Would You Willing To Sacrifice Performance For Better Immersion/scenery/gameplay?

So the topic newly came up due to the preview stream of Fife Circle, where Jasper from Rivet Games mentioned that they sacrificed parts of the nearby scenery like the airport and ships for performance.

We know that this is a common issue at the creation of routes by first and third parties for a long time.

Unreal Engine is really a beast in this regard. The more you add scenery or gameplay elements like AI traffic, the worse the performance. The level streaming architecture of Train Sim World also is quite problematic in this regard.

A developer kind of need to find a balance between them two to get the majority of players satisfied. At least this is what developers say and kind of is common sense.

However, what is your very personal point of view in this regard?

1. Do you prefer performance over immersion?
2. To which degree would you willing to sacrifice performance for having better gameplay/scenery/immersion?

 

A balance is preferred but sometimes performance does take an unfortunate hit on unexpected routes. Goblin for example, on PS5 I usually get between 35 - 50fps but I’m ok with that because the route is still a good experience and very detailed.

In regards to Fife, I can understand the area around the Fourth bridge being performance intensive but does that really affect the entire route? I don’t think it does.

 

One thing I don't get:

BML is, according to DTG, unstable, so much so that they haven't returned to it to fix or add anything. Yet, there are planes at Gatwick Airport.

Fife Circle has nowhere near 1000 services running on it compared to BML, so how are planes performance heavy on this route?

 

Agreed. With the level of scenery of Fife Circle there are no performance problems at all, that’s just an excuse for the state the route is in. We’ve all seen Niddertalbahn, Goblin, Glossop, Blackpool and a host of others. It is absolutely possible to have good performance and good levels of detail

Remember the guy from Skyhook said the exact same thing when that mess was first released. MML is now one of my favourite routes and it is unrecognisable from that time in almost every regard. It turned out to be complete nonsense then, and it’s nonsense now.

 

I know its that old saw, but with the wide range of equipment being used, its hard to find a balance. I run a fairly high end gaming pc, so more detail is not going to impact me as much as saw an older console. DTG is still playing that game of trying to design something to appease to every level and it hurts them on the upper end of things. When the eastern European moder came out with his schedule for LBM that includes most of the 700 services along with all the other Southern and Gatwick Express schedules, and I don't see any difference in performance from the original schedule without the 700 services.

 

I do not mind a few less fps. When playing tsw I prefer playing max settings and getting 30-35 fps rather than medium/high and getting 50fps. In my eyes tsw does not need a super high fps rate like a shooter does.

 

Probably not to be honest.

As much as scenery, traffic and other gameplay features can add and enhance the enjoyment and immersion of the simulation, that all becomes useless if the game is choppy and laggy. In my opinion having a laggy / choppy game is the biggest immersion breaker.

 

Well, let me share my bit of experience I gained through the last months building a route in the public PC Editor, which is not exactly the same tool they use but still very similar.

There is indeed a performance loss dependent upon the amount and level of detail of scenery and timetable traffic as well as other factors. So, it's not a lie.

And dependent upon the route circumstances, different factors have different outcome to the end result. Therefore it's a bit problematic to compare different routes with each other, but I get your point.

So the basic thinking to reduce scenery for performance is not wrong, it's true. However, in case of some aspects of Fife Circle I doubt myself whether it really has that strong impact on performance or not.

However, I kind of asked in a more general approach how players would perceive a dropdown in FPS for the gain of gameplay/immersion/scenery. With the statements of the Fife Circle preview in mind, but not as the main topic to debate about. Sorry, if I didn't made that clear enough.

 

I have a fairly high end PC too and I can play most everything at ultra graphics settings and still get at or close to 60 fps.

But not everyone can attain that level of graphics and performance and that does present DTG with a problem finding a happy medium.

But, to be brutally honest, the onus is really on the player and his/her equipment.
If you have a less than stellar PC, you're going to have to turn down your graphics settings or accept an FPS in the 30's.

If you're a console player, you don't have the same options to adjust these settings and the ball is more in DTG's court to make the game playable.

Either way, DTG have chosen to make compromises which sometimes hurts the few in order to benefit the many.

In the specific case of FCL and the compromises that Rivet claims to have had to make with the planes and ships and scenery, I simply don't buy it. Plenty of routes have those kinds of assets.

