How RS can't simulate even basic train movements

[MartinvK]

That is what was supposed to have set it apart from the highly limited Train Simulator and Trainz, neither of which are realistic in their operational aspects.

Just to correct you there, the allegedly 'highly limited' Trainz can do everything in your initial post.

Mike.

It still amazes me that no matter how badly they describe some aspect of RS, some people always mange to bring in an unfavorable comparison with Trainz. I think it is the favorite strawman in these cases, whether or not the alleged fault is true or not seems to be besides the point. Just saying it means it must be true.

[stewart]

Just to correct you there, the allegedly 'highly limited' Trainz can do everything in your initial post.

Mike.

I havent had any experience with Trainz since its first incarnation which I found a bit of a dissapointment. Could you tell me which version is capable of the things mentioned in the initial post please? Also is there steam and proper UK routes available now? Sorry for going of topic with this post. I would also be interested to know if the points raised in the initial post will be addressed by RSDL and if it is planned for the features mentioned to be implemented. If Adam or Derek could respond, I would be most grateful.

[PaulH2]

What worries me is that, so far, it seems that scenarios in KRS can't even be as complex (for want of a better word) as in MSTS as specific paths and accurate timing don't seem possible. The scenrio system looks like it can be more flexible, and timetabled scenarios should be very interesting, but I have my doubts that the system will allow complex activities of any type due to many of the reasons listed (not just lack of complex AI manouvers but the static consist issues too).

Only time will tell, but at this point I remain concerned that the limitations in the scenario and signalling systems may be a long term problem with KRS and can only hope that a lot of unfinished code is lurking in the engine to allow more of the features sooner or later.

Paul

[Mike10]

Here's the gory details Stewart.. Sorry it's a long post but it is quite detailed. This covers the whole concept of terminus operations and loco release rather than addressing individual points mentioned in the first post as they serve, to a certain extent, to highlight what can't be done rather than what is desired to be done

The post quoted here was made by Bloodnok on the UKTrainz forum, I'm sure he won't mind me re-posting it here. This should be possible to achieve in TRS2004 or newer versions. There are a few thing that are outlined that need to be setup carefully to make things function reliably, but it is possible to have the AI perform prototypical terminus operations if these 'caveats' are observed.

As some of you will know, I've been building stations recently - and the big terminus is now nearly ready. Just another ~1250 attachment points to place, and it'll be ready to go.

Anyway, as part of this, I've been experimenting with running loco hauled services into terminus stations under AI control, and have them a) cope, and b) manage to depart in a sensible fashion. I concluded that it was impractical to provide enough space for the AI to complete a run-round, as it seems to need at least a complete train length either side.

Investigating what happened for real in various different space-constrained places resulted in the following sequence of events:

A loco will arrive hauling an inbound service. Another loco will come from a stabling point, couple to the stock and depart with the outbound service, leaving the original loco at the buffers. This loco will then depart to the stabling point.

I've now got a sequence of orders that reliably implements this (reliable enough that several express trains are now running on a repeating pattern on my route using these orders).

How the map is set up:

A station with (at least one) terminus platform.

A trigger near the buffers of the terminus platform. About 15m back from the buffers seems to be good - though you may need to tune this.

A trackmark near where the end of train will be in the terminus platform. Accuracy isn't essential here - it seems to work OK within +/-50m or so.

A stabling point, marked with a trackmark. (In my experience, the route from the stabling point to the platform must be unsignalled for this to work reliably).

The sequence of events, as seen by an observer (i.e. coaching stock / service centric) is:

A train arrives at terminus.
The passengers alight (unload).
The loco is uncoupled from the stock. Note: the loco is NOT moved, just uncoupled. You won't actually see any change from this - but doing it first is a requirement (two AIs in the same consist == session go boom).
The 2nd loco arrives from stabling point, and couples to the stock. Note: This must be an unsignalled route to work reliably.
With the 1st loco still half attached, the incoming AI giving the coaches it's typically hard shove doesn't recouple the 1st loco - at least, I've never seen it do that, even with a variety of different traction (and thus enginespecs) in use.
Passengers are let on the train (load), and the service departs.
The original loco is left at buffers, and departs to the stabling point a few minutes later.

