Video of Ground level detail from MSFS X

[bf124]

One of the MSTS developers (http://yardlimits.spaces.live.com/) references a video on You Tube that shows some of the ground level detail in FSX. Looks to be a pretty decent starting place to build a sim on.

Video: http://www.youtube.com/watch?v=EJW9BbZgyY0

[Poltergeist]

Eh I'll believe it when I see it on the shelf.

[Nels_Anderson]

Well, that really is how good Flight Simulator looks. Hopefully it will look just as good when adapted to trains :-)

[SCL-A-Line]

to get the most out of the new sim if and when it comes out you will most likely need a minimum of 2GB Ram and a 512MB video card in order to get decent frame rates, it will run on older systems but i from what i have seen and heard putting FSX on my system you do need a beefy setup to enjoy it, granted framerates are more important in a flightsim than a railroad sim


if you are having framerate issues with MSTS 1 then it will most likely be worse in MSTS X

[Hobart]

Can't wait to get my mitts on this thing. :)

[WaltN]

Did anyone else notice that all of the ground level shots were of flat terrain? It looked like, well ... Flatland. That video tells us nothing about the adequacy of default FS X terrain for a train simulator.

Walt

[robertreedy]

Walt, if you haven't seen it, you wouldn't believe it .. if you have the power to run it.

We have about a 250 ft high shelf ridge a few miles away from us in a very obscure place and even it is modeled as close as to look like a photograph. Has the trees, vegetation and even the fire tower and small buildings around.

It is a major improvement over MSTS if they can pull it off.

RobertR
ADF-M&RM
Cascade & North Western at http://www.vcnwrr.com/

[tcain1]

>Did anyone else notice that all of the ground level
>shots were of flat terrain? It looked like, well ...
>Flatland. That video tells us nothing about the adequacy of
>default FS X terrain for a train simulator.
>
>Walt

Don't worry elevation won't be a problem! FSX defaults in the US (And whereever else data was available to 38m mesh, and there are already 3rd party mesh which takes it down to 10m. FSX has the capability of handling 1m mesh also. Here's a shot perched on the tracks (Or where they would be, FSX currently represents most main line track as texture. Makes a good template to add detailed routes.) along the Marias Pass. You could also get an idea of the detail in FSX by viewing videos on youtube, just search FSX. Or you can view mine here, which shows detail in city areas!

http://www.youtube.com/profile?user=tf51d




http://forums.flightsim.com/vbts/up1/116592.jpg

[CRQ5508]

with THIS level of detail, can you imagine how low frame-rates will be??? :+ :+

[OTTODAD]

Hi Thomas !

I was watching your Manhattan video on my Internet audio/video links and GOOGLE Earth only WIN-XP Home SP2, the main WIN-98SE Internet computer not being able to download the latest Flash player and during the plane banking to the left at almost 90° the computer re-booted.

It did that also when I tried MSFS-X Pro and in the end had to un-install it !

Could it be that this WIN-XP computer is not up to the job ?
AMD 64-Bit 3000+, 512 Mb RAM and RADEON X300SE PCI-E.

O t t o

[OTTODAD]

MSFS cockpits and fully working composite instrument panels are far more complicated than that and my WIN-XP Sims Only AMD 64 X2 4800+, 1.5 Gb RAM handles them beautifully, having to use it's Mobo ATI 200 Express chip until I get another ATI PCI to replace the one which expired and I can be sure that it can handle MSFS-X !

Observe the frame rates in MSFS-2004, this flight loading almost instantly ! ;-)

Can't create create such composite instrument panels in MSFS-X any more ! :-(

O t t o

http://forums.flightsim.com/vbts/up1/116594.jpg

[WaltN]

All I am saying is that, from my FS X visits to favorite sites, default FS X terrain resolution and textures are not good enough for cab views in a train simulator. I have to admit that your Marias Pass view is pretty good (much of it due to good texturing), but similar terrain in the eastern U.S. anthracite region is not nearly as good. Moreover, I've seen samples of sloppy terrain in FS X. For example, there's a big thousand-foot lump about where Lansing, NY is, and the mountainous Caribbean island of Tortola is there, but about one inch above sea level.

