Real Effects - An FX mod for Realism Overhaul[DEV]

[Guest]

Yeah, the previous method seemed far too easy. The huge document on nozzle design I read noted the flow separation phenomenon to be highly unpredictable and hard to predict in practice. BTW, anything assuming 1D flow will be imperfect to a certain degree, since it's also been noted to be very asymmetric. This is why it's problematic, this asymmetric behavior induces huge lateral loads on the nozzle, which the gimbal usually isn't built to handle.

The paper is Apollo-era, so we've likely got better at predicting this stuff since then. I don't know how elusive the precise point of flow separation might be, but it could be hard to precisely find with a 1D model.

[pingopete]

Hey dragon could you get some pics up on the OP?

[ferram4]

Since the asymmetry in the flow is due to turbulence and instabilities in the boundary layer near the separation shock, any model that doesn't include a turbulence model and high resolution of the boundary layer won't predict asymmetrical or transient effects at all. However, the models that I've implemented seem like they'll give good approximate methods for the average sep point; any asymmetries can probably be estimated from videos if we can get ahold of them. The models are from AIAA 2005-3940 if you're interested, which is much more recent.

I've implemented all the ones that directly relate the separation pressure to chamber pressure, back pressure, and Mach number; the other ones require more complicated relationships involving either the nozzle geometry or the pressure directly downstream of the separation shock (as I haven't yet figured how to relate that to the exit pressure).

[sarbian]

I m so happy to see more SmokeScreen use

Feel free to hit me with any idea you have that may help. I did not do much on SmokeScreen for a while but I'll resume dev.

[Guest]

Think of adding the possibility of having the "atmo density" slider (in the testing interface) go past 1 into values you'd find on, say, Eve or deep Jool atmosphere. Currently all the curves stop at Kerbin/Earth SL pressure. Some planets can go much higher than that, there'll likely be a lot more flow separation shenanigans were those to be implemented.

Also, could you add an option to toggle the SmokeScreen button via a config (in the plugin's folder), and move away from Blizzy's toolbar? I prefer the stock one, (SS is the only reason I use the toolbar plugin), but I understand that most users don't need the SmokeScreen interface.

Hey dragon could you get some pics up on the OP?

Will do, but I didn't have much time lately and pics I took so far aren't as good as I'd like them to be. I'm planning on introducing a few more changes in the next version, BTW:

Dim the high-altitude kerolox plumes, make them more transparent.

Increase low-alt hypergolics plume expansion.

Fix the flow separation to be more in line with reality (thanks, Ferram). This will likely result in most engines losing the effect.

Tweak the curves somewhat, especially underexpansion.

[Namolis]

Will this join several exhaust nozzles together as one? Eg if I have 100 nozzles together producing the same amount of thrust and Isp as one large nozzle with more ore less the same surface area, you'd expect the resultant flame to be about the same length.

Also - did the stock game attempt to do this at some (very early) point?

[Guest]

No, this will not join exahaust in any way. I simply can't do that. This is why I'm not eager to start doing the N-1. Beside the fact I'd have to configure it for RO before staring, 30 kerolox engines might very well turn out unflyable from sheer lag. Saturn I is already hard on my GPU with its 8 engines. N1 has that many in its second stage. Needless to say, this is gonna cause huge lags, especially if using Denny's version and not BobCat's (the latter might be more tolerable, I'll be doing them side-by-side).

[JebThrillMaster]

Is it possible that a version of this could be made that works on stock and not just FASA? Also, a KW version would be nice aswell

[Guest]

I'm not really interested in unrealistic engines. Maybe later, but not anytime soon. This is for RO, so my main focus is on engines that look like their real counterparts. Which means Soviet Engine Pack, OLDD Proton and Antares (when I'm done with configs for those two), as well as other realistic engines that get made for RO.

[NathanKell]

On the bright side, Ven's new stock engines are pretty much spot on.

(We're debating simply requiring that, for RO...)

[Guest]

That would be good, but would also mean adjustments to engine types they represent. For example, mainsail would be F-1B and I'd prefer it to stay that way (we could even remove this variant from other parts, it's really different from the original F-1). I'm using Stock Part Revamp and it's good, but has some way to go. If it gets to the point it can be called a realistic mod, I'll do it.