That's all very well and good, but we aren't giving orders to the coaching stock, we're giving orders to the individual drivers. And we have to code things very precisely, with very limited communication.

The sequence of events as the loco (i.e. the AI driver) sees it is:

... whatever the incoming service is required to do (e.g. call at intermediate stations)
'Drive to' terminus platform.
'Unload'
(Optional 'Wait for' a few minutes to make it look good)
'UncoupleLoco' (IIRC this order is a built in one now, though may need to be added to the session)
'Wait until minute' (this is to allow the other loco to remove the coaching stock for us, so we set the time to be the other loco's planned service departure time plus a few minutes. The "Wait until minute" order is available on the DS.)
'Go to trackmark' (stabling point)
'Wait for trigger' trigger at buffers, Any Train. (i.e. wait for the next incoming service to arrive - this order is available on the DS)
(Optional 'Wait for' a few minutes - same as, or slightly more than, the previous optional 'wait for')
'Couple at trackmark' trackmark near rear of train (order is available from DS, search under the username 'Marzipan')
'Wait until minute' (service departure time)
'Load'
... whatever the outgoing service is required to do (e.g. call at intermediate stations, disappear in portal)

I use a portal to generate these trains, and sink them into another portal that doesn't regenerate them when they are leaving the map. Once this is running, it will form a repeating (hourly) pattern, with each loco departing with the coaching stock from the following service. If you need more frequency than 1tph, you'll need to set up several incoming trains. I use a 3rd party portal (available from the DS) that allows me to specify more accurately when each train will be generated, and in what order.

On my route, the portal generates a sequence of four express services at 15 minute intervals. Each one has a different departure minute set in the order. This needs careful planning to make sure the loco that has just arrived waits long enough for the previous loco to take the stock away, but not too long so it is still blocking the platform when the next service arrives.
(There is no requirement that all these trains come from the same location, just that they arrive at the terminus at reasonably predictable times.)

Turning a train round in 15 minutes is a hard task - so I alternate these into two platforms. There is a catch in that the loco must return to the same platform it was in before - otherwise the 'load' order doesn't work properly. I've timed the loco moves to and from the stabling point such that only one location is needed for these two platforms, even though there is always two 'floating' locos lurking in the vicinity. Depending on timing, you may find you need two stabling point locations for this.

This sequence won't start to run all by itself though. You need to prime one loco (per platform, if you are using more than one platform) in the stabling point so it can depart with the first set of stock. This has unusual requirements if you are using a passenger interactive platform. It needs to have done a 'drive to' to the platform concerned before it gets to 'load'. The 'couple at trackmark' doesn't count - and you can't do it on the way in (or afterwards) without causing problems. I do this by making the loco leave the depot at the start of the session, drive to the platform, then reverse into the stabling point. This produces quite a realistic shunt move, with the loco just at the far end of the platform - particularly if you give it a 'wait' order of a minute or two for the driver to swap ends.

Orders for this loco are:

'Drive to' terminus platform
'Wait for' 1 minute (timing not critical)
'Drive to trackmark' stabling point
'wait for trigger' (as above - i.e. the first incoming service)
... and the rest of orders as for the outgoing part of the main repeating pattern.

With this loco primed ready for the first service to arrive, the repeating pattern will then start up from the portal generated trains from the 2nd arrival.

This assumes locos that don't need turning - e.g. pretty much all UK outline diesel and electric locos, and quite a few steam tank locos as well. If your chosen traction needs turning before it departs, you'll have to concoct a triangular or semicircular route to your stabling point. Provided you've got the space, this shouldn't be a problem - just use two (or more) 'drive to trackmark' orders to make sure the locos go the right way. (I haven't yet been able to get an AI to stop precisely enough to use a turntable, let alone be able to turn themselves round on it as well.)