What's sticking in my craw are the many, many rave reviews of FS X scenery in the context of a train simulator, coupled with expectations that maybe, just maybe, they'll do my favorite route. (And run on my PC that I upgraded from Win98 to WinXP.)

For sure, FS X is an improvement over FS9 -- and a significant improvement once you've added a couple of hundred bucks worth of add-on terrain and texture. (The same add-ons or upgrades will improve FS-X scenery.)

I found a notation that I installed MSTS on my old Win98 machine on June 11, 2001. I continue to read forum musings that the new Train Simulator will probably ship in time for Christmas 2007. (I doubt it, boys and girls. Lord, the ink isn't even dry on the specs, judging from the blog feeds. Judging from the blog feeds, they don't yet have any real, live programmers coding away and consuming pizza.) My bet is Christmas 2008, or later. At any rate, that's 7-8 years of game technology evolution, and I expect to see that represented in a new train simulator.

Of course, my expectations are probably unrealistic too. The new train simulator will not feature my favorite routes. (They're all "fallen flags.") I hope the route and rolling stock development tools are sound. (FS X seems to have a good SDK, that actually shipped with the product this time.) I expect that I will have to develop my own favorite routes and that I will have to upgrade terrain to 10-meter resolution and find satellite coverage for texturing. I also hope my old investment (money for add-on routes and sweat for personally developed routes) will be preserved via an import facility for "legacy routes." But, somehow, I have my doubts.

Will I buy it? Yes. Will there be disappointments in spite of this pessimistic view. Yes.

Walt

[WaltN]

Otto,

I have come to the conclusion that you are near-sighted. In a train simulator, you focus on the track. In a flight simulator, you focus on the cockpit.

Walt

[tcain1]

>Hi Thomas !
>
>I was watching your Manhattan video on my Internet audio/video
>links and GOOGLE Earth only WIN-XP Home SP2, the main WIN-98SE
>Internet computer not being able to download the latest Flash
>player and during the plane banking to the left at almost 90°
>the computer re-booted.
>
>It did that also when I tried MSFS-X Pro and in the end had to
>un-install it !
>
>Could it be that this WIN-XP computer is not up to the job ?
>AMD 64-Bit 3000+, 512 Mb RAM and RADEON X300SE PCI-E.
>
>O t t o

Your memory is kind of lite for FSX! Minimum you should have 1GB and it runs best on 2GB.

[tcain1]

>with THIS level of detail, can you imagine how low
>frame-rates will be??? :+ :+

If you notice the yellow number in the top tight hand corner of my shot, that is my framerate. Generally in most areas I get high twenties to 30's + with higher in mountainous areas. The only locations that I drop down in the teens is areas around the high density airports like JFK in NY. Much of that is the high concentration of air traffic, which includes all scheduled flights for most airlines. If I turn AI traffic down, my FPS jumps back up in the 20's. Plus I could lower scenery density or autogen to get more. It will still be more scenery then we're used to in MSTS! Also there is alot of overhead in a flightsim, that just won't be needed in a trainsim, so we should see a boost just from that. For example, a train doesn't require as many gauges to operate, and they are not as complicated. The AI system, would be greatly simplified, since it will no longer have to deal with plotting flightpans for 100's of aircraft. There would also be additional challenges, like implementing a signaling system, tunnels etc... Also MS is releasing a performance patch which should give us even more of a boost. They are targeting atleast a 20% boost in the patch.

[tcain1]

