Australian steam locos

[Altoona+BeachCreek]

Wow, how stupid. It's not like you pay for the site or anything I'd be pretty mad about that.

[Altoona+BeachCreek]

I just downloaded the double car mad, and it has taken my breath away! It's truly awesome behind twin N&W SD40-2's and Y and A class mallets. I was thinking, though, wouldn't you have to be able to make a triple-sectioned car to make a Garret? And, if we can manipulate engines like that... couldn't be set up one diesel to look like 2 without using the tender file? That would allow us to then copy-paste that already double diesel into the tender, thus giving us a quad-header! Or, we could add a normal single engine to the tender file and make a triple header. Or, could we merge the tender of a steamer into the head engine file and then copy that into the tender file? Maybe then we could finally have a steam double header!!! If there's anything I could do to help, Gumboots, anything at all, just let me know, and I'll see what I can do. Also, is WPandP still active on this forum? I don't remember him posting anything for a while. Anyone else notice how, whenever something awesome is about to happen on the internet, it just kinda... dies? Don't get me wrong, though, because WPandP gave us a LOT of great things already! It would be nice to have double hopper cars, too, though, or any other car. I really want to see these projects come through, and I hope Ned can find the time to, at the very, very least, share the files for that glorious Soviet P36 Northern, which would instantly become my favorite engine in the game! I think you all deserve an applause for all the hard work you've done!

[Hawk]

It's not like you pay for the site or anything

What do you mean I don't pay for the site. It runs me about $350.00 a year to keep this site going. That's why there's the option for folks using this site to donate.

There's no reason to get mad. Stuff happens.
I used to be in the construction industry and I know how accidents like that can happen.
Either the guys digging didn't contact the underground locators or the locators didn't find the cable. It happens.
Last time they cut the connection from that data center to most of Kansas and some of Colorado. I don't know how extensive it was this time.

[Altoona+BeachCreek]

It's not like you pay for the site or anything

What do you mean I don't pay for the site. It runs me about $350.00 a year to keep this site going. That's why there's the option for folks using this site to donate.

There's no reason to get mad. Stuff happens.
I used to be in the construction industry and I know how accidents like that can happen.
Either the guys digging didn't contact the underground locators or the locators didn't find the cable. It happens.
Last time they cut the connection from that data center to most of Kansas and some of Colorado. I don't know how extensive it was this time.

That was sarcasm, I'm not sure if you picked up on that. That's why talking on the internet can get annoying sometimes... I was just joking that I sure wouldn't appreciate losing service that I pay for to an error like that ;)

[Hawk]

That was sarcasm, I'm not sure if you picked up on that. That's why talking on the internet can get annoying sometimes... I was just joking that I sure wouldn't appreciate losing service that I pay for to an error like that ;)

Ah! I see now.
Sometimes the written word doesn't come out quite as planned.

[Gumboots]

I was thinking, though, wouldn't you have to be able to make a triple-sectioned car to make a Garret?

Been thinking about how to tackle it. I haven't got into the actual coding and game testing yet, but will make a start soon. Yes, basically a triple section car is one way to do it.That will work, since the cars behind it are standard single section, so they wont cause the "wraparound" problem shown in my diagram.

The wraparound problem is caused by the fact that by introducing extra segments, you are effectively making the train a closer approximation to a circle. A circle has a shorter perimeter that any polygon which shares the same overall dimensions. So, given that the pivot points between the game's "single cars" have to be coded to have the correct length in straight line, the more segments in each "car" that you try to wrap around a curved track, the more overlap problems you will get. This is basic geometry and we're stuck with it. So, it comes down to what you can get away with before it starts looking too silly too often while playing the game. Like I said, I haven't played around with it yet in live testing yet, so I don't know how far the concept can be taken.

The problem with doing it this way for a Garratt loco will be that the first car of the load will want to attach to the length point of the loco body, which will be outside the track on a tight curve. So, it may tend to look as if the car is attaching to somewhere off to the outside of the loco, given the rear truck of the loco will be more or less following the curve. Also, on a tight curve the central body of a Garratt really should be slightly inside the curve, but we can't model that in RRT3. I'll try this "one body, two trucks" idea first though, since it's the simplest option.

Another possibility would be to model the Garratt loco as one truck (for the front unit) plus loco body (required anyway) for the central unit, with the "tender" making up the rear unit. That should also work, and would probably give a better looking connection between rear unit of the loco and first car of the load. However, it would rule out double heading Garratts by cheating with the tender slot for the second loco. OTOH, I'm really not sure I care about double heading Garratts in the game, even though in real life it was done sometimes.

