Austral Bay - activities - AI trains @ 10minute hea

Gentlemen -I have just uploaded a file with activities for AI trains on Austral Bay route. It should appear in the file library as:
ab_act_1.zipThere are two activities, one commencing at Chippewa Falls and the other at Sundermeyer.Both have AI trains running at 10 minute headways (westbound, long and slow for the Chippewa activity; eastbound, short and fast for the Sundermeyer activity).At this frequency the player train encounters an AI train at most, if not all, passing loops.I have included a document with full instructions,including how these AI trains may be incorporated into any activity that you may create.Only default MSTS rolling stock and consists are used. The "Austral Bay" route is, of course, required.Comments are welcome.Best,
Phil Skene

Attachments

http://www.train-sim.com/dcforum/User_files/40c6951751a0c653.jpg

http://www.train-sim.com/dcforum/User_files/40c6953152221adf.jpg

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

RE: Austral Bay - activities - AI trains @ 10minute, brucek, Jun-09-04, 05:14 AM, (1) RE: Austral Bay - activities - AI trains @ 10minute, philskene, Jun-09-04, 06:56 AM, (2) RE: Austral Bay - activities - AI trains @ 10minute, longislandmike, Jun-09-04, 07:35 AM, (3) RE: Austral Bay - activities - AI trains @ 10minute, philskene, Jun-09-04, 06:04 PM, (4) RE: Austral Bay - activities - AI trains @ 10minute, brucek, Jun-10-04, 02:04 AM, (5) RE: Austral Bay - activities - AI trains @ 10minute, longislandmike, Jun-10-04, 06:32 AM, (6) RE: Austral Bay - activities - AI trains @ 10minute, hotmetal2, Jun-11-04, 11:46 AM, (7) RE: Austral Bay - activities - AI trains @ 10minute, niknak, Jun-11-04, 12:36 PM, (8) RE: Austral Bay - activities - AI trains @ 10minute, niknak, Jun-16-04, 03:55 PM, (9) RE: Austral Bay - activities - AI trains @ 10minute, sniper297, Jun-17-04, 00:55 AM, (10) RE: Austral Bay - activities - AI trains @ 10minute, philskene, Jun-17-04, 04:24 AM, (11) RE: Austral Bay - activities - AI trains @ 10minute, niknak, Jun-18-04, 02:15 PM, (12) RE: Austral Bay - activities - AI trains @ 10minute, sniper297, Jun-19-04, 12:31 PM, (13) RE: Austral Bay - activities - AI trains @ 10minute, Colin, Jun-19-04, 12:52 PM, (14) RE: Austral Bay - activities - AI trains @ 10minute, sniper297, Jun-19-04, 04:35 PM, (15) RE: Austral Bay - activities - AI trains @ 10minute, sinclair97, Jun-19-04, 06:24 PM, (16) RE: Austral Bay - activities - AI trains @ 10minute, sniper297, Jun-19-04, 09:21 PM, (17) RE: Austral Bay - activities - AI trains @ 10minute, philskene, Jun-20-04, 00:08 AM, (18) Ooopz, you said the "T" word!, sniper297, Jun-20-04, 01:17 AM, (19) RE: Ooopz, you said the "T" word!, philskene, Jun-20-04, 03:36 AM, (20) RE: Ooopz, you said the, sniper297, Jun-20-04, 06:38 PM, (21) RE: Ooopz, you said the, philskene, Jun-21-04, 00:26 AM, (22) KILL KUJU!!, sniper297, Jun-21-04, 02:19 PM, (23)

Conferences | Topics | Previous Topic | Next Topic

Quote:- Gentlemen - Phil, shouldn't that be "Lady and gentlemen"?I know there's at least one female regularly posts here; she keeps going on about "pumpkins" or some such thing! Bruce.

AustRail

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Hi Bruce -You are absolutely correct. My apologies. I promise to be more politically correct in future.Not being a North American I'm not sure about the "pumpkin" reference - is it something to do about the colour (color?) of locomotives, or about my bedtime?Best,
Phil

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Guys,She likes the BNSF painted engines in that scheme. And she is a fine and very helpfull person! Mike

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Hi Mike -I was only kidding.From what I have read in her posts the lady in question brings a lot of excellent information and considerable wisdom to the site.If it the BNSF colors, I hope she likes my screen shots above. Me - my preference is the classic SF livery.Phil

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Quote:- She likes the BNSF painted engines in that scheme. And she is a fine and very helpfull person!Yes, I realise that, Mike.
My reference was made in gentle humour.Bruce.