>All I am saying is that, from my FS X visits to favorite
>sites, default FS X terrain resolution and textures are not
>good enough for cab views in a train simulator. I have to
>admit that your Marias Pass view is pretty good (much of it
>due to good texturing), but similar terrain in the eastern
>U.S. anthracite region is not nearly as good. Moreover, I've
>seen samples of sloppy terrain in FS X. For example, there's
>a big thousand-foot lump about where Lansing, NY is, and the
>mountainous Caribbean island of Tortola is there, but about
>one inch above sea level.
>
>What's sticking in my craw are the many, many rave reviews of
>FS X scenery in the context of a train simulator, coupled with
>expectations that maybe, just maybe, they'll do my favorite
>route. (And run on my PC that I upgraded from Win98 to
>WinXP.)
>
>For sure, FS X is an improvement over FS9 -- and a significant
>improvement once you've added a couple of hundred bucks worth
>of add-on terrain and texture. (The same add-ons or upgrades
>will improve FS-X scenery.)
>
>I found a notation that I installed MSTS on my old Win98
>machine on June 11, 2001. I continue to read forum musings
>that the new Train Simulator will probably ship in time for
>Christmas 2007. (I doubt it, boys and girls. Lord, the ink
>isn't even dry on the specs, judging from the blog feeds.
>Judging from the blog feeds, they don't yet have any real,
>live programmers coding away and consuming pizza.) My bet is
>Christmas 2008, or later. At any rate, that's 7-8 years of
>game technology evolution, and I expect to see that
>represented in a new train simulator.
>
>Of course, my expectations are probably unrealistic too. The
>new train simulator will not feature my favorite routes.
>(They're all "fallen flags.") I hope the route and rolling
>stock development tools are sound. (FS X seems to have a good
>SDK, that actually shipped with the product this time.) I
>expect that I will have to develop my own favorite routes and
>that I will have to upgrade terrain to 10-meter resolution and
>find satellite coverage for texturing. I also hope my old
>investment (money for add-on routes and sweat for personally
>developed routes) will be preserved via an import facility for
>"legacy routes." But, somehow, I have my doubts.
>
>Will I buy it? Yes. Will there be disappointments in spite
>of this pessimistic view. Yes.
>
>Walt

The way I envision it is this. Unlike MSTS, FS scenery can be layered. This means that the base scenery it already has can be used as a template to build off of. You would have your detailed routes, which would ship with the sim or developed by 3rd parties already placed in their proper world coordinates. Track (or at least the current representation of it) would continue on the base scenery, making it fairly easy to expand these routes by 3rd parties. (Provided proper tools are provided. To this end FSX already has a pretty good scenery builder included in it's SDK. It also serves as a mission editor, which can be used to add special effects to the scenery based on time, or by action. To get an idea of what can be done in this respect take a look at these videos. They are quite a few, but very informative on just what is possible in FSX in this regard!!

http://fsxmission.com/live/modules.php?name=LearningCenter

[djt]

“Also there is alot of overhead in a flightsim, that just won't be needed in a trainsim, so we should see a boost just from that."

"For example, a train doesn't require as many gauges to operate, and they are not as complicated.”

Really, not if your modeling it accurately.

With a flight sim you are modeling the physics of one vehicle in flight. With say a 150 car train you would be modeling 150 individual vehicles at one time. These 150 vehicles also interact with each other in a very complex way.


“Also MS is releasing a performance patch which should give us even more of a boost.”

That performance patch is nothing compared to what DirectX 10 could bring the “X” platform.

[landnrailroader]

Performance Patch? I assume this is for FS-X?

I would beg to differ on a number of counts. IMHO, a train simulation, done properly, puts a lot more pressure on a processing system & OS than flight sim. Why, unless you are really close to something in FS (and I have all versions up to 2004), there does not need to be much detail. In CFS-1 or CFS-2, if I am in my P-38 closing on a Jap Zero, I don't need to see the rivets in the zero's wings, but I do need to see my own aiming device, and I probably won't be watching the gages very well either. Most of this stuff and the scenery below could be very low resolution without affecting the visual imagery.

Now about TrainSim(s). These are right down on the ground, and the speed is relatively slow. Even 200kph on a TGV is low compared to the P-38 and so a lot of relatively high density stuff can be thrown at the processor quickly. While I don't need as many gages and stuff, I do need a pretty detailed "outside" the windshielf view or the quality of the sim is reduced to nothing but a simple game. Now I cannot see details of cars in my train further back than 8-10 car lengths, but all the cars interact with each other, with the track, and with the locomotive, no matter what the length of the train is, so each cars physics become very important even if it's external appearance can pretty wash out beyond 10-15 car lengths. Now an exception would a tight horseshoe (think the one above "bispo" next to the "California Mens Colony". That sucker is 14 degrees or so and yes, I can easily count the rivets in a car coming into the upper end of the curve as my engine leaves the lower end and heads across the trestle. Consider the 19 degree horsehoes around Coleman in NC on the climb to Ridgecrest, the cars will be pretty close there and missing details would be noticed.

