska (Swedish)
Türkçe (Turkish)
Tiếng Việt (Vietnamese)
Українська (Ukrainian)
Report a translation problem
> There's no friction in space
That's not very achievable in practice, because physics engine has finite precision and at high speed it breaks down, which is why almost all games limit maximum speed. See Elite Dangerous. The ship auto-brakes even with FA off, that is handwaved as a "safety feature" which you cannot be disabled.
> remember that much lauded 'Newtonian mechanics" from the SC kickstarter
No. A lot of people here never played it.
a= t/v−u Acceleration.
Imagine it.
At 1G; 9.8 meters per second squared (m/s²) or 32.174 feet per second squared (f/s²)
At 1G of constant acceleration for 1 minute, you would be traveling at approximately 1,020 meters per second. or 2281 Miles per hour, or 3670 KPH.
- What game would be able to handle that. :D
- Using a sustained 1G of acceleration, the journey to Mars could be completed in about 1 day and 7 hours.
- However if you didn't want to fly by mars, You'd have to Decelerate halfway there. So it would take 2.1 days, or 50.4 hours. or about 2 days and 2 and nearly a half hours.
The guy behind the new flight model for X4 was their (CIG/SC's) main 'flight model' guy/flight combat director for at least for or five years.
The pretext behind this update is that this supposed 'rockstar' dev from CIG would reconfigure X4's flight-model to be super cool and appeal to all the refugees from that failing project (of which I saw many names cheer-leading his appointment).
There's nothing wrong with this - provided we get a good "flight model". This one could use improvement in a few key areas to make it much more robust but at the same time more rewarding.
Actually its absurdly achievable. We can use interpolation, extrapolation and dead reckoning to efficiently assume a ships trajectory - and its not nearly as performance costly than relying on a series of constant input changes to predict/extrapolate vector/location.
For example, we are not moving and we inject thrust to bring us to 50m/s forward. The engine can extrapolate "Until I receive another input, targets vector is 0/360 at 50m/s and thus will be at xyz * deltatime."
Obviously it's a little bit more complicated - but linear projections are pretty efficient in this context. Your autopilot already does this in many respects.
Besides, using xFA/decoupled provides this linear extrapolation specifically because you have a mode that lets you travel without any additional inputs. The thing about 6DoF movement is that it's 6DoF and therefore requirements measurement and indication of vectors in 3d space.
But in such an environment much of what's going is relative. We don't want to increase or decrease speed to an absolute max or min, we want to adjust it in relation to things, even if those things are ourselves. It would be much easier on the pilot, if rather than injecting 60% total thrust, we were able to adjust various axis to treat inputs as relative, whilst otherwise maintaining our vectors.
For example, in Star Citizen, you could cruise at some limited speed using their decoupled mode, but to engage any kind of pitch or yaw would automatically engage an auto-braking mode where the flight system sought to return the ship to a zero energy state (no momentum relative to nearest moon or planet or whatever) - which doesn't make a lot of sense.
To conclude, the use of techniques of dead reckoning and extrapolation work much better with consistent speeds and vectors; versus the consistent manual/constant input often required.
Yeah it can, more so than manual flight over the same distance.
Because your physics calculations can dead reckon faster than they can manually configure for constant manual input.
For example at t0 you add enough thrust to bring you to 100m/s and at t1 your v is constant at 100m/s
However manually, at t0 you might have 99m/s, then t1 you have 101ms, due to variation in inputs from the user. Even assuming only 1% variance on average, compared to a realistic spaceflight injection which might just be t=10 at 100m/s compared to that slight variation from manual flight, would be far less computationally expensive.
But this is more of a Starwars kind of game and less a moon landing simulator.
* Flight model improved, successfully. Previously flying was a chore with predictable and boring combat and ships were moving in straight lines. Now flying is fun. Did you know boost used to drain ship shields? They finally removed this nonsense. So the game did get a good model.
* There was no goal to attract SC refugees.
* Everything you listed is not really related to high velocity problems in high speed engines.
The issue is that high speed objects require special handling, and that quickly results in escalated number of calculations, because even with logn broadphase algorithm, high speed movement causes number of queries to shoot up. Basically, if you're flying at 30 fps at half lightspeed, that's a 4500 km cube, with all stuff within to check. EVEN with optimizations. Add to that the issue where objects do not move into straight lines, that becomes even more fun.