AustRail

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

I know you guys were kidding. I was just stating my feelings, not jumping on you guys (there is far to much nastiness around here to begin with). No vileness in me! Sorry you guys thought different.Mike

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Phil,many thanks for the Austral route, I've always like Jim Ws PO&N but find your rebuild has forged a great compromise between shunting and main line work.I really like the Gums, great touch.Does the sun really come up that quickly in Oz?Alan

Hotmetal

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Thanks for the very quick response on the activities Phil. Have d'loaded them and will run them over the weekend.Regards, Nick (UK) (aka Mike.)

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

About halfway through running the first AI traffic activity and it's great! Many thanks Phil. I like the fact that it's possible to run red signals without stopping the game, even if there's an AI train approaching - Very lifelike. Also the frequency of the AI traffic means it's an ideal route to run if you're changing AI sound files etc and want to hear the result without waiting ages for another train to appear. One thing I have noticed though is a few stationary AI trains backed up around Anderson Transfer Yard - Could this be the dreaded AI stack syndrome that Jim experienced on version 8 of his great route? Or it could just be that I've swopped the diesel motive power on the long consist for two BigBoys which has increased the consist length.Regards, Nick (UK)

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

AI traffic is like nitroglycerin, if you breathe on it too hard it blows up in your face. Try opening the original .con file in wordpad, copy the maxvelocity line, and paste it into the new .con file. AI trains develop SFNRS (Stop For No Reason Syndrome) at the slightest excuse, and considerable fiddling is necessary to cure it. Check in the AE VCR display, I've had wierd situations (particularly around Anderson!) where the first 10 AI trains sail thru with no hesitation, then #11 stops and waits for #10 to get 18 signals away before it will move again, with #12 yapping at its heels.

Attachments

http://www.train-sim.com/dcforum/User_files/40d1242128b2a506.jpg

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Nick -Why TWO Big Boys!? If you are using my Big Boy patch, which you should be, just one of the Boys should have enough grunt to haul the 50 cars up the North Face of the Eiger.I fear that you have bumped into the same unforgiving wall that I did when testing AI activities on Austral Bay and which I allude to in the readme file.I think that it's all to do with "complexity" - the more complex the activities/player train/AI trains the greater the task for the Simulation, and at some stage it just throws in the towel and quits by forgetting to clear a signal or set a switch.But as Jim says above, it never ceases to amaze me/us that it works well for numerous repetitions of the AI trains, but then just falls over with a bank up of AI trains at a red signal somewhere on the route.I believe that Her Majesty's Inspecting Officer used to report on Signals Past At Danger (SPAD) on British railways. Perhaps we need a similar acronym for signals that fail to clear in MSTS?Two suggestions:
Either cut the headend power to one loco;
Or reduce the size of the consist, maybe in half or even less.Then try the video player function at 128x speed.If it does not work in the video player it is unlikely to work in the Simulation, but unfortunately the converse is not always true.Good luck,
Phil(Postscript - can we take it that you are experimenting with AI steam train sounds?)

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Hi Phil. Well, I guess the two BigBoys are for looks/sounds rather than pulling power! Anyway I think I read somewhere that AI trains disregard normal physics rules and will just run irrespective of the actual engine power available - Can you confirm? So I'll try with one loco and then cut down on the long consist if no change.Regarding the sounds-I'm just doing basic stuff like adding compressor sounds to Stuart's 2 cylinder sound which I use. Nothing ground-breaking I'm afraid. btw I've added 2D figures to various cabooses, I'm now trying to add figures like the orange/grey/blue workmen you have on the route to engine cabs using shape manager but when they appear in the sim the heads are missing!! Interesting sort of "Headless Ghost Train" effect but not really what I was looking for!!........Regards, Nick (UK)