Yes, IMHO, it takes more to do a train simulation.

Jerry Sullivan
42 years in the railroad industry.

[OTTODAD]

I have come to the conclusion that you are near-sighted. In a train simulator, you focus on the track. In a flight simulator, you focus on the cockpit.

I don't know whether you have ever driven a real loco or have flown a real plane as I have done and occasionally still do, being a real Private Pilot, still having 20/20 vision passing the medicals, where you have to focus not only on what the control panels of whatever you are driving or flying tell you but also have so scan intensively what is going on outside at ground level and in the air around you until you have reached your controlled and assumed safe flight level ! But don't ever take air traffic controllers for granted and use your eye balls all the time ! ;-) I almost got had by an F-16 once, it doing low level flying practice in my free air space ! :-( I am sure he did not see me but I saw him !

I am responding to this [/b]With THIS level of detail, can you imagine how low frame-rates will be???

This "Level of Detail" is nothing compared with what I am showing in my picture of my MSFS CESSNA cockpit, the runway ahead of it and the scenery in the distance. And that is MSFS-2004 and not MSFS-X ! ;-)

I wonder what a similar cockpit scene looks like in MSFS-X, waiting on the runway of a major airport with loads of buildings and ground traffic everywhere !

O t t o

[tcain1]

"For example, a train doesn't require as many gauges to operate, and they are not as complicated.”

Really, not if your modeling it accurately.

Care to reconsider this statement :)

http://forums.flightsim.com/vbts/up1/116600.jpg

Seriously, though what I meant about FS gauges being more complicated is in train sim gauges are used to monitor current operating environment ie speed, air pressure, fuel quantity. While FS gauges also do this, some of them such as the PFD and MFD are tied into the navigational system of the autopilot. They monitor the flight according to the planned flight, as well as external navaids like VOR's and ILS's, there can also be a gauge that monitor other aircraft and warns if they get too close, called TCAS (Traffic Collision Avoidance System).

With a flight sim you are modeling the physics of one vehicle in flight. With say a 150 car train you would be modeling 150 individual vehicles at one time. These 150 vehicles also interact with each other in a very complex way.

Except that 1 vehicle can be 400 tons. Many of the elements that make up the physics needed in a train sim exists in the sim in some form, and can be adjusted to work with trains. The main areas that need to be modeled in a trainsim are acceleration based on power, weight, and traction. FS models the first 2, traction would need to be developed. Momentum based on Weight, FS models this, but I think it would need to be refined. Braking based on weight and braking power, while FS models braking, it is currently on off braking. Incremental braking would need to be modeled as well as the bleeding effect of the air brakes. Trailing units, this would be completely new and would need to be properly modeled especially going around curves. FS includes stress failures so that may be able to be modified to identify conditions that would result in a derailment. Reverse, FS does this via reverse thrust, but I think this would need to be refined for TS. The one thing I think the TSX will be able to do that MSTS didn't is model super elevation or tilt mechanisms, using FS Roll capability. Obviously modifications would be needed, and new functionality would be required. but many of the hooks for this exist already.

[tcain1]