Anyway, I hope this is of use to someone out there, I've certainly had a lot of fun experimenting with how to set it up

There's plenty of steam available, and tons of routes on the Download station. You'd have to define 'proper routes' for me to answer more authoratively on that point though.

This may be of interest: http://www.trainzclassics.co.uk


Apologies for hijacking the original thread.

[Lad491]

2) the services that one is able to drive have realistic operations

Realistic operations could, of course, just mean the train starts from point A and runs to point Z stopping at points along the way, to a timetable. if you are driving partway you can leave your train at point F and jump into another train and take it onwards to point A again. That seems pretty realistic in an operational sense to me, albeit very simplified. Im sure that is what was meant.

Anyhow, here I am defending RS when there is really so much else that is wrong with it. Get that lot sort first and maybe a couple of years down the line upgrades will do what you want.

[AdamsRadial]

I reloaded TRS2004 onto my machine during the long wait for RS to arrive, and was just starting to get to grips with the driver commands, using an excellent route called UK Somewhere, and Paul Gausden's O1 engine and SECR stock. The Trainz sounds and steam were poor (in my opinion), but other aspects of the sim were excellent, and it was very enjoyable to watch the O1 haul a train into a loop, uncouple, run round, couple again, and propel the wagons slowly down to the quay, while a saddle tank hauled a batch of empties back to the colliery, and a King Arther or two ran past on the main line.

This doesn't mean that I shall be putting my copy of RS in the launching trap and yelling "Pull!". I can see that I shall end up with all three sims on the machine for different types of experience.

MSTS, which I am still running often, is very much a driving experience, although I have learnt a lot from several activity writers about creating activities where you can just sit and watch the trains go by. TRS2004, now that I have bought a download ticket and built up a collection of stock and routes, is going to be fun for scripting and playing the Fat Controller. RS, well, who knows what we can make it do?

I suspect that in a year from now there will be very few of the disgruntled threads, and lots more talk about the finer points of amending the physics of the rolling stock and pushing RS beyond the current (partly psychological) limits.

But, without people like James to point out the limitations, there wouldn't be the impetus to find ways around the problems, some of us just don't think of doing the things that others want the sim to do, and others don't know the railway ways of yesterday.

I do think, however, that there should be a stray dummy forum where vexed posters can go and let off steam (sorry), after a frustrating bout of trailing point oversights or whatever else it was that got their goat. In this forum alone, they could be allowed to hurl abuse at each other, on the understanding that it is a virtual war-room, and the insults do not carry over to the outside world. I am worried that there will be long-term damage to relationships if the current issues are allowed to grow unchecked, and that some valuable posters are going to turn their backs on these forums and go elsewhere.

[phill70]

I suspect that in a year from now there will be very few of the disgruntled threads, and lots more talk about the finer points of amending the physics of the rolling stock and pushing RS beyond the current (partly psychological) limits.

Roll on

I do think, however, that there should be a stray dummy forum where vexed posters can go and let off steam (sorry), after a frustrating bout of trailing point oversights or whatever else it was that got their goat. In this forum alone, they could be allowed to hurl abuse at each other, on the understanding that it is a virtual war-room, and the insults do not carry over to the outside world. I am worried that there will be long-term damage to relationships if the current issues are allowed to grow unchecked, and that some valuable posters are going to turn their backs on these forums and go elsewhere.

That is a no go, because of the "younger" members.
If anyone see's anything that the think is out of order, hit the report to moderator button, its far quicker that waiting for us to wade through this lot to see the problem.

[AndiS]

I think that the post about trains in the terminus in Trainz shows very nicely why Trainz fans shout that Trainz is great and at the same time urge KRS to become a better alternative. In theory, you can do a lot in Trainz, in practice, it seems very tedious and you do not exactly get were you want to.