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

"Anyway I think I read somewhere that AI trains disregard normal physics rules and will just run irrespective of the actual engine power available - Can you confirm?" More or less. Try starting the default GP38 with 17 mixed cars at the Johnsburg wye, run up the South Pass. Can't be done, not enough ooomph for that 4.2% grade. Now use the same consist as an AI train, watch it motor on up to the top.Steamers are a different animal, I forget who it was, but somebody once gave me a tip for steamers - create an AI only clone, then find both places where the wheel diameter is defined in the .eng file (two different values, only KUJU knows why), and change those values to 6.28 times the original. That stops the wheelslip and rapidly spinning drivers on AI and helpers, I dunno why. (6.28 probably wasn't pulled from a hat tho, since divided by two it's the short version of pi.)On PO&N 9 I moved Anderson to get it away from the center of the route, and changed the AI route back to what I had with PO&N 6, where the spawn points are near the bottom and it loops around Anderson/Paludan.Here shown with clockwise traffic.I had smoothly flowing AI traffic in both directions, then KUJUed it up by trying to get some green lights for the player. PO&N 8 was uploaded with a very simple set of activities with westbound (counter clockwise) only traffic, that's because I couldn't get the eastbound (clockwise) traffic to work without SFNRS. Now I have the opposite problem, clockwise works no matter what, counter clockwise goes to hell if I change/add/remove one signal. What's really odd is I'm using the PO&N 8 services/consists, which are nearly identical other than having the cars rearranged. freight1, freight2, freight3, freight1, freight2, freight3, and so on. The first 13 trains (four repetitions of each, plus a fifth freight1) make it all the way thru, then #14 (the fifth repetition of freight2) stops for no reason at various places near Anderson depending on which signals I changed lately. Some variations in the loop around Anderson/Paludan cause #1 to stop dead 3 to 7 junctions past Anderson. It don't matter to the clockwise traffic, whatever I do to the signals in the loop it just flows on smoothly. Why any of that should make any difference I dunno, all the junctions and signal arrangements at those junctions are identical.Most places where I have a short passing siding I eliminate the first signal that triggers the first switch, leaving the "cutoff" signal at the opposite entrance to the doubletrack section to perform that function. Works fine everywhere except the bridge in Saint Marys that goes over the mainline between Port Ogden and Johnsburg, if I leave out the trigger signal it gets SFNRS at that junction (again only counterclockwise, don't affect clockwise traffic).Point to all this, if you get SFNRS, sometimes just adding another spacer signal for the AI trains will cure it. Sometimes you gotta add a lot of them.

Are we having fun yet?

Attachments

http://www.train-sim.com/dcforum/User_files/40d478742636376c.jpg

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

So PO&N 9 is going to have AI traffic going in both directions. I haven't seen that before. I hope we don't have to wait too long for "9".
Colin

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

"AI traffic going in both directions", no, not at the same time. What I mean is the option of traffic in either direction, like I had in PO&N 5 and 6, where you could drive from Fowlis to Sundermeyer going against the traffic, and clone the same activity so that you're running with the traffic. Makes for more variety, the exact same activity can be a lot different depending on which way the traffic is going.I did experiment with actual two way (opposite directions at the same time) AI traffic when I first started adding signals, but then you get into conflicting paths and serious complication, and everything comes to a dead stop. The westbound is waiting for the eastbound, and the eastbound ain't gonna move until the westbound passes, might as well just place loose consists for all the movement you're gonna get out of them. Bad enough when they're just following each other on the same path and stop for no apparent reason.As for how long til V9 is ready, I'm (knock on wood) finally at the point where I've added enough signals to the long bypass sections to get that balky #14 to keep moving, so it looks promising. Why the hell it has no problems in one direction I have no idea, but I've added about 30 extra spacer signals for the CCW traffic just to get that one blasted train to keep moving. It's the wierdest thing to watch in the VCR, I can see the signals changing from red to green as each of the previous 13 trains approach a junction, then #14 gets a red and stops, gets a green 2 minutes later and starts again, then stops momentarily for another red several miles beyond. Meantime here comes #15, closing the gap and having to wait at red signals because #14 hasn't cleared yet. I could live with it since it don't get far enough behind to affect #16, and 16 to 37 are all on schedule. Problem is the nitroglycerin factor, I hate to leave in little bugs that could blow up unexpectedly when something else changes.Anyway, probably a few more days (toss salt over left shoulder Luck be a Lady tonight) to finish signals, another week or so to test the activity templates (if nothing goes wrong with those!), then I got at least another month or two cleaning up and finishing scenery. So probably sometime in the Fall, unless TMTS comes out before that and causes me to flush the whole mess down the commode where it shoulda been in the first place.