>Performance Patch? I assume this is for FS-X?
>
>I would beg to differ on a number of counts. IMHO, a train
>simulation, done properly, puts a lot more pressure on a
>processing system & OS than flight sim. Why, unless you are
>really close to something in FS (and I have all versions up to
>2004), there does not need to be much detail. In CFS-1 or
>CFS-2, if I am in my P-38 closing on a Jap Zero, I don't need
>to see the rivets in the zero's wings, but I do need to see my
>own aiming device, and I probably won't be watching the gages
>very well either. Most of this stuff and the scenery below
>could be very low resolution without affecting the visual
>imagery.
>
>Now about TrainSim(s). These are right down on the ground,
>and the speed is relatively slow. Even 200kph on a TGV is low
>compared to the P-38 and so a lot of relatively high density
>stuff can be thrown at the processor quickly. While I don't
>need as many gages and stuff, I do need a pretty detailed
>"outside" the windshielf view or the quality of the sim is
>reduced to nothing but a simple game. Now I cannot see
>details of cars in my train further back than 8-10 car
>lengths, but all the cars interact with each other, with the
>track, and with the locomotive, no matter what the length of
>the train is, so each cars physics become very important even
>if it's external appearance can pretty wash out beyond 10-15
>car lengths. Now an exception would a tight horseshoe (think
>the one above "bispo" next to the "California Mens Colony".
>That sucker is 14 degrees or so and yes, I can easily count
>the rivets in a car coming into the upper end of the curve as
>my engine leaves the lower end and heads across the trestle.
>Consider the 19 degree horsehoes around Coleman in NC on the
>climb to Ridgecrest, the cars will be pretty close there and
>missing details would be noticed.
>
>Yes, IMHO, it takes more to do a train simulation.
>
>Jerry Sullivan
>42 years in the railroad industry.


Have you flown a cessna or a helicopter? These are just as slow or even slower than some trains. They alo tend to be flown at low altitude. As far as texture resolution, FSX defaults to 1m resolution and can support 7cm res. So Outside detail won't be a problem, also it uses textures and autogen 3D scenery based on the land class for an area. This can either be from data, or you can create your own. The data covers the entire world based on geographic data.

As for modeling and controlling the physics or multiple objects at once, you forget the AI system can control 100's of AI aircraft, (and now boats and cars) for miles around, so I don't think this will be an issue. What will be new is doing it while attached to another. Do to the necessity of modeling high altitude flight, atleast 60 and up to 110 mile radius needs to rendered in some level of detail. This would be cut down in a Transim to a few miles radius. This will probably allow this sim to recover a good amount of performance relative to FSX. The greatest thing about FS is it's versatility! Want to fly low and slow, fly an ultra light or cessna. Want something faster, fly a lear jet. Feel like being a commercial pilot, fly a 747 among all the other airline traffic. Want to be Tom Cruise, hop in your F14, and if that's not enough, if you don't want to fly hop in your boat for a trip down the Hudson, or the Mississippi. Wanted to take that Alaska cruise? Take the QM2 or Island Princess through Glacier Bay! Adding Trains to this is a natural extention!!





http://forums.flightsim.com/vbts/up1/116604.jpg

[djt]

“Many of the elements that make up the physics needed in a train sim exists in the sim in some form, and can be adjusted to work with trains.”

Depends on how accurate you want the physics and how much do you want to model.

Elements from an existing flight sim are not going to create an accurate simulation of in train forces or the air brake system. Just ask the aeronautical engineers that tried years ago to come up with an accurate train simulator.

[djt]

“Adding Trains to this is a natural extention!!”

Yes creating a train simulator from the “X” platform is a natural.

Trying to realistically simulate running a train on the current DirectX 9 / FSX engine may be another story.

FSX is currently almost completely CPU bound. Adding complex train physics is not going help the situation. The cure for this will come in the form of DirectX 10.

[tdragger]

>All I am saying is that, from my FS X visits to favorite
>sites, default FS X terrain resolution and textures are not
>good enough for cab views in a train simulator.

I agree.

mike @ msft

[tcain1]

>>All I am saying is that, from my FS X visits to favorite
>>sites, default FS X terrain resolution and textures are not
>>good enough for cab views in a train simulator.
>
>I agree.
>
>mike @ msft
>

I don't think the problem is as much resolution as it is the type of textures. Textures in FS, especially in urban areas, are made to look like they have buildings, houses, roads from the air. This gives a relatively convincing view from the air, even when no actual 3D objects are present. I agree this approach wouldn't work in a Train Sim, where you would need the actual 3D objects to be believable. Textures that just represent the ground would be needed.

[rdamurphy]

If you had a hard re-boot, I would seriously consider a better CPU fan and a chipset fan. Sounds like a heating/cooling issue.

As any of us old time Flight Sim fans know, EVERY version of Flight Sim and CFS are built for the NEXT generation of computers. It's always been that way.