If something has to be sacrificed, let it be the gimmicky things like the escalators which don't advance the game and the idiotic fluffy toys which would be more appropriate for a toddler's tablet game.

In fact DTG and partners should quit trying to bring in gimmicks or the whims of some childish coder. That would free up some memory and we can focus on driving trains.

 

Every dev has to consider that most of their player base does not have a high end PC. Just look at Dragon’s Dogma 2 for a good example of what happens when a game actually requires high specs, toaster gamers and mid range gamers are throwing an absolute fit

They also have to consider that half the people that think they have a high end pc don’t

 

Within reason I don't mind a constant lower frame rate to allot more computation resources to graphics and immersion, but I don't want frame rate drops as those are very noticeable and distract massively.

I prefer a constant 30 fps over mostly-60-with-sudden-drops-to-five-or-ten.

Tune per route and system how much fps to sacrifice to bring in all the desired immersion, but keep it steady.

 

Exactly. I think if DTG decided to just prioritise those with higher end equipment and left those with low/mid end equipment behind, I feel that could equally have a huge damaging effect, which is why they haven’t

 

Personally I think it's just an excuse from Rivet to paper over the deficiencies of their route builder(s). Most other routes that are packed with detail run fine and I've even just bought the Goblin line despite saying I wouldn't so if anything is going to stress my potato PC that will. I mean BPO is packed with detail, varied terrain textures as are just about every other route in the collection. What about Dresden to Reisa, the Meissen branch - that runs fie and again is full of assets.

 

You can’t escape from the fact that more assets in an area = worse performance.

I do not however, believe it means everything has to barren to achieve good performance.

I don’t really believe Rivet ever tried to add the planes, better scenery or more populated areas, I think they simply just never put anything there & are playing fast & loose with the performance card.

Another issue I think Rivet have is that they’re clearly using an outdated asset library, the grass, ballast & trees alone are ones you find in older routes. Everything has that horrid mushy, 1 shade of green texture to it.

Pretty much every route in TSW has areas where it dips in FPS as it’s a busy area, the Forth Bridge might be a big asset, but there’s no way I believe it means the route has to be empty and have TSW2 scenery.

 

Personally it is the consistency of FPS rather than the actual figure that bothers me more and I think that this means there does need to be a balance as having some areas at 60 and others at 40 would be worse than having a solid 30. Sorry if that doesn't make sense, but I hope it does!

 

never. performance always won for me. but as a TSC player, which game is...not so good on optimization, what affects in final moment might be ballance between factors.

 

I know this will sound clichè but I would expect DTG to be able to have a consistent good performance while having as much good scenery and detail as possible

 

As I said before, Unreal Engine is the problem. And DTG's hands are literally tied to the limits of it.

It's not about how much money and ressources they could spend to get it right, it's an Unreal Engine characteristic which can't be changed by them, nor any other developer except maybe Epic Games, who develop it.

Yes, you have better detailed maps in other games, however the size of the maps of them is not nearly as big as it is in TSW, which could reach up to several hundred square kilometer. And TSW routes are kind of build on this elementory requirement of level streaming to make all happened. Not sure on which engine TSC was built, but TSW is a different game and has (moreover) superior graphics to TSC.

So everything in the end relies on Unreal Engine, not DTG. And Unreal Engine just can't have an indefinite number of assets and level of details whilst having top-notch performance. Two contrary opposites. The main opposites of Unreal Engine and maybe every other game engine as well.

I really would wish we could create way better detailed maps with way better gameplay elements and way better immersion, but it's simply not possible with Unreal Engine, at which TSW is build upon.

I said before, I'm not quite sure whether it even was the right choice to build Train Sim World on Unreal Engine. However, this engine might have been the only way, how a simulator like TSW even could have been created. Level and tile streaming is an important topic to it, and not sure if any other engine could handle it in the way Unreal does, even it means we get this annoying pop ups of scenery and stutters during the gameplay.

 

I play always on HD instead of 4K. I don't get stutters when I play on 4K but the camera is so much slower. So long I don't get stutters or noticable framedrops, I'm fine with less FPS.

