Steam loco smoke too slow?

[scefhwil]

I not convinced that the issue lies with the particles. I'm inclined that it lies with the emitter. Here's what I'm seeing on my screen.

The emitters for the chimney smoke/steam are located correctly; in the middle and level with the top.



When static smoke is OK.



At slow speed movement is starting to show.



At high speed the origin of the stream of smoke has clearly moved to "downstream" of the chimney. Imagine a best-fit line through the centre of the particals in the stream. It no longer passes through the original emitter location.



In the sim particals are actually just viewer facing polygons. The first time a partical is drawn it should surely be drawn at the origin of the emitter with no displacement appied due either to it's own motion or any external forces applied. In all three cases shown above the vertical possition is correct in relation to the emitter position, ie it is at the emitter's height origin suggesting it is being drawn initially without any physics applied. But with speed it starts to displace horizontally. Given that the steam is correctly rising upwards it seems to me that it is inheriting the horizontal speed from the loco, but only 99.99% when it should be 100%.

Stuart

[Sly401]

Hi Stuart

As pointed out earlier this has been tested and shows that particles have zero intial velocity inherited from the loco.

Check out the sv set up on the 9f I have added emitters with a Y velocity of 1,2,3,4,---12 mph and each switches on in turn matching the loco speed via lua.

the sv stream follows the loco perfectly up to 12 mph, remaining vertical and not inclining and stringing out as is normal.

Sly

[AndiS]

Just a footnote: When Sly says Y, I guess it would be Z. Anyway, he is talking about longitudinal velocity, speed in the direction of the movement of the engine. Y seems to be up, by my limited reading of default XML files.

[scefhwil]

Hi Sly,

As I don't own a copy of your 9F I can't check out what your've done, but I understand from your description and don't doubt it hasn't proved a solution to the problem you are trying to solve. You have a separate SV emitter for different speeds each set with a forward Z velocity equal to the speed to counter-act the forward motion. But what happens when the loco travels in reverse? Doesn't this mean that instead of counter-acting the displacement the situation is compounded by double the displacement? Not really a problem with sv effects I guess as travelling in reverse and the sv blowing is a small occurance , but not so good for the main smoke effect.

Any how, I found something quite significant today. By adjusting the AA and AF of my graphics card I can significantly alter the fps achieved in RW. Guess what, the amount of displacement achieved (as in the forth screen shot I posted earlier) varies with the fps. With fps in the 100's the smoke was just across the back edge of the chimney. With fps in the 20's the smoke was some 1m behind the stack. This was the case with the loco travelling at about 18m/s. Simple bit of maths and it becomes apparant that the displacement is the single frame travel distance. Nice one RW.com!

Stuart

[Sly401]

Hi Stuart

Yep the Z velocity ( wrongly referred to as Y in previous posts )

The 9F SV hack was to prove the point that Particle effects would be greatly enhanced if they inherited the vehicles forward motion on release ( this then decays as per physX) .

Yes the hacks steps are each limited to a set speed and direction...
the idea is that the z parameter should be made to equal absolute speed controller ( which can be negative) , obviously I can't do this until either its hard coded or a call is made available to us in Lua.
But I could simulate its effects

Sly

[scefhwil]

Just to wrap this issue up from my perspective. The problem I had (smoke moving to behind the chimney as speed increases, but displacement related to fps) has been overcome by a bit of lua scripting. This thread gave me the idea as it turns out that a particle effect child can also be moved using Call("setNearPosition", x, y, z).

http://forums.uktrainsim.com/viewtopic. ... 4&t=108319

Stuart

[AndiS]

Incredible, how simple things can be once someone puts one and one together!

That's what I said since the release of KRS: Make the thing modular, give us some docu, and we will get it further.