[Hades]

He's not talking about a trail moving, he's talking about the actual plume coming from the engine, which would look pretty silly and unnatural if it rotated with the ship, IMO.

[Dragon]

From the physics POV, the plume should act like trail (hot, glowing gases ejected from the engine).
In fact, the plumes you see on modern jets appear only when using afterburners and are composed of burning fuel (which is injected into the engine after the main stage to increase thrust).
And you're right about the fact that it may not look good when being firmly attached to the ship.

[Herra Tohtori]

From the physics POV, the plume should act like trail (hot, glowing gases ejected from the engine).

Except the gases are flowing backwards at velocity v, in which v means "very high".

That means that from any practical standpoint, the plumes can be handled as rigid models that always point to the direction opposite from the thrust vector. If we're dealing with longer trails, then the particles' travel time from the thruster to the point where it doesn't glow any more will of course matter but for short thrusters, it's more or less like a rigid object as far as appearance goes.

[Droid803]

Wouldn't something more "realistic" require some volumetric particle system?
Modelled thrusters don't fix the fact that they stick out in one direction, always.

I mean, particle thrusters look great but LOLPERFORMANCE.

[Nohiki]

@Had--: But because the particles, according to newton's 1st law, tend to remain in straight motion, maneuvering thrusters and thrust vectoring would only turn the deformated cone, not prevent it from deformating, because the vector of speed of the ship would change angle to the speed vector of the particles in the cone anyway, and the difference will grow with distance from the engine, as the cone of particles is not a solid object.

I think the trail-type is so far the best.

[Fury]

There was a debate of sorts between me, Hades, Vasudan Admiral and Zacam on irc about thrusters.

Main points were as follows:
- Adding modeled thrusters back to ships and missiles is according to Zacam, too much work for too little gain.
- Zacam and Vasudan Admiral favor particle thrusters over modeled thruster cones.
- There are problems with particle thruster code that needs to be addressed.

These problems include but are not limited to:
- $Max created seems to either directly, or indirectly affect rendering distance of particles
* The higher value used in $Max created, the further away you can be before rendering of particles stop. But the closer you are, the more particles is rendered.
- Despite of $Max created, you have no real control over at what distance particles are no longer rendered
- When particle rendering stops, it is abrupt
- When particle rendering starts, it is again abrupt
- Missiles have no particle thrusters, only $pspew
- Some ship pofs have clipping issues, some are really serious such as GTD Orion's capital01.pof
- Rendering issues when particle effect is partially, but not fully out of view
- Heavy fps drop the closer you are to particle thrusters when value of $Max create is high(ish), which is needed for particle thrusters to be visible beyond an arm's reach

In addition I thought of something I thought to be clever. In the end it didn't work out, perhaps I messed up somewhere so I'll detail it out here.

Since particle rendering starts and stops abruptly, I came up with an idea of using current thruster bitmaps to smooth out the transition. I stole the design behind missile_flare, which uses mipmaps for its effect. The first mipmap is empty or black, it is practically invisible. The real flare begins from second mipmap and becomes stronger a few mipmaps, then starts fading out. Now, for thruster bitmaps I tried using black or empty mipmap for the first two mipmaps, with actual thruster bitmap starting from third mipmap.

This didn't work out as I had hoped. Thruster bitmaps were rendered even up close, but they were really small and thin. However, mipmaps worked exactly as they were supposed to and the thruster bitmap gradually became stronger the further away you got from the ship. In theory, this still can be used to smooth the transition between particle spew and thruster bitmaps, I just don't know where I went wrong with them.

To illustrate most of these issues mentioned above, I have uploaded a small mod. The mod includes necessary tabling for GTC Fenris and GTD Orion, mipmapped terran thrusters and a demo mission.

[Nohiki]

System looks good, but if this goes into the game, make sure it can be turned off just like the radar icons, because the fenris cut me down from 37 FPS to 23, and the Orion to 7  There are still people who use AGP and have virtual memory next to none 

[Fury]

Please re-read my post. The mod was there to present the PROBLEMS as listed in the post above yours. Even performance issue was listed in the known problems caused by $Max create that needs to be a high value to render the bloody particle thrusters anywhere that is not within arm's reach.

Not to mention the demo uses shockwave01 for particle thruster since there was nothing better suited for it in the mediavps. It's horribly inefficient effect to be used in particle thrusters.

[Galemp]

I'd prefer thruster cones for ships such as the Aten and Thoth, that have linear engine arrays and are thus inadequately served by point-and-radius thrusters. (Would be nice if we could scale thrusters on their X and Y axes relative to model space, instead of an absolute radius, but that's wishful thinking.)