 

I’d say that things get noticeable below 45 fps on PS5, which does not currently ever have a stable fps and of course not everything is actually running at the same framerate in TSW. Some things are capped at 30 fps, like reflections I believe, and may be even lower on PS5. I wouldn’t want to have things get worse than they are now. Even on the best performing sections, the movement of scenery when looking sideways the scenery is very jerky. I guess if Rivet are using a smaller, older library of assets then their routes may just run worse so I wouldn’t want them to be trying to cram too much foliage in for example. It would likely make things very bad. I have no idea how asset sharing works between DTG and third parties so I can’t comment any further on that without the necessary knowledge. I’m just a guy on his sofa. I vote for no change.

 

I’ve always held the view that performance is their to be spent, not to make it run awful that’s just unbearable, but happy to sacrifice quite frankly for the most part I often find “the unnoticeable difference”

i did vote depends as I think it should be used sensibly on graphics/scenery etc with respect to performance.

if a section is already bad on performance best not to make it worse as one example, but excess should be spent or it’s just wasted.

 

The Goblin line is good candidate for this topic. After having sunk quite a few hours into that route now, It’s a fair statement to say that DTG favoured resolution more than FPS imo. The route definitely hovers around the 30-50fps mark on the PS5 for me.
I thought it would bother me, but it doesn’t. The scenery, stations, assets, the loco etc. Everything is so well done. I wouldn’t want DTG to sacrifice the resolution for a slight bump in performance.

 

I really fail to see what about Blackpool would make it a better route for detailing than Fife like Robert's alluding to. The routes I'd see it being possible to do higher trackside detailing on would be something like Clinchfield or even Sand Patch. Those have high hills close to the track which means you don't need to do distant scenery like a flat route where you might actually need to model landmarks a mile out. Fife is maybe a bit more hilly, but it ain't anything that your external camera couldn't look above. So I don't see what would be vastly more demanding that it would need to hold back, but Blackpool doesn't.

And at any rate, even if you have to limit detail for last gen consoles that doesn't mean you can't do a next gen version with extra detail. I've been playing Hogwarts Legacy lately, and that game is a open world game on Unreal Engine 4 that's astonishingly detailed. But it's worth noting that for all the issues people like to complain about with UE4 one of it's major perks is that it's very scalable. To the point that HL, despite being clearly made for next-gen platforms first has competent versions on Switch and the last-gen consoles. Like yes, you're taking hits on detail especially on Switch, but the fact it's playable at all without massively changing the game is impressive. And importantly it means they don't need to hold back the PS5 or PC versions just to make sure the older platforms can handle it.

 

Bumping this thread after almost a year (how time flies) because of the on-going discussions about performance issues especially in regard to Train Sim World 5 since release but which clearly seem to even increase over time and now filling almost every thread.

While I'm not denying that there might be severe issues with Train Sim World 5 in particular which increased stutters and other performance issues for many players, a thing a lot of player (including those who calls us out as DTG defenders or apologists) don't seem to understand still is that Unreal Engine is simply not a very performance-suitable engine for the purpose of building large worlds with high-detailed scenery and lots of other performance-eating stuff happening at the same time.

This is fundamentally different to any other games you use to play, which were built in a different engine or if they're built in Unreal Engine, use smaller levels with less performance-hungry tasks going on in. So comparing Train Sim World with something like Forza Horizon is unfortunately like comparing apples with pears. Train Sim World will never be a game that handles performance as these, and this will come clear pretty soon to anyone who just invested some time with the Public Editor seeing how much of an impact already a slightly increased density of grass near the trackside (not even distant scenery) has for example.

However, and this is the benefit of Unreal Engine, due its feature of level streaming, it allows to generate almost indefinite worlds in 1-1 scale, which other engines can't do.

So was Unreal Engine a good choice for Train Sim World? Performance wise absolutely not. However, even if Train Sim World isn't nearly as powerful on this topic as other games we know, we at least got the game how it is now. This doesn't mean the code couldn't get enhanced or the performance couldn't brought to a state it had before Train Sim World 4, but simply it says that demanding more performance while also demanding dense and detailed scenery and more and more features which come with each iteration of the game, simply is illusional.

Doesn't matter how much you complain, DTG can't suddenly transfer the game to a new engine and build it from scratch there, this wouldn't nearly make any sense.

So what could DTG and developers do to enhance the experience for the player and get some more FPS for them?