Did I understand that you must not have signals at the platform to make the described thing work? KRS would be flamed for such a limitation. Train end must be within 50m of a fixed point? This means to me that Trainz can not do AI shunting, but AI engines have the ability to couple into the unknown world around them. Do the coaches jump the 50m to where the engine waits for them? Or does it slowly approach the coaches where they actually are? Passenger trains in the stations I know are 5 to 10 8-wheel coaches. Most likely you would have to write different scripts for different train lengths, or/and place different stop markers. Not too much different from placing static stock sensors every 200m in KRS.

In my view, what we see currently is the sobering out from a nice dream (which was furthered by Kuju, of course). People are unhappy with MSTS and with Trainz (why else would they bother about KRS?) Everyone hoped for a great new thing, assuming (with some reason) that they will take the best ideas from existing software and implement all that. Now we find out that this is not the wonder we were hoping for. Next step is "let's condemn it", plus/minus praise the next saviour.

What I try for myself is to simply find out how much of a railway simulation I find in this game, with as little emotions as possible. I can only recommend this to anyone. Let's give all the marketing lies a last bitter laugh and try to find a personal answer to the question "is it good enough to make me switch from X" where X is your current favourite(s).

Doing so, I am afraid that we need to start in the middle ground. For the higher grounds, we have some clear answers and mourning about it will not help is one iota. You might still want to do it, I love to fight for truth, even in advertising, which by definition is the opposite of truth, but it is such a boring exercise. So lets start with the assumption that we will not get multi-player, super-elevation, user-friendly and complex timetabled scenarios (seeing the network as a whole), AI shunting; that the terrain grid indeed is 8m and track-based structures ("procedural lofted geometry", e.g., viaducts) are pushed into each other in curves instead of adapting the shape; there is no coupler slack and the nice feature to push wagons without coupling is useless at the moment; and the interactive loading is a joke (constant animation, no change in weight). Monotonous exhaust and wagons stopping dead after being uncoupled is on the issue list, so not all is lost here.

Ok, so what about bankers, single line working, trains reversing at a station, various form of shunt movements, engines running around the train. Each of that to be answered with respect to AI and player train separately and all should be possible with prototypical signalling and dense AI traffic; and if possible without too much of special hacks like modifying rolling stock or limiting the station layout (or rolling stock used) to special cases. How to arrive at prototypical track layout; reliably working turntables; acceptable overhead wiring using procedural lofted geometry (the overhead wire with its poles seen as a fence should span straight between support point). And then the passengers, they might have potential, but we know nothing about them.

While I fully support the quest for the one real simulator, it is pretty clear to me that RSD are the wrong address to send your petitions to. I myself restrict myself to hoping for a patch which fixes some of the serious issues. E.g., since installing the developer tools, signals vanish if you move their track links. This means that you can choose between signals but no scenarios (after reinstalling KRS without the tools), or scenarios with signals only for one direction (the default direction); or firing up Mike's bin editor and hunting down track links, replacing "forward" by "backward", which is a pain as soon as there is more than a handful of signals, and you do not want to have the look all in the same direction. Things like this endanger the future of KRS much more than the absence of intelligent AI movements.

[CaptScarlet]

I think that the post about trains in the terminus in Trainz shows very nicely why Trainz fans shout that Trainz is great and at the same time urge KRS to become a better alternative. In theory, you can do a lot in Trainz, in practice, it seems very tedious and you do not exactly get were you want to.

Did I understand that you must not have signals at the platform to make the described thing work? KRS would be flamed for such a limitation. Train end must be within 50m of a fixed point? This means to me that Trainz can not do AI shunting, but AI engines have the ability to couple into the unknown world around them. Do the coaches jump the 50m to where the engine waits for them? Or does it slowly approach the coaches where they actually are? Passenger trains in the stations I know are 5 to 10 8-wheel coaches. Most likely you would have to write different scripts for different train lengths, or/and place different stop markers. Not too much different from placing static stock sensors every 200m in KRS.