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Jim,I have my hands full at the moment making another tutorial, but as soon as I have finished I will try out both yours and Phil's route. In the meantime, though, some questions:1. You said that by providing more spacer signals resolved the issue of an AI train stopping; therefore, would it not resolve the AI stalling problem by simply adding more spacer signals until all recalcitrant trains were doing as they were told?The second question does not relate directly to your route's problems, but is still, none the less, intriguing:2. I have modified the Pro-train4 route by taking out the first half of the route's speed posts and signals after which, I increased the speed limit of the route to 2000mph with corresponding changes in both the consist and trainset folders of the trains concerned. I then observed two interesting things from this arrangement: First, when the AI trains reached the second half of the route, which was unaltered in both the speed and signals, they continued to go, after stopping and starting, at up to speeds approaching 2000 mph right up until the very end of the route.Second, that within the first half of the route that had no speed posts or signals, I noticed that the AI trains STILL behaved as if their were signals i.e. I could not get a faster AI to crash into a slower one .In addition to all of the above, I also noticed that AI trains would either be servile to the superior speed of the player train by slowing right down to speeds of below 500 mph, or if the player train itself slowed down to around mach1 or lower for any period of time(plus or minus 50 mph depending on what part of the route you were on), then the AI trains would sneak up on you and streak past you.My conclusion at present on these points is that there appear to be pre written rules that one cannot change.Mike

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Maybe the rules can't be changed by the user, but I suspect they change themselves according to the phase of the moon.Lessee, "would it not resolve the AI stalling problem by simply adding more spacer signals until all recalcitrant trains were doing as they were told?"I dunno, I had to go and open my mouth about no problems with the clockwise traffic. Since that last post, I got the counterclockwise traffic running smoothly, then backed up the route (my Mama didn't raise no dummies) and tried to add some signals on the crossovers in Anderson yard so that the switches at both ends wouldn't throw at the same time. Didn't affect the CCW traffic, since I added a ridiculous number of spacer signals that runs smooth either way. But the clockwise traffic now lets 4 trains thru, and #5 stops dead after running thru both crossovers on the south track, around the loop, and hits the west crossover on the north track, stops dead there and the signal just stays red. Added a like number of signals and more on the bypass loops facing the other way, but haven't cured that problem yet. Makes absolutely no sense whatsoever.I'll probably restore the backup if I can't get it working tonight, it would be nice to space out the throws of the crossover switches to reduce the chances of the player getting caught in a trap between switches, but the number of signals might reach infinity as far as I know. I notice Phil isn't tossing in his two cents worth, he's too busy laughing at me while I do all the work. As for #2 (2000mph are you nuts?!), if you drive around with the track monitor open, you'll find that every time you drive thru a green signal (or flashing yellow if it don't have a reduced speed defined in the script), that's the equivelant of "resume speed", the track speed goes back to the default set in the route properties."I noticed that the AI trains STILL behaved as if their were signals i.e. I could not get a faster AI to crash into a slower one ."Yup, on a route with no signals at all, an AI train following the same path as the first one will stop at the first switch set against it until the first AI train reaches the end of its path or passes a signal. That's because AI trains don't actually respond to signals, they just follow the path, and the path is advanced or stopped from advancing by (1) red signal, (2) switch set against, (3) conflicting path, AI or player, that hasn't been run yet.You want AI trains to run into each other, try mixing permissive signals in there.

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Sniper -"I notice Phil isn't tossing in his two cents worth, he's too busy laughing at me while I do all the work."Not quite! - I've also got a foot in the Trainz narrow gauge door (with a few notable exceptions, Trainz standard gauge is, in my view, only for Toy Tainz players) and have spent an equally frustrating weekend sorting out steam locomotive sounds. For those interested, though, I have managed to get a reasonable sound scheme and locomotive performance for that lovely Welsh Highlands "Garratt".Back to AI trains. Yep, it is frustrating, irritating, completely irrational, ... . In the end with "Austral Bay" I just had to accept that in some circumstances it worked, in others it did not.I sort of have the feeling that it is something to do with the changes of heading that occur in this complex route. If only "westbound" actually did equal "heading to the west" rather than "now we go south, now north, now east before we go west". It would be interesting to hear if other more linear routes suffer from the same affliction.
Onto another MSTS irritation - BLUE POLES in Track Database Rebuilds.I've discovered that in Trainz the simulation recalculates it base origin every 5 baseboards. It does this because, otherwise, the rounding errors in calculations causes objects to be incorrectly located.I've always had the lurking suspicion that this could also be the cause of the dreaded blue pole in TDR. My guess is that in complex routes the Database Rebuild circulates one way around the route but then, as rounding errors accumulate, finds that when returning to a location visited before but by a different route the ends do not match exactly.If this is so, eliminating alternative routes by temporarily removing track sections could be an answer to the problem.
Just some wild ruminating, of course.
I'm looking forward to seeing your v9, though being used to barren South Australian landscapes I think I prefer the stark simplicity of the earlier versions of PO&N!Best,
Phil