If we take hot topic Frankfurt-Fulda, which was slayed for its performance impact, for a closer look, we can see that the factors scenery and traffic both being very high in the area around Frankfurt, and it absolutely make sense out of my perspective that hard stutters or even crashes do happen when the tiles for Frankfurt being loaded in, which is more or less somewhere between Offenbach and Hanau and ta-da said to be the worst location of the route in regards of performance. What a suprise!

If the few of us recall, DTG said before doing Frankfurt-Fulda that Frankfurt would never be likely to come into the game because of its performance impact at Frankfurt. Yet they did but after that all people complain about how bad Frankfurt is on performance.

Almost same thing with Mildmay Line now which covers large parts of London and skyscrapers can be seen from the distance plus lots of traffic services (especially AI) and yet again people complain how bad performance is on that.

So what is that sending for a signal now to the developers? People ask for more metropole stuff like Berlin, Paris, London, New York but these will never come if performance is indeed more important for you all than the scenery and the route itself and you can't live with the hit on performance in return.

For me as someone who spent the last 15 months of building routes in the Editor and spent quite some time putting enhanced scenery on my levels (and yeah indeed I have no clue how bad performance is in the game currently because I'm mostly spent my time in Editor), my perspective on the demands of the players is as split as yours.

Because if one side you worship developers for getting as much detail and traffic on the routes but on the other want that everything runs as smooth as peanut butter, which is impossible for me (or us) to achieve.

Retrospectively, taking Fife Circle as example was maybe not best choice for my poll indeed, as performance cannot be seen as an excuse for everything, but it still is a fundamental and real problem every developer has. Because the more you detail a route, the worse it will perform.

All what I just wanted to ask for is if people would be generally okay with less detail of scenery, with less gameplay and maybe less traffic, if the game runs smoother then.

As I saw from lots of posts in this forum during the last 6 months, there absolutely seem to be a lot of people which would prefer performance over immersion.

So that made me think: Should developers now switch back to route-building in a way simpler style just building the nearby-scenery on complex metropole routes, and omitting much of the AI traffic, just to enhance the performance experience for the player?

Because as all of your complaints show, the majority of you clearly want that.

 

How about they optimize their game and their assets so it doesn't run like garbage?

Then there is no need to sacrifice anything.

 

Indeed. A game that looks as bad as TSW has no excuse to run at such low framerates. The options in the game are minimal and so there's only so much we can do even on PC. There are barely any anti-aliasing options, so for the game to look marginally smooth it has to be run at above 125% screen percentage, which kills performance as well.

 

I voted for the first option, but if given the choice I would have clarified I wouldn't want the framerate to go below 30 FPS.

My Ryzen 7 5800H and GeForce RTX 3060 Laptop with 32 GB RAM can handle old routes perfectly and new routes (such as Mittenwald) just fine on 1080p (even 4K with some compromises), with some exceptions at large stations, such as Köln and Bremen (or is it Oldenburg). In those cases the framerate dips from 40-60 (which I find totally acceptable) to 20-30, which I can live with.

On Frankfurt Hbf and at Hanau, the framerate dips into the single digits and time actually slows down. That is not acceptable to me. If my rig can run every route mostly without that much trouble then I would expect no different from Frankfurt. Especially when the rest of the route runs at an acceptable 30-60 FPS. Not as stable as I would want, but OK.

That, or they should indicate the minimum and recommended specs per route.

 

Wouldn't it be a better option for dtg to introduce the same graphics settings that other games have. Like lod, draw distance etc... dlss, fsr.
On the pc you can mess with the ini file, favour performance over quality and vice versa. On console your stuck with what dtg pre define in the settings.
As far as I'm aware you can disable layers to free up memory, but what about distant scenery?
If planes flying around uses resources, and some players don't care about having planes in the skies, wouldn't be a good Idea to have an option to turn it off?

 

As far as I'm aware it is not possible to turn specific scenery elements off this way, no. I guess if that would be possible and we think further that way, we could have way more enhanced scenery on PC than consoles. The only thing I do know in this regard is that there are possibilites to set separate timetables for PC and Gen9/8 consoles if I recall correctly, which makes sense since we have cut down timetables on some routes for Gen 8.

The advanced settings in-game do have a "View Distance Quality" parameter though, which adjusts quality and loading of objects in the distance and can be set to "Low", "Medium", "High" and "Ultra". However, most people want it to be as high as possible or simply don't know it could be turned down I guess. But I'm also not quite sure how strong the effect of this particular parameter even set to "Low" is.