This method would also introduce another glitch, in that the connection between tender and loco tends to be a bit rubbery. If you watch locos and tenders you will often seen the connection between them jerking around a bit as the train moves over lumps and curves. This looks kinda natural with individual tenders and cars, since trains do behave like this IRL. It might look decidedly weird if the rear end of a Garratt was doing it relative to the central unit though. Again, will have to try it and see.

Then there's another option, of using an invisible "body" that is pared down to just the required attachment point to the track, with front unit, central unit and rear unit all modelled as "trucks". That should still work, and may be useful for getting a better approximation of real Garratt geometry around curves, if it is necessary. My instinct is that this would be more trouble than it is worth though.

And, if we can manipulate engines like that... couldn't be set up one diesel to look like 2 without using the tender file? That would allow us to then copy-paste that already double diesel into the tender, thus giving us a quad-header! Or, we could add a normal single engine to the tender file and make a triple header. Or, could we merge the tender of a steamer into the head engine file and then copy that into the tender file? Maybe then we could finally have a steam double header!!!

Yes, all of these are theoretically possible, bearing in mind the caveats about the geometry around curves. You could, in theory, have ten steam engines and tenders all in a row, with a hundred load cars behind them. It should look perfectly fine on straight and level track. The less straight and level the track is, the worse it's likely to look.

The thing is that each segment within each car will have to be long enough to look like a real car. This means that for every segment you add, the distance between the game's single car ends get longer. This will only exacerbate the geometry problems around curves, because from a geometrical point of view increasing car length is the same as decreasing corner radius, which has the effect of increasing the angle between cars, which is the root cause of the overlap problems (short version: as long as the cosine of the angle between the game's single cars is close enough to 1, it'll look fine. The more the cosine diverges from 1, and the more segments there are, the worse the overlap will be).

If there's anything I could do to help, Gumboots, anything at all, just let me know, and I'll see what I can do. Also, is WPandP still active on this forum? I don't remember him posting anything for a while. Anyone else notice how, whenever something awesome is about to happen on the internet, it just kinda... dies? Don't get me wrong, though, because WPandP gave us a LOT of great things already! It would be nice to have double hopper cars, too, though, or any other car.

If you really want double hopper cars, learn to make them. With the info WPandP left behind, it should be possisble to convert his double furniture cars into any other sort. Once I get a Garratt running I'll put it up for beta testing.

[arop]

Message to Gumboats: At the moment I work on a v1.06 Tasmanian Map, and it would be very nice if later I could add a Beyer-Garath to this scenario as soon it becomes avaiable.

[Gumboots]

Well it's going to take a bit of time to get the whole thing finished, but I'll make a start on the basic coding this week (wheels, chassis pivots, stuff like that).

I'm not planning on trying to patch it up out of bits of existing locos. A Garratt is just too different from anything the game already has in the locomotive files. I thought about it and decided it would be less work to code the whole thing from the ground up. At least that way I don't have to figure out which points are which in an existing .3dp file. I can just write my own as I need them, with the correct coordinates, and can document them as I go. I can mock some of it up first in another CAD app and just transfer the coordinates to hex. This has to be easier that trying to work backwards and round in circles in an undocumented file. It should also result in a better model.

I had an idea which may make things easier to track through the hex. The following string of bytes:

00 00 00 00 45 4E 44 47 52 45 55 50 00 00 00 00

will render as this in the text columns at the right side of the editor:

....ENDGROUP....

And this:

00 00 00 00 4E 45 57 47 52 45 55 50 00 00 00 00

will render as this:

....NEWGROUP....

I haven't given it a lot of detailed thought yet. It would have to declare valid values that can be assigned to an unused portion of the .tga image, and that wont conflict with vertices that need to be used by the model, but it should be possible to think up some standard strings of bytes that can be used purely as markers between different parts of the locomotive. So the body could be split down into chassis rails, baseplate, boiler, firebox, cab, etc with standardised markers between them. This should make going back to a file, and trying to figure out *** it all means, a lot easier.

Any Garratts I make will have world availability anyway, so wont need to be added to any scenario as long as it allows world locomotives.

[Altoona+BeachCreek]

That's some complicated stuff there, Gumboots. I do plan to look up and see if I could make double hoppers, but technical stuff has always been daunting to me. I am a year older than when I posted that list of stuff that never got done, after all. If I follow WP&P's instructions as best I can, it should be possible. I wish you the best of luck with this project!