In Flight Sim 2004, I flew a helicopter (A really nice freeware Blackhawk) and followed the Moffat Tunnel route, one I'm familiar with, from Denver to Glenwood Springs, then came around and followed Tennessee pass through the Royal Gorge. All Visual Flight, no waypoints, and guess what? It's easy to see and follow the "track." The funny thing is that the track is just lines drawn on the terrain, so the Moffat Tunnel went straight up OVER the Continental Divide, and back down the other side. Pretty funny!

Robert

[Erick_Cantu]

No, but modified elements will. The presumptuous assumptions that the 'they can't do it right' crowd has been hinging on are based on a flawed mentality that 'MSTS X' or 'TS2 reloaded' or 'TSX' will be FSX with the stock models replaced with trains over the same exact terrain using the same exact physics.

ACES isn't stupid, and that's not going to happen. The speculation of most of the crowd is positively absurd from the viewpoint of someone who actually develops for the series. I hate to be blunt, but most of the crowd seems to have positively no clue what they're on about.

The exact system is irrelevant. No, there is little doubt in my mind that the new sim will not use the same physics for braking or movement as an aeroplane. Why should it, and why would ACES do it that way? Dealing with the physics of multiple objects in the same space is not a foreign matter in flight simulations. AI aeroplanes do not simply fly on a rail. They are subject to the same physics as player aircraft. As stated above, the main difference is that the mechanics are tied together into a train - a multiplier and feedback loop, and I somehow doubt ACES is going to skip the changes required to cope.

Performance? Do you lot have any idea what the performance overhead required to model and render and give physics to the world at ground level and at every altitude to well over hundreds of thousands of feet is? When people speak of the fewer resources required due to operating at ground level, this is what they are talking about. The difference is huge. This is why a flight simulator can never run as fast as a ground level only game.

I like how people are trying to somhow imply that there's no way it can be done properly because everything about a train is somehow more complicated. The constituent part of a train are simpler in all means than aeroplanes are. Only in numbers can they become complex and in this regard I somehow think the capabilities or a simulator being devised in 2007 are somewhat more capable than those of the benchmark of 2001. The most complex Walschaert valve gear pales in comparison to a 737's flap drive system. The internal workings of a steam engine could never compare to the internal workings of the CFM56 engine, let alone a GE90. The ES44's glorious computers can't hold a candle to an FMS or EICAS.

Get over it.

It is really quite that simple. Do not assume that you know what they are doing or will be doing. Do not assume that the end product will somehow be MSTS with .air files and FSX default terrain. Until a beta or demo is produced, you and I and most everyone else know positively nothing, however, the assumptions being made by some are positively absurd.

[OTTODAD]

The assumptions being made by some are positively absurd.

Absurd they may be, but don't forget that those making them do not have your knowledge on what is involved and will appreciate information about what it takes to create a Train Simulator using parts of the technology used by MSFS-X.

And don't forget that a BOEING-747 taking off at a V2 speed of 144 knots at ground level travels faster than most passenger trains, the scenery and it's ground level objects have to cope with !

O t t o

[djt]

“The speculation of most of the crowd is positively absurd from the viewpoint of someone who actually develops for the series. I hate to be blunt, but most of the crowd seems to have positively no clue what they're on about.”

Did you expect any thing less?


“No, there is little doubt in my mind that the new sim will not use the same physics for braking or movement as an aeroplane.”

Finally, some one who has caught on?


“I like how people are trying to somhow imply that there's no way it can be done properly because everything about a train is somehow more complicated.”

It can be done properly if you go to an experienced source for the information.

I’m currently working with a train physics simulator every day that has been well recognized in the railroad industry for decades. It is used for railroad accident/derailment recreation, capacity projects, training, you name it.

I don’t know if I would say that it would be more complicated then say a flight simulator but it is far from being simple.

A pilot is not going do a great job at explaining how run a train just like an engineer is not going do a good job at explaining how to fly a plane.

[Erick_Cantu]

You are indeed correct. Both trains and aeroplanes are subject to the same physics, however, they encounter them in different, non-interchangeable ways. What annoys me is the group who can't seem to recognize that ACES knows this.