In my view, what we see currently is the sobering out from a nice dream (which was furthered by Kuju, of course). People are unhappy with MSTS and with Trainz (why else would they bother about KRS?) Everyone hoped for a great new thing, assuming (with some reason) that they will take the best ideas from existing software and implement all that. Now we find out that this is not the wonder we were hoping for. Next step is "let's condemn it", plus/minus praise the next saviour.

In regard to Trainz you misread it a little, the buffers in trainz act as a stop signal at the end of the line and you can have signals at the platforms. The Ai will shunt to the consist but what can happen is that the train will go a bit further away than it should from the consist before reversing and slowing to 4mph before coupling up and it doesn't matter how long the consist is for this to happen. As in RS you can help it along with markers and drive to commands before the couple up command. What trainz does allow is the scripting of commands which can be quite powerful if you now how. Though it can get quite complicated to set up.

Yes I was hoping for more of a mixture of Trainz and MSTS and to a point we have that. I hope that RSDL and the community can build upon that to fill in the gaps but its early days as we are still learning mode so to speak.

I like RS even with its limitations and bugs for a number of reasons but I wouldn't dump any other sim just for this at the moment or perhaps even ever. They all have there pluses and minuses as Iam sure MSTS2 will have as well.

As you say If you are always looking for the pot of gold at the end of the rainbow you will always be disappointed when it doesn't turn out the way you hoped. And given how game development works in the real world one could easily be disappointed most of the time if you buy into the marketing hype.

John

[Mike10]

Did I understand that you must not have signals at the platform to make the described thing work? KRS would be flamed for such a limitation.

I said it was possible, I didn't claim it was perfect

Train end must be within 50m of a fixed point? This means to me that Trainz can not do AI shunting, but AI engines have the ability to couple into the unknown world around them. Do the coaches jump the 50m to where the engine waits for them? Or does it slowly approach the coaches where they actually are? Passenger trains in the stations I know are 5 to 10 8-wheel coaches. Most likely you would have to write different scripts for different train lengths, or/and place different stop markers. Not too much different from placing static stock sensors every 200m in KRS.

The trackmark is an initial target point to identify the rough location required, the same as the one in the stabling point, it gives you a fixed location to instruct the train to move to. It does not need to be exact as once your loco reaches that point you will give it further instructions. Once it arrives at the location you tell it to couple to the coaches. The loco knows where they are and slowly moves to them. They could be three miles away, but if you've already moved the loco to within 50 metres or so it makes more sense.

Trust me, this setup works. I've seen it working and other people on the forum have used that post to setup their own implementation.

Trainz can definitely do AI shunting. AdamsRadial describes a shunting move in this very thread, and in another thread Matt describes a session he created for AI controlled banking engines.

Mike.

[jamespetts]

Realistic operations could, of course, just mean the train starts from point A and runs to point Z stopping at points along the way, to a timetable. if you are driving partway you can leave your train at point F and jump into another train and take it onwards to point A again. That seems pretty realistic in an operational sense to me, albeit very simplified. Im sure that is what was meant.

That makes no sense: simplistic is the opposite of realistic.

Anyhow, here I am defending RS when there is really so much else that is wrong with it. Get that lot sort first and maybe a couple of years down the line upgrades will do what you want.

Actually, getting the basics of signalling and operations right are second only to a 3d exporter and stability issues: in no other respect is Rail Simulator as deeply flawed as in signalling and operations.

[jamespetts]

So lets start with the assumption that we will not get multi-player, super-elevation, user-friendly and complex timetabled scenarios (seeing the network as a whole), AI shunting...

For me, those features are the very essence of railway simulation. Without them, the product is totally worthless - if I had have known that they were absent, I should never have taken any interest whatsoever in Rail Simulator, and certainly never purchased it. I am most certainly not alone in that view.