[DennyTX]

No, this will not join exahaust in any way. I simply can't do that. This is why I'm not eager to start doing the N-1. Beside the fact I'd have to configure it for RO before staring, 30 kerolox engines might very well turn out unflyable from sheer lag. Saturn I is already hard on my GPU with its 8 engines. N1 has that many in its second stage. Needless to say, this is gonna cause huge lags, especially if using Denny's version and not BobCat's (the latter might be more tolerable, I'll be doing them side-by-side).

well, i really surprised. when i started modding carreer i thought about lag with models, but when i have tested my huge and ugly first models (seem you'll agree that my models are much heavier than others) i faced no ugly lag. Well, i have about 100 different modes installed, gamedata is about its limit for 32-bit version, but lag occured only when i start 2 N-1 first stages simultaneously. (60 engines) - about 10 fps.

PS: PC is old enuff, has been bought (by details) about 1.5 years ago

[Guest]

That's due to both Squad particles being vastly less demanding than what RealEffects do, and the fact that you're not using SmokeScreen at all (which is much heavier on performance than stock system). Also, Squad's "lagsbane" system does it's part, too. You can't see the lag until it gets really bad, because time slows down to keep it from occurring.

That said, if you'll find motivation to finish LOK and LK, I will do the N1 right away. Even the WIPs you shown are just too awesome not to have RE treatment done.

[pingopete]

So are you not including more persistant smoke? I tried launching yesterday with this, the flames looked more realistically size for the size of engine/SRB but the smoke was still the weird stock stuff :/ If it's due to some people not being able to run with it for CPU/GPU reasons, could you include an option for those who can, I really think it makes a massive difference (or am I missing that's already there?)

One other thing, IIRC, with hotrocket/smokescreen it is possible to add velocity based deflection to flames, so when pitching over hard during accent the flames don't just fly out in a perfectly straight line against the several thousand meter per second cross wind flow?

Cheers again for this, I'm really glad this is being dedicatedly worked on for RSS

[Guest]

I don't think that this kind of velocity based deflection like you want can be done. Besides, exhaust will usually fly a lot faster than the rocket. You don't usually get to a few thousand meters per second (the exhaust velocity for a chemical rocket is between 2000 and 5000 m/s) before you're in space, anyway.

As for smoke, that's weird. I need to double-check, but I didn't use stock smoke at all. You either get no smoke (hydrolox and hypergolics) or a great big trail made of gray smoke (kerolox, as seen in OP). Are you seeing the launchpad smoke effect? If you're not, there must be something wrong.

[pingopete]

I don't think that this kind of velocity based deflection like you want can be done. Besides, exhaust will usually fly a lot faster than the rocket. You don't usually get to a few thousand meters per second (the exhaust velocity for a chemical rocket is between 2000 and 5000 m/s) before you're in space, anyway.

As for smoke, that's weird. I need to double-check, but I didn't use stock smoke at all. You either get no smoke (hydrolox and hypergolics) or a great big trail made of gray smoke (kerolox, as seen in OP). Are you seeing the launchpad smoke effect? If you're not, there must be something wrong.

Ah that's a shame, any how yeah no nice smoke on my end? I got the latest hot rockets, but I assumed the stuff for smoke screen was already included, do I need to grab that individually?

BTW, here are those SSME & SSSRB sounds I started working on. The SSSRB start sound was apparently too long given SRB's obviously go straight to full throttle in ksp, where as the Lqd fuel SSME's can have the engine activated on 0 throttle then once the start sound is half way through throttle up so it merges with the running sound.

http://www./download/i8obh2cff6zxa2o/ShuttleSounds.zip

Edited December 2, 2014 by pingopete

[Guest]

Remove HotRockets from your install. That might be the cause, RealEffects is self-contained, it shouldn't be ran along HotRockets. Also, are you using FASA? Currently, there's no support for anything else. I'll check out your sounds.

[eggrobin]

I don't think that this kind of velocity based deflection like you want can be done.

It can be done, but the issues is speed: we can't have particles flying off at 2 km/s (you'd need a lot of particles).

This may be useful for slow turbopump exhaut however.

In order to do that, set Simulate World Space in Unity on your flame (in other words, make your flame as you would make smoke).

The following illustrates the kind of flame I'm talking about:

Note that for this you'd also want to set some buoyancy, etc.

[pingopete]

Remove HotRockets from your install. That might be the cause, RealEffects is self-contained, it shouldn't be ran along HotRockets. Also, are you using FASA? Currently, there's no support for anything else. I'll check out your sounds.

What about stock parts? I can't really run any major part mods when testing RVE.