 

FPS drops at Frankfurt are down to your CPU, not your GPU

For what it’s worth the minimum spec on Steam isn’t worth the paper it’s written on. They say Windows 8, 10 or 11 then spec a system that Windows 11 won’t run on. Windows 8 is no longer supported and Windows 10 support ends in October this year. Like it or not, the minimum spec of this game, all games in fact, will shortly become the minimum spec to run Windows 11 (basically a GPU compatible with DX12 and TPM v2)

TSW5 might run on that spec machine, but it would look absolute guff, to the point of being unplayable. Personally I think it’s high time that was updated to something realistic

 

I kinda feared that :/ Nothing much I can do then, right? Running it on DX12.

 

DX12 is all you can do to alleviate it, but your GPU needs to be able to cope. TSW has always been really heavy on CPU’s in areas with many tracks, I’m not sure why it is so intensive but it always has been that way sadly.

 

Thanks for the insights! I might have to build a PC in future then (because even after everything, I really love this game)…

 

This relates directly to why Gen 8 consoles aren't as supported now.

 

I was doing some experimenting with the in game settings vs ini settings.
If I set all the in game settings to low, with a big difference in the quality of the world, I mean almost all vegetation was gone, shadows gone, I only managed to gain 15fps. However I did notice that distance objects were still visible and track draw distance remained unchanged.

Then I reduced some settings in the ini file, all vegetation was still present although it didn't become fully rendered until I was on top of it. However distant scenery was now gone and the track draw distance was reduced and I gained 50fps.

So there must be some error in the way the in game settings work.

 

I think the whole thing needs a huge overhaul personally.

That would make sense though. Track draw really hurts CPUs where there are many tracks, if it isn’t decreasing with the settings that’s undoubtedly problematic

 

Experiementing with settings alone is difficult, because quite often the cause can simply be some badly set up or unoptimized asset.

Good example are the original LIRR M7 units which were known to cause framerate drops when on screen.

Same with one of the Clinchfield boxcars. Matt even acknowledged it once, saying it's probably set up incorrectly in some way (could be something to do with the animations I believe?), yet it never got fixed.

Or the framerate killer grass asset that Sherman Hill used to have.

Things is, a small unoptimized asset can have huge impact on performance when you have tens or hundreds of them in a scene.

 

I voted for yes since it's the rason I upgraded my GFX two weeks before MIT release and with JetWash's beast ini and the better shadows mod by SpaceyKat, this game, after so many years, looks incredible and runs ultra-smooth
A G-Sync Screen helped me the years before to mitigate micro-stuttering.

I would choose more and better scenery over FPS every time and fight bottlenecks with (informed) hardware upgrades. Like I did the last 25 years. That's PC gaming as always. And console hardware is (out)dated within one year after release, like the last 30+ years..?

¯\_(ツ)_/¯

Get yourself decent hardware if you plan to run all the layers (that actually flesh-out the game and make it more true to life) and immerse yourself in this

 

Yeah I think it's the way tracks are drawn. It seems to me the tracks are drawn in big sections, like I'm driving along and I can see the sections of track being added up ahead rather than them being drawn as a line. So I guess the cpu has a lot of pre rendering to do before each section is added.

I guess that dtg makes modifications to the ini settings when they release a route to make the balance between image quality and performance.
This is why console players get blurry textures.

Personally I believe there should be more graphics options in the game menu, preferably tied to these hidden ini settings, to give players more control over quality vs performance.

 

But of course.
Sell us a slideshow....
I'm already buying fewer and fewer roads precisely because of the stuttering and the low FPS, so if it's going to drop even further, I think I'll stop buying anything.
We don't all have racing PCs for the latest generation consoles.
To be honest, a pixel art game would be much nicer; it would consume less power, and we could have details like never before!
Today, some games like "Art of Rally," for example, are very beautiful, yet not as "realistic" as the "next generation" games.
I play a lot of Densha de Go; they may not be as pretty, but they're so much more immersive, with lots of AI trains. The Yamanote line is simply superb, on Switch or even on the PCSX2 emulator (PS2), we can see that there's no need for NASA's gimmick to have a decent and fluid game.
If immersion is what we're looking for, then why not limit the game to the internal view like in Densha de Go?
Everything that we can't see from the cabin view becomes obsolete, and we can therefore remove it and gain resources for more details and AI trains.
I still think that TSW5 is very good but also very, very poorly designed from the start.
For me, it's time for a new, better-designed train game.