There is no free lunch. If it was absurdly achievable, everybody would've been doing that. But that's not the case. Which means it is not absurdly achievable. Games that have high speeds often feature small number of objects as a result. Like in ED you maybe have 30 ships that actually exist in the same bubble as you.
Why do you also bring up high speed? I'm not talking about high speed, but rather 'orbital' or docking manuevers at relatively modest speeds. For me, cruising isn't about achieving some max high game engine speed, but simply achieving the desired course and heading then being able to either let the ship fly on its own momentum/inertia; or rather than try and target my throttle's position, simply target its own velocity - so I can make minor adjustments whilst travelling under my own energy.
Vector math isn't that expensive - think about it: The game is doing vector-calcs every time you walk around and use mouselook. This argument about supossedly the engine breaking down under some arbitrary numbers doesn't feel correct to me. These are not Mersennes Primes - what we are calculating is fairly modest, and only becomes more modest when objects are moving on consistent trajectories.
You can probably cut down on a lot of AI and ship-brain overhead by letting them simply drift between way points before 'injecting' a new trajectory via thruster burning. Obviously, in astro-combat manuevering, you are not going to be moving so sedately; so these arguments about combat seem inapplicable.
This is more about the ability to use unassisted flight to manage your flight path more efficiently by using simply Newtonian mechanics, without having to engage in any funny artificial acceleration/deceleration business.
* "Friction in space" exists to cap high speed.
* Orbital mechanics are outside of this game's scope.
* Collision detection and high velocities have implementation costs.
Your ideas are not correct.
One NPC on station, just being displayed is in ballpark of of 30 matrix operations every frame, just to update the skeleton.36 multiplications 48 aditions, per matrix, 30 bones for a basic skeleton with fingers. About 2500 floating point operations?
Ship collision is a calculation between multiple arbitrary convex hulls. Normally games use discrete collision, at high velocities that results in tunneling, and that complicates things and needs special handling. "Given two convex hulls of arbitrary configuration, moving from starting point at corresponding direction, determine whether they collide, and if they do, at which point of time".
Walking and looking around have nearly zero cost.
If you doubt my point copy-paste my response into a chatbot that supports search and reasoning, and ask it to verify. Read the links.
i can defeat every corvette and frigatte with simple s ship like takoba or mamba.
and im a very bad player...
edit:
i couldnt do this before patch 7.5
fights are crazy good now, cant get enough...
Now this new model has some problems of course, but it's a cosmically large step in the right direction. As for problems to solve or things I don't like in the new flight model...
-Independent Velocity cap on strafe with FA/ON needs to be removed, it feels bad. You should be able to strafe up to your max listed velocity with FA/ON just as you can with FA/OFF.
-Janky and non complete vector transition towards your thrust direction if it's within 90 degrees of your vector while at max drift velocity with FA/OFF. Both space engineers and star citizen do this correctly, this game does not.
-Optional absolute velocity read out instead of angular or whatever is currently being used.
-Optional "coming from" vector indicator
-Option to turn off adjustable throttle state with FA/ON. Think space engineers with dampeners on behavior, or star citizen with FA/ON and no throttle state set.
With your setup and interest in (almost)real-space-physics, you should give "Evochron Legacy SE" a try. Setting-up your input-devices might be a pita, but it's worth the effort imo.
Here there is (allmost) no speed-limit in any direction if you disable the damping-thingy.
Give it up. You'll not succeed here, I can promise you that.
I had a similar conversation with this guy recently; he's convinced that X4 is realistic, and that it uses real Newtonian mechanics to calculate. In some way this appears to be connected to speeds being capped in X4 to prevent the engine from failing.
Firstly I'm minded that computers handle orbital mechanics in RL no problem, and secondly that they managed to land a man on the moon with the equivalent processing power of a ZX81 (assuming you believe that kind of thing, lol).
But sure, that's RL and not game engines; who actually knows how this game engine works?
In which case I'm also minded of a recent thread of a guy who glitched and was moving at something like 750 km/s. The game engine appeared to be handling this fine, rather the issue was how to stop moving at such a high speed (answer was in editing the save files for certain variables).
I mean if you want to continue then go ahead, but my advice is just stop now. Give up. Preserve your sanity. You'll get precisely nowhere with this ;)
Edit: The above aside, I do agree with your basic point that they could have left the non-flight-assist mechanics as they were before. I haven't played around with it much, but insofar as I can tell they've applied the "space drag" thing to non-flight-assist mechanics, which is unhelpful.
I thought it was written `Nicht gut.`?
I'm doing some writing and trying to get my German right.