[AndiS]

Well I for one would be content enough to invest time in route building if the graphics engine can dense scenery (which it does far better than MSTS), if there is a way to lay the track prototypically enough for my taste, and if the limitation on operations are not too frustrating for me. No AI trains is unacceptable. No AI shunting is very sad. Timetabled scenarios are important, but once we have secured that (and how) the standard scenarios (and the AI dispatcher's doings) work, then we can come up with some additional programme like Activity Changer. Yes, I would like to have it all included, without paying much. But then again, the reality is that we have what we got in the base package for a moderate price, plus parts of the rest for free (with reportedly microscopic caveats) and now we must see where we get from here. RSD will have to put every £ from the not exactly sky-rocketing sales into a patch, which is not even announced. That's the brutal world of capitalism.

Re. Trainz, I seem to have interpreted it in a rather restricted way. Anyway, this "a lot is possible, but it is very complicated to set up" is what I keep hearing from many sides. Reminds me about MSTS physics. The public perception of a system is often much worse than its dormant powers. However, if something is too esoteric, it is just as good as not there at all. The condition for something to be of any use is that it be used. Therefore, I was very happy to hear about the signal scripting long ago. Dividing the whole system into isolated subsystem with predefined communication channels is the way to go. Unfortunately, signalling seems to be the only subsystem where we have a say. The dispatcher, engine physics, passengers, scenery placement rules, to name a few, are areas where the determined hobbyist can get much further than a game producer on a tight budget (and there are not other game producers). Ups, fell into dreaming mode again ...

[jamespetts]

But then again, the reality is that we have what we got in the base package for a moderate price, plus parts of the rest for free (with reportedly microscopic caveats) and now we must see where we get from here. RSD will have to put every £ from the not exactly sky-rocketing sales into a patch, which is not even announced. That's the brutal world of capitalism.

One cannot blame capitalism: after all, no other economic system would have any prospect of producing computerised railway simulators for entertainment purposes at all. What is to blame is not capitalism, but unprofessionalism (not necessarily of the people who presently work for RSDL, not that that really matters for this analysis). There is a vast untapped market of people interested in railway operations, not simply driving trains around a world that has no more than a superficial appearance of reality, which, if tapped, would be capable of leading to substantial profits.

The reality is that the people who made Rail Simulator have simply failed to make what they set out to make, and have made something instead vastly inferior to the point of being of very limited worth indeed. Worse than that, there is no information about whether the ongoing development by RSDL is realistically capable of, let alone specifically plans to, correct the basic flaws and make the product resemble the thing the creation of which was the raison d'etre of starting the Rail Simulator project in the first place. The fact that implementing the features that would allow the most fundamentally basic of railway operations (as described in the first post) is not even confirmed to be on the agenda for the developers at all is of very great concern indeed.

[growler37]

So lets start with the assumption that we will not get multi-player, super-elevation, user-friendly and complex timetabled scenarios (seeing the network as a whole), AI shunting...

For me, those features are the very essence of railway simulation. Without them, the product is totally worthless - if I had have known that they were absent, I should never have taken any interest whatsoever in Rail Simulator, and certainly never purchased it. I am most certainly not alone in that view.

Hi,
why not just uninstall it!
these threads seem totaly pointless to me,nothing anyone writes here will, make any difference to the majority of people who are enjoying KRS.
surely your best bet would be something like SIG SIM if you like DOS programes with a collection of dots moving around a black background.
Also if as you say KRS is "worthless" why not uninstall it,wipe your hands and move on.
remember that KRS is aimed at the general public as was MSTS TRAINZ and what MSTS2 will be aimed at! because like it or not!
if train sim developers relied on train sim enthusiasts for there sales it would totaly flop,
i think i read somewere that train simmers make up only 1% of total sales.
In short ,the majority of people who will buy KRS dont give a monkeys about accuracy as long as they can drive trains.
Thats why stores in my locality "KRS "is along with" Call of duty" the best seller at the moment, and in my oppinion deserves to be!
with kind regards
kevin