[Gumboots]

Well the way I look at it, none of it is really difficult as such. It's just a bit tedious. So the thing to do is just proceed logically, at a comfortable pace, and take a break when it gets too much. That way I should be able to get through it with my sanity intact, and have a cool loco to play with when I'm finished.

[Gumboots]

K, just to get myself a bit inspired before digging into the hex, I whipped up a basic model of the Class M. Still needs work, but it's off to a good start.

This pic is the original beastie, with 8 cylinders and all, shot in works grey.




And this is the CAD model so far.

Basic_Class_M.png (45.67 KiB) Viewed 6547 times

[Altoona+BeachCreek]

That looks like quite the speedster! So, is there a way to directly use that 3D design while creating the in-game model? Originally I thought building a CAD was all you had to do, which was why I thought it would be so easy when I wrote up the list of ideas a while ago. When I found out about hex editing, I was at a complete loss (and still am.) Will a 3D design help at all?

[Gumboots]

Yes, it'll help to some extent. The thing is that I have the original GA drawings for the loco, which means I can get dimensions off those in Photoshop. I know the old caveat about "Don't scale off the plans", but in this case it's ok. They're in good condition, and some dimensions are noted and are just legible, plus I know the dimensions of some components from other sources. This allows me to check for distortion and apply corrections.

So, I can use dimensions from the plans to build the CAD model. This gives me a good idea of how it all goes together, what will need to be included to make it look convincing, and what I can leave out. I can then use the mesh from the CAD model as the basis for the mesh in the RRT3 .3dp files. It's not as quick and easy as I would like, but it's quite workable and will give a well-proportioned model with a minimum of guesswork. Added a few more details last night.

Basic_Class_M_2.png (54.56 KiB) Viewed 6531 times

The thing about these early Garratts is not only are they quite good looking, but they're very simple. Hardly any crud hanging off them, and front and rear engine units are identical, so doing a good model isn't really all that complex compared to something like an A1 Berkshire, which has all sorts of crud all over the place.

Re speed: IRL this thing had the same top speed as the default PopTop Consolidation (60 mph). Having done a little bit of research into late 19th century Connies, it seems they were generally rated slightly lower in tractive effort than the Class M. They wouldn't have been superheated either (the Class M was) which means that for the same nominal TE you actually get less work out of the steam. So in the game, I reckon this thing should out-grunt a Connie, but not by an awful lot. It should also have better fuel economy, acceleration and reliability (Beyer Peacock were noted for the quality of their work).

Connie stats are:

Cost: $120,000
Annual maintenance: $9,000
Fuel economy: Below Average
Acceleration: Poor
Reliability: Average
Passenger appeal: Acceptable
Top speed: 60
Free weight: 91.0
Pulling power: 5.0

Provisional guesstimates for Class M are:

Cost: $150,000 (it shouldn't be too cheap)
Annual maintenance: $12,000 (it was complex)
Fuel economy: Above Average (sounds about right)
Acceleration: Above Average (it wasn't blazing fast)
Reliability: Above Average, or possibly Good
Passenger appeal: Ultra Cool (Look Ma, it's got eight cylinders!)
Top speed: 60
Free weight: 110.0 ish
Pulling power: 5.5 or 6.0

[Gumboots]

I haven't forgotten about this thing. Gave it some time last night.

The Berkshire project is just something else I wanted to finish, since it's a simpler project so is a good way of getting familiarised with how the modelling works in RRT3. That's largely done now, so the Garratt should move ahead in the next couple of weeks.

It's interesting doing a model from scratch. I'm going to have to do a lot of manual coding, so in the interest of making that easy it makes sense to keep the geometry as simple as possible. This is also best for in-game performance.

A lot of the default PopTop models are remarkably crude in their geometry, yet still manage to look surpsingly good in the game as long as you don't look too closely. This is largely down to gfx details taking the eye away from the geometry, and also to the way the game engine uses the normals to fake the look of rounded surfaces from polygons.

There are still some things which I think are too crude though, at least for my tastes. Boilers and firebox ends are one obvious point. The face count for those is too low in most models, so they look pretty crappy when viewing them from the front. My 2c is that to get a decent looking smokebox it really needs around 16 faces (most are 10 or less). That's enough to get rid of the built-out-of-old-bricks look, and only costs another dozen triangles for gfx. Could go higher (hey, what's two more points and triangles?) but I'm not sure it's necessary. Will try it out with 16 faces and see how it looks.