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Heretic. Infidel. Hee-hee, with all the MSTS problems and MSTS 2 being cancelled, no more patches forthcoming, Trainz (oopz, now I said it!) is starting to look better and better, hey? Can you do shunting with the front coupler?Yeah, I accepted some limitations too, PO&N8 wasn't anywhere near where I wanted it to be, but I figured the overwhelming majority never play the activities anyway, just explore route at 2000mph. So I uploaded it as is, but of course I always get the itch to try and make it work the way I want. This latest problem wouldn't be cured by fifteen gazillion spacer signals, so I tossed it in the scuppers and restored the backup that did work. Then I came up with another wierd idea, I can't have home signals or intervals in the crossovers to sequence the switch throws, how can I warn the player that both switches are about to throw? I wrestled with that one two years ago, no way to link thru all the possible paths to another switch or signal miles down the line. So what if the AI train was already visible when the switches throw?Signalnumclearahead 5, place 5 permissive signals from 100-200 feet before the interlocking. As long as you don't run a train shorter than 200 feet (engine and 2 cars), there's no potential for the permissive signals to allow one AI train to rear end another. Wanna cross the AI main, nothing up ahead, quick look over the shoulder;Hmmm, better wait, here comes Crazey Jones.This works from both ends, since you can see the AI train from the junction long before he hits the signal trips. He has 5 greens before the red interlocking, so he don't even slow down.As soon as he hits the trip for #5 both switches throw curved, but by that time the player is well aware of the need to stay out of the crossover.Obviously that only works for these very short crossover-type junctions, with a half mile long single track section, especially around a curve or thru a tunnel, the more traditional 4 spacers plus one 2000 feet or so further back still needs to be used.And Phil, don't get me started on that unholy TDB rebuild and them demon spawn blue poles, you'll never hear the end of it!

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Jim -"Can you do shunting with the front coupler?"
ABSOLUTELY!What you should have also asked:1. Do turntables work?
Yes!2. Does the signalling and AI traffic work?
As far as I can tell.3. Do couplings behave like elastic bands?
No.4. Is there a map view that can be called up as trains are running?
Yes.5. Can you throw any switch any time?
Yes - by a mouse click on it or from the mini-map.6. Can you jump between route construction and running a session?
Yes. 7. Should I try it?
Indeed you should. But only narrow gauge.8. Will it be initially as disappointing and frustrating as my first encounter with MSTS?
Yes it will. Perhaps even more so.Phil

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Ha! Well, #1 I don't really care, if the front couplers work the way they're supposed to there's no need for a turntable. Anyway you talked me into it, next time I got 50 bucks burning a hole in my pocket I'll go over to the Dark Side.Better get back to MSTS before them MSTS vs Trainz jihadists come in here and get this thread locked.Well, as has been my experience since version 6, 3 out of the 4 crossovers worked just fine that way, but the west end of the north track (upper left corner) gave me SFNRS with the clockwise traffic.
AI traffic RED, Player BLUE, crossovers ORANGE.
Wasn't nearly as bad as the previous, I added two signals facing the CW traffic (far upper left), and that cured it.So on to Sundermeyer, where I notice you didn't make any changes to the mainline or crossovers to the yard lead, but there are no player signals? Hee-hee, cross at your own risk, Abandon Hope All Ye Who Enter Here? Anyway, same setup as at Anderson works for these two crossovers (wondered if it would since they're only 10 meters long), but I ran into a snag at the east end of Sundermeyer.Here we have a very short section for the AI between the east crossover and the east mainline merge (GREEN), problem with that is a player approaching Sundermeyer from Van den Hoven. There's one passing siding (BR/AUSSIE "passing loop") between Sundermeyer and Van den Hoven, and when the player hits the west end of that coming to the green single track section, if an eastbound AI train was going thru the east Sundermeyer crossover at the same time, it would throw the switches at both ends of the single track mainline. To sequence that out I added spacer signals at the east end of the shorty track, and even moved them within 10 meters of the interlocking to get the AI train to momentarily slow down, but the best I can get is 16 seconds between switch throws. Possible solutions, some kind of sign to get the player to slow way the *&%$#! down when approaching Sundermeyer on that passing siding, find some way to slow the AI train way down without triggering SFNRS, or put in a disclaimer that really fast reflexes are required to play the activities.Problem with speedposts is they automatically jump across to any parallel track, what I need is a speedpost that affects only the AI track. As sinclair observed tho, AI trains have their own set of rules as to whether they pay any attention to speed limits or not.
Maybe I can hack a "slow the *&%$#! down!" signal for the AI train, where the least restrictive indication will be a 5mph speed limit. Assuming they'll go that slow without SFNRS.