 

I think a certain dev speaks with forked tongue. I don’t trust anything they say. An airliner or two should not cause the problems alleged. I think it’s more the extra time and work involved, they like to do the bare minimum to enable the price tag.

 

I prefer better FPS. I like a smooth ride.

 

He can you have performance issues with reused assets?

Id also wager that the performance problems and Bugs are due to simply rushing out content.

Even on my favourite German routes there’s clearly reused assets. Buildings, cranes, most of the moving traffic.

The roads are not detailed either, neither are the distant hills and trees.

Traffic is only pathed in the most basic of fashion, cars don’t even turn properly at junctions.

I just think not enough time is spent on optimisation. Some of the games you can run on Xbox look stunning, but TSW is a mixed bag. It should be well better.

 

Part of the issue is different platforms (not even two XBox are identical given different updates, versions, ages, etc let alone different computers), and part of it is that there are so many "cooks" in the kitchen. By that I mean there are a ton of "teams" at even DTG, let alone the third party DLC, and the assets many use are often taken from yet other third party vendors/sources. That individual person probably didn't create every tree or bush they use, they likely got it from some other person as part of a "pack" and then that person may have gotten it from someone.... so it's not the same people making the "whole game." Anyone who has been on a team knows what I mean. It doesn't even have to be computer game development.
Imagine even a restaurant kitchen. One person doesn't make the whole meal. They each make one item, and it comes together at the end to you as a "meal" but if even one thing is sub-par, the diner will often will say "well that meal sucked" as a whole. The cooks are also dependent on the food from the distributor, which is just the truck delivering the food who then have to depend on the quality of the produce from the farmer, etc.

That's why it's not all "optimized" and often varying in quality.
It takes a lot of work to monitor inputs, process and outputs, and it takes a willingness to toss bad product to make new if found to be sub-par....but both things tend to raise prices. I can't speak to game development since that's not my experience, but restaurants and dining facilities are. IRestaurants are VERY price conscious (because the customers are) and it's often a battle to keep the quality higher because that often leads to higher prices which can cost you customers. There's not a lot of "wiggle room" without losing money and/or going out of business.

Given what I've heard from DTG on the forum, they are very price conscious too, and they are very cautious about tossing bad content or redoing it because it could push it over the cost that players are willing to spend. So if the food is just kinda "meh" but people are still buying it... they'll keep sending it out.

The truth is a lot of people will keep going to an "okay" restaurant if the food "isn't too bad but it's cheap." Video games are like restaurants in that they are both luxury items. People can always go make their food at home, or go find something else to do than buy a video game.

 

Wish Train Sim World was imported to Unreal Engine 5 as Unreal Engine 4 is old and outdated

 

It wouldn't change a thing. The game is simply unoptimized in Unreal Engine 4. It would be just as unoptimized in Unreal Engine 5 as well.

 

As a predominately PS5 player I would like to see quality/balanced/performance options which would do whatever was needed to deliver a smooth 30/40/60. A route running at a steady 30 will play better than something lurching between 20 and 50 FPS.

Having said that, at the moment the biggest problems on console are the memory leak problems causing blurry textures, a blurry mess at any frame rate is the worst situation and needs fixing before I'll buy any more routes.

 

What is it with this mentality that upgrading TSW to UE5 would change anything.

 

Can't guarantee this as I can't evidence it (can't remember when it was said) but I'm pretty sure that Matt even said switching to UE5 wouldn't change anything as things would need to be re-worked to take advantage of it

 

Also some may not have seen it but we finally got a response from DTG (Chris) in the Troubleshooting Forum the other day which pretty much concedes they are scratching their heads at the stuttering and sound squelching in particular. So unless they can bring in people with the necessary skills and talent to untangle what's gone wrong with TSW's use of the UE4 game engine, there's only so much upgrading the hardware can do.

I still think the new 2D map is one of the performance hogs, so really there needs to be an option for those of us on lesser PC's or consoles to revert to the old style map, along with a great many more options to tune graphics performance.