Cylinders (meaning the ones with pistons in them) would probably be fine with 12 faces. Most of the defaults are built out of hexagons, which may have been ok 10 years ago on crappy little monitors, but doesn't cut the mustard now. Similar comments apply to things like steam domes, chimneys, and cab roofs. For obvious features which can be viewed end on, I think it's worth adding an extra face or two just to get something satisfying.

I'm also going to experiment with image sizes for skins. The largest ones that come with the game are 1024x1024, but I have no idea if this is a hard-coded limit. If the game allows images of any size that would be great, because it would mean you could use the optimum image for each model. A tender is not really worth a whole 1024x1024 by itself, but often chews too much space if it has to fit in an image that size with all the loco bits. Using a 512x512 for a tender would be possible, but would probably require splitting that sides into two panels each. If trying to make one panel do the whole side from a 512, the problem will be that the gfx will be blurry at close range due to the image being stretched by the game engine. A lot of the default models suffer badly from this, and I hate crappy gfx.

Anyway, Garratts are still coming. So is Christmas. Both may arrive at the same time, or not. Just have to see how it goes.

[thietavu]

Counting days... :)

[Altoona+BeachCreek]

This has me interested now: were there any particularly successful passenger Garrats? The logic I had of two Atlantic classes being fast seems to not have quite live p to expectations. Which ones had large drivers? Were there any streamlined Garrats? Which was the fastest?

[Gumboots]

There was only one class of streamlined Garrat built, AFAIK. That was the French 231+132 BT-1 (double Pacific) which was built in 1936+ for the Algerian runs. Drivers were 5'11" and it was clocked at over 80mph. That was the fastest Garratt. Yes I do have some pix and drawings. Yes, I had already thought of that.

Generally, Garratts weren't particularly fast, because they were built to run on really crappy lines. For example, a lot of the eastern African lines were laid with 60lb rail, with 2 feet of unsupported rail between sleepers, and hardly any ballast at all. You don't want to be doing 100 mph on track like that, unless you have a death wish. The Rhodesian Railways 15th class (yes, have drawings of them too) was specifically built for express runs, but even they weren't any faster than the Class M.

Without wanting to rant too much about stuff I haven't done yet, it's obvious that after doing the earliest Garratts it'd make sense to do some later ones. It'd also be cool to continue with the theme of one express and one GP/freighter. So, I had thought that I should do a BT-1 and one of the large African grunters for the 1930's.

Then that brings up non-Garratt streamliners from the same era, some of which were rather cool...................

Oh and there was a proposal to build a streamlined NSWGR C38 Garratt, instead of the Pacific that class ended up being. Never built though, so doesn't really count.

[arop]

Keep on with the good work, I'm longing to see and use the output. By the way: There is one thing missing in RRT3: It would be nice to have a 2-6-2 tank engine to do the minor jobs on short lines or to be used in suburban traffic. We only have the odd and ugly looking 2-6-4 tank with cab forward, made from the Italian class 500

[Gumboots]

I hadn't given any thought to suburban tank engines, but there's the NSWGR C30. A bit unusual in that it's a 4-6-4. Several are preserved, and there is plenty of information available on them. Although old, they were in use until quite late in the steam era, and were specifically designed to handle the fairly steep grades around parts of Sydney.

[Gumboots]

Ok this is officially awesome. Get this: there was a Berkshire tank engine back in 1893!

Yup, Lima wasn't the first company to think of a 2-8-4. Neilson & Co. in England were building 2-8-4's thirty years before Lima got in on the act.

I started doing a bit of idle research on tank engines. Apart from the C30 in NSW, it turns out that the WA railways used several classes at various times. One was a 4-6-4 like the C30, but the 1893 K class was a 2-8-4. Now my guess is that this thing would not have been "superpower" in the Lima sense. It wasn't superheated, and the extra rear wheels would probably have been for carrying the weight of the coal (no tender) rather than for supporting a large firebox. Still it was apparently the most powerful locomotive in WA at the time, since the line it was intended for had some nasty grades.

The steep 1 in 30 gradients over the Darling Scarp presented a major problem to the early railway system, so in 1893/94 the K class 2-8-4T tank engines were introduced for traffic on this Eastern Railway.

So there you go.Someone built "Berkshires" thirty years before Lima, for the same reason: because they wanted a loco that could haul loads up serious grades. They should be called "Darling Scarps" instead of "Berkshires". WA's hills got there first.