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Jim –I dispensed with the player signals at the Sundermeyer crossovers because I figured that the AI semaphore signals would provide a visual indication of the approach of an AI train. The good thing is that the AI signals, unlike some of the player ones, do their semaphore act perfectly.I still subscribe to the “complexity” theory, and if this is correct then adding yet more signals to a route will just add to the problem of AI signals that fail to clear to green. So it will be interesting to see if all the signals you are adding solve the problem of exacerbate it. If you look at my signalling I’m damn sure that each passing loop has a few less signals that yours.A (politically incorrect) thought occurred to me last night watching a documentary on TV – we really need to get the programmers who did the MSTS code locked away in some dungeon in Iraq with the thumbscrews applied. It might make them talk about the inner workings of the Sim. Might give us all some pleasure too! Another thought, has anybody here ever bothered to ask them that question?Phil

Alert | IP

Printer-friendly page | Edit | Reply | Reply With Quote | Top

Hee-hee, if I got ahold of them boneheads complete with thumbscrews, I think the first question would be "WHAT IN THE NAME OF LUCIFER WERE YOU THINKING WITH THIS OBJECT LINKING AND EMBEDDING NONSENSE?!!" I made a train simulator back in the 80s using VB 3.0 for Windows (Windows 3.1 riding on MSDOS 5 at the time), with page-flipping bitmaps loading and unloading as fast as a 40mhz 80386 could run. Looked like crap, ran like crap, but OLE would have been a lot worse.I totally agree with you on the complexity thing, the less signals the better. For my purposes I need a minimum of 5 at each junction, and that's what I started with. The other additions are to cure other problems, such as SFNRS or the player looking at a red signal for 9 1/2 minutes and getting 30 seconds of green after an AI train passes.I've just been looking at the other option, speed limits. My previous response to sinclair was apparently wrong (oh quit laughing guys, I know my theories are wrong 9 times out of 10, it's the nature of trainsim), the AI trains pay no attention to signal speed limits. I hacked a signal to display 5mph regardless of yellow or green, and when I drive thru it I do indeed get a 5mph limit in the track monitor. AI trains just bucket on thru at full tilt boogie, so apparently they're ignoring the speed restriction made by the signal. Place an actual speed limit marker on the track, I go thru and get the 5mph, next green signal I go thru gives me 70mph again (the route limit). AI train goes thru a speed limit he'll stay at that speed thru ten thousand greenlights, until he hits a resume speed marker.Biggest problem with speed limits is the bug (oops I mean "feature") that automatically stretches the blasted thing across to any parallel tracks, that makes it impossible to control any track individually, especially in yards. Or does it? Latest crazy experiment, delete the parallel player track(s), place speed limit and resume speed markers, then replace the deleted track sections. Can't autostretch to include tracks that ain't there, and as long as I don't fool around with the speedposts after replacing the player tracks it works fine. That opens a whole new idea, if the AI trains can be speeded up and slowed down at will, a lot of signals designed just to space out the timing of the switch throws could be eliminated. An AI train coming onto a single track section with, up ahead:

one facing point home signal before the next diverge
one "cutoff" signal after the diverge
one spacer signal before the next interlocking
interlocking for next mergeThat sequence will throw the first switch at the far end of the next passing track when the AI enters the single track section (interlocking is the trip) and then throws the switch at the far end of the next single track section when it passes the home signal before the diverge. Problem with that is the timing of switch throws is dependent on the length of the current single track section, so if speed limits are placed just before the interlocking, with a resume speed after the cutoff signal, the number values can be altered so that the AI train runs thru long ones at speed, and short ones at reduced speed. If the AI train takes the same amount of time in each case to transit the single track section from the interlocking to the home signal, then the timing of the switch throws would be exactly the same no matter how long or short the distance between the two signals. You could make them run faster in general without hammering the player with redlights, and slow realistically when passing thru or around yard regions.Lotta screwing around to test this idea, got a lot of signals to delete and speedposts to place, but if it works the end result would be worth it.

Alert | IP

[ Home | MSTS Forum | MSTS FAQ | How To... | File Library ] Copyright © 2004 by Nels Anderson. All Rights Reserved.