[markus_GE]

I´m no real expert in programming, and even less of an expert in physics, but I think, after watching some "loads of smoke" - videos on youtube, I should share what I saw:

Both with stam and Diesel locomotives, the smokestacks are usually vertical. That means, with (what will rarely be done - this is just imagery) full throttle / bad firing at standstill, smoke will be blown out vertically (asuming no windspeed, as this is currently not modelled in OpR).

If the smoke blown out would keep its initial velocity, it would rise up straight into the sky, like a rocket´s smoke colums - no plume would be formed. The plume is the product of the exhaust gases slowing down the farther away from the stack they get. To make space for more smoke, they are pushed to the side, forming the plume.

For the next part it is easier to not imagine smoke as gases, but rather as the different patchs OpR uses to produce diesel smoke currently.

The above paragraphs can e transcribed for these patches as well (this is NOT, what OpR models - I just use the idea of patches / particles to illustrate my thoughts): Patch starts out vertically, locomotive standing still, moving along a vertical line at a specific initial speed. As a matter of air cooling down and other air resistance, the patch loses speed and thereby its volume rises (it expands).

If the locomotive now starts to move, it will still emit patches along a vertical line - no angle will be changed there! Instead, the locomotive will move away below the patch emitted when stopped, and will start the next patch maybe one ft down the track.

This patch will raise along a straight, vertical line too, parallel to that of the first patch. But it´s one foot down the track, and the patch was started out later. As a consequence, the patch emitted earlier will already be higher in the air and back a little.

If you now imagine this being done for multiple patches you will get a great plume, with no need for difficult physics (ias far as I can tell). All that is needed would be a particale system which for every particle (smoke patch) could define the following options:

• vertical path to follwow (can be other than vertical (as defined by the .ENG file), this won´t influence the process)
• speed at which it rises initially and current speed
• rate at which the speed declines
• rate at which the particle expands, forming the typical conical (somewhat...) shape of a smoke plume (at standstill)
• some sort of lifetime to get rid of the particle again at one point

I´ll post a schematic drawing later, as soon as I can assemble one.

Cheers, Markus

[markus_GE]

Got the schematic assembled (yes, the background pic is taken in OpR).

The red and green arrows belong together, as well as the orange and greensih-yellow arrows, indicating movement of the locomotive, which will cause the smoke to move back, relative to the locomotive (relative to the ground, the smokw patch won´t move, of course)

The vertical lines of different color indicate the path the smoke patches move along, with the size of the arrow at the upper end (arrow only means the inverted-V-shaped part at the top) indicating the speed at which the patch moves along.

The small colored areas inside the smoke patches indicate which vertical line and arrow belong to which patch.

Click to watch in full resolution, so you can see the details.



As can be seen from the bottom quarter of the pic, a plume will form, that at speed is bent backwards.

Of ocurse, in the end, a higher somke patch emission rate would have to be used.

The schematic is only drawn for a diesel loco here, as I already had that pic available and did not have to shoot it now. The same pronciple, however, is also applicable to steam locomotives.

Cheers, Markus

PS: As an enhancement, to make this even more realistic, the smokw patches would, especially directly after being emitted, have some momentum in the direction of the train too, which would decrease sa the patches move away from the stack. Thisin the end will definitely provide the "bent plume" shape that cna be observed in real engines.

[CrisGer]

ah good some great discussion going on here. That pic from Trainz is encouraging. I was finally able to approach decent smoke in MSTS by fiddling with the Smoke main ace for years...and with a good engine WAG it can generate very reasonable smoke as per my earliest posts in an thread below on smoke. Well here's hoping we can get closer... i know there are many other more important issues bot for a steam era MSTS and hopefuly OR person the smoke is an essential part of the immersion . thanks for the great thoughts.

[James Ross]

I've committed an improvement to the particle simulation in X1946. This in particular has particles starting at one velocity and decaying to another towards the end of their lifetime, which gives steam especially a better-looking curve. There's still more tweaking to do, and I would like to ship our own steam/smoke textures at some point (which will allow for better control).

[Genma Saotome]

copperpen, on 03 January 2014 - 03:22 AM, said:

The main problem with both steam and diesel exhaust effects is that OR does not allow for the initial vertical component when running. As soon as the locomotive begins to move, the exhaust plume assumes an angle from the point of origin which increases with speed increases. There is a need to include an element of vertical velocity before the plume lays back, and this second component should have a more horizontal movement as well as expansion to allow the cloud to gain volume. There is also a tendency for the point of origin to move backwards as speed increases.