[Guest]

Try LoadOnDemand, then. Stock parts... I suppose I could do them, most of them duplicate FASA ones (so no need to recalculate expansion ratios), and Ven's Stock Parts Revamp makes them look somewhat decent. This pack was conceived as an addon for realistic mods, but I suppose some people don't have enough memory for mods.

It can be done, but the issues is speed: we can't have particles flying off at 2 km/s (you'd need a lot of particles).

This may be useful for slow turbopump exhaut however.

In order to do that, set Simulate World Space in Unity on your flame (in other words, make your flame as you would make smoke).

Here's the problem: I'm not making those flames. This is all HotRockets content, reused according to the license. It was planned to be a dependency, but I found it more manageable to ship the effects with the mod. Turbopump exhaust would indeed be nice to have, but I simply don't have this kind of effect (though I do have a rough idea on how to set it up, if I had it).

[eggrobin]

Here's the problem: I'm not making those flames. This is all HotRockets content, reused according to the license. It was planned to be a dependency, but I found it more manageable to ship the effects with the mod. Turbopump exhaust would indeed be nice to have, but I simply don't have this kind of effect (though I do have a rough idea on how to set it up, if I had it).

I'll see what I can do, hopefully we can make that configurable on the SmokeScreen side of things, otherwise it shouldn't be too hard to make a decent enough flame using HotRocket assets (I did that when testing SmokeScreen after all).

I've started trying this out by the way, the effects are indeed pretty!

Looking at photographs, I now remember that I also wanted to see what I could do as far as shock diamonds are concerned...

[Guest]

Shock diamonds generally appear when the plume is underexpanded, so remember to allow simulating that, too. If you manage to implement that, I'll try to have them in HotRockets. Will probably end up working with Ferram on that one, he's the to-go person as far as shockwave behavior is concerned.

BTW, I'm considering reducing the intensity of overexpanded kerolox exhaust. It looks like a giant trail of flame, while pictures clearly show the flame dilute a lot. I'll try to replicate that, though obviously the effect will still be pretty impressive (especially when using an engine cluster). Thoughts?

[eggrobin]

Shock diamonds generally appear when the plume is underexpanded, so remember to allow simulating that, too. If you manage to implement that, I'll try to have them in HotRockets. Will probably end up working with Ferram on that one, he's the to-go person as far as shockwave behavior is concerned.

BTW, I'm considering reducing the intensity of overexpanded kerolox exhaust. It looks like a giant trail of flame, while pictures clearly show the flame dilute a lot. I'll try to replicate that, though obviously the effect will still be pretty impressive (especially when using an engine cluster). Thoughts?

Regarding useWorldSpace: Will be dealt with in #17.

A note regarding smoke: as mentioned above, the HotRockets models predate my neat physical features (buoyancy, drag etc.), and in particular they use damping (which leads to smoke awkwardly accumulating some distance away). Once I make damping removable, it would be preferable to remove it from your smoke effects, and replace it with adequately chosen buoyancy/drag etc. (your cfgs are not on GitHub, so I can't send you a fix easily).

Another remark on smoke: from what I saw in the cfgs (I only quickly glanced at them, this may be incorrect), you seem to use grow (which results in exponential growth) with clamping.

Clamped exponentials may not be the nicest-looking smoke shape we can do, linGrow (linear growth) would probably yield something more realistic.

Regarding that which you call underexpansion, which I'll call overexpansion just to confuse matters further: at the moment you can make any combination of exponential, logarithmic, and linear growth (the combination is by adding the derivatives, which yields strange functions when you use both grow and logGrow, finding out which is left as an exercise to the reader and Mathematica).

Something has to be added to allow for wobbly elliptic arcs in the jet boundary (in addition to the effects for the disks/diamonds themselves), but what should the long-term behaviour be? Exponential decay to zero? Exponential decay to a nonzero asymptote (that would have to be a new kind of 'grow')? Constant width of the flow, just fading?

[Guest]

I use logGrow, mostly. Used grow sometimes for nozzle overexpansion effects, though I'll change it to use logGrow, too, as this is how the gas actually behaves after leaving the nozzle. I'll try linGrow for smoke, currently the smoke used is standard HotRockets fare. I didn't give it much thought, but now that you mention it, I should probably try for something nicer-looking.

As for buoyancy/drag, could you PM or post an example of a fixed config? I don't use GitHub, but I'd like to see how your idea looks like. Of course, you need to make damping removable first.

[thyriel]

Is the FASA Rocketdyne A-7 Engine supposed to have only a small flame and no smoke ?