Wouldn't that motion be based on the differences in velocity between exhaust and train? IOW the exhaust moves vertically only for as long as the velocity behind it (roughly put) exceeds th forward velocity created by the moving train.

For example, say you are in the back seat of a car and you toss a tennis ball against a side window. The car moves forward slightly while the ball is in the air towards the window and again while it is in the air on the return so its motion, relative to the moving car vs. a stationary car, is not an exact straight line... close, but not perfectly straight. The faster the velocity of the ball the less distorted is it's path. Now open the window and toss the ball out. The path of the ball from your hand to some point outside the car is the same as it was before but once it starts to leave the window it's path is further distorted by the pressure difference between the moving car and still air -- what you feel when you put your hand outside the window. Increase the velocity of the ball and it will go a bit further out of the window before it's path is greatly distorted... increase it a whole lot (like a bullet fired from a gun) and it might go a very great distance.

And so if the example correctly describes the problem we'd have to know something of the velocity of the exhaust and how quickly that falls off over time in order to calculate how high its vertical motion might be at any given velocity of the train. Is there any way to do that?

[markus_GE]

Genma Saotome, on 18 January 2014 - 11:32 AM, said:

Wouldn't that motion be based on the differences in velocity between exhaust and train? IOW the exhaust moves vertically only for as long as the velocity behind it (roughly put) exceeds th forward velocity created by the moving train.

Dave, see post #13 in this thread: What you´re saying is exactly what I (tried to) describe(d) in this post.:)

Cheers, Markus

[BB25187]

Hi,

Besides any other possible improvements, adding a movement of the particles linked to the wind speed and direction might help to create a more natural effect. For the moment, the wind speed is hard coded in OpenRails. If a connection is made at some point in the future with environment files, smoke will be directly dependent on wind conditions. Its movement will be consistent with that of clouds.
Before the latest changes implemented by James, I made a couple of tests. I read the wind speed (from viewer3D) in the ParticleEmitter.cs file. I then computed a wind-dependant displacement to be added to the movement of particles. Combined with a very simple randomization of the time of emission, the effect was not bad. This was very basic and not mature enough to be submitted. The trajectory of the plume wasn't stuck right above the track though...
Of course this only one small thing amongst other.

Regards

[PA1930]

James Ross, on 18 January 2014 - 08:44 AM, said:

I've committed an improvement to the particle simulation in X1946. This in particular has particles starting at one velocity and decaying to another towards the end of their lifetime, which gives steam especially a better-looking curve. There's still more tweaking to do, and I would like to ship our own steam/smoke textures at some point (which will allow for better control).

James, so far its a superb modification!
I had a loco a bit more tweaked for ORTS's smoke looks and I had a run with it and I was quite satisfied! ;) But I guess there are still indeed quite some tweaking to do. :)

[Genma Saotome]

markus_GE, on 18 January 2014 - 11:54 AM, said:

Dave, see post #13 in this thread: What you´re saying is exactly what I (tried to) describe(d) in this post.;)

Cheers, Markus

Understood.

WRT the graphics tho... those circles of exhaust. I know that was just an example... What we should be seeing is a series of thin slices of smoke particles emitted perpendicular to the exhaust vent with each slice getting it's own calculation...perhaps a slice per frame as the smallest unit of time with the height based on the amount of time in that frame. Each slice gets pushed up n distance by the velocity of the exhaust and as that velocity tapers off is moved (stretched?) laterally due to the train speed until that too isn't much of a factor and then it should be able to bloom fully as it dissipates. I suppose a certain amount of bloom occurs earlier too. But however it is done it should never look like a circle.

[rdamurphy]

James Ross, on 18 January 2014 - 08:44 AM, said:

I've committed an improvement to the particle simulation in X1946. This in particular has particles starting at one velocity and decaying to another towards the end of their lifetime, which gives steam especially a better-looking curve. There's still more tweaking to do, and I would like to ship our own steam/smoke textures at some point (which will allow for better control).

Glad to hear that! I worked on that for quite a while and couldn't get it to work. I'll have to take a look at that and see how you accomplished it...

BTW, there are a number of VERY much better smoke textures available for MSTS that could be replaced making OR smoke look much better as well...

Robert