NV3 whats the future of Trees and shrubs for Trainz?

Tony_Hilliam said:
None that you really want to hear. If we concentrate on trees, we'll be accused of being a "vegetation simulation",
Click to expand...

I would NEVER accuse you of such a dastardly deed as that ! :eek: There has to be a balance though.

Just curious....what's the attraction to playing a vegetation simulation. Do you to sit there and watch fake grass grow or is there more to it ?
 
Last edited:
Tony also failed to address whether the programmer(s) could come up with a way to stop the transparency artefact on the older trees. As I said earlier, it is ironic that an Australian based sandbox programme has but a handful of its own indigenous tree and flora types that meet current standards.


I mean obviously if it comes to tossing a coin between trees and a proper player timetable mode/editor as to where the development $$$ go then I would plump for the latter. But a bit of dynamic thinking as to how the current hiatus could be resolved would not go amiss.

Also as I mentioned, some of the RMM trees are now showing with either transparency issues or white artefacts on the leaves, so they also get crossed off the list.
 
Vern said:
As I said earlier, it is ironic that an Australian based sandbox programme has but a handful of its own indigenous tree and flora types that meet current standards.
Click to expand...

In my current project, outback South Australia, I have a picklist containing more than enough suitable local SpeedTrees that work perfectly. Likewise suitable shrubs and grasses.

I agree with your point about N3V getting their priorities right. They should concentrate on the programming, the users should be creating the trees they want to see. If we rely on others to do that job then no-one would get the exact trees that they want.

This "discussion" over the "tree gap" could easily be extended to just about every other type of asset.
 
pware said:
In my current project, outback South Australia, I have a picklist containing more than enough suitable local SpeedTrees that work perfectly. Likewise suitable shrubs and grasses.

I agree with your point about N3V getting their priorities right. They should concentrate on the programming, the users should be creating the trees they want to see. If we rely on others to do that job then no-one would get the exact trees that they want.

This "discussion" over the "tree gap" could easily be extended to just about every other type of asset.
Click to expand...


Well I'm glad you are perfectly ok, I just had a look at whats available for Australia as I am probably going to make a South Australian route, however, I have to differ from your opinion , as I think the auran Eucalypt speedtrees look most unconvincing, I cant even think what sort of gums they are based on, they really are very generic. I am however very picky and perhasp unrealistically so given the possibilities with the software.

I never actually use any of the larger Auran items apart from bushes. The older Australian trees that could be used for an australian route are now transparent , but they would have looked much more like the actual thing when they were solid . Its a shame they haven't been updated, but if you are happy with the built in stuff than that's nice for you.
.
The only trouble with the strategy of letting the users create the content is that not enough people are creating trees . As far as I know, we no longer have anyone creating payware , I'm quite prepared to pay for trees and have done so in the past , but now, appears to be no source that is currently making trees .

I cant think of any other area of scenery objects that is this ubiquitous and which is this neglected, this is probably due to the difficulty involved in making a good looking tree or grass that doesn't slow down the frame rate.

Anyway we now know NV3 are not going to let us know what their plans are ( if any ) so I'll see if I can manage to commission someone to make trees and hope that NV3 don't make more changes that make much of the vegetation look even worse than it does now.

I wonder if people would be so sanguine about this sort of problem if engines and rolling stock started to become transparent ? , I'm darned sure NV3 would fix that sort of problem pretty sharpish .
 
Last edited:
Well I'm glad you are perfectly ok, I just had a look at whats available for Australia as I am probably going to make a South Australian route, however, I have to differ from your opinion , as I think the auran Eucalypt speedtrees look most unconvincing, I cant even think what sort of gums they are based on, they really are very generic. I am however very picky and perhasp unrealistically so given the possibilities with the software.

I never actually use any of the larger Auran items apart from bushes. The older Australian trees that could be used for an australian route are now transparent , but they would have looked much more like the actual thing when they were solid . Its a shame they haven't been updated, but if you are happy with the built in stuff than that's nice for you.
Click to expand...

Pretty much the point I was making, apart from the RMM euca's as mentioned, I certainly never found much that looked right in TRS19.
 
There are, I believe, over 1200 different species of eucalyptus trees in Australia, from the Alps to the edges of the deserts. Do we expect that all these different species should be available in all their different variations? Then what about all the other species, the bottle brushes, acacias, wattles, etc?

I have no problems with the quality of the available built-in and DLS SpeedTree gum trees on offer, but then I am obviously not as picky. Scenery is meant to provide a backdrop to the railway, to be seen from a passing train, not studied in close detail to pick out its imperfections otherwise nothing would pass muster.

Good luck (seriously) with finding someone to take on the job of designing perfect SpeedTrees. If you succeed and it becomes payware then I will not be interested. I have no problems with session assets (e.g. rolling stock) that are payware but not route assets such as trees.

I am of the opinion that N3V have more important issues to resolve, and issues that are much more in demand, than trees.

We are simply going to have to disagree on this matter.
 
Hi All
Unfortunately trees are a major issue for most games developers, and a LOT of money goes into this area in most cases. This is why companies such as IDV/SpeedTree have become so extensively used, as their tools allow for extremely good tree generation relatively easily (compared to trying to hand make the same tree in a 3D modelling tool, or trying to make billboards look like 3D trees).

Unfortunately, in modern gaming environments, flat planes for trees are exceedingly hard to handle without making them stick out like a sore thumb. This is why they do change when changes are made to the rendering methods/systems in Trainz. The more realistic you make the game engine, the more they look like flat/flattish planes.

In particular, if any shadowing is applied to a billboard tree, it suddenly looks poor due to the flat-plane nature being highlighted by the shadows. Shadows can be disabled on specific meshes, but it isn't currently possible to prevent a shadow being cast onto a mesh (and doing so may not work visually as you would now have shadows behind that mesh but not on it!).

The same applies to shading (ie if the surface is lit or unlit), as your light vs dark has a hard line.

Twisted billboards, as in Clam1952's trees, can make big steps toward mitigating some of this, but they are still effectively flat in nature. Unfortunately this means that the shading still shows up the flatness.

There are some possible options to work around this, but these can introduce major problems as well. Either for the creator, or for the visuals, or for everyone...


One is to remove all directional lighting from these objects; which is how billboards were handled in earlier releases. This however introduces a lot of issues with matching them with the surrounding environment, as surrounding objects (including other trees) may be lit by the sun and yet the billboard tree is only being lit by ambient light. There are also issues with other areas with this, including limited or no normals map ability, and a potential to break the PBR rendering (since ambient lit surfaces have different reflections to direct lit surfaces).


Following on from this is to reduce the directional lighting by adjusting the object's normals. This can allow you to change how the object appears to be lit (ie does the object act as if it is simply pointing upwards, rather than horizontally), but going too far with this can cause issues, such as it appearing too bright or too dark compared to surrounding objects. Finding the balance can be hard, but can produce helpful results.
.

Another is camera facing planes. This is used for SpeedTrees for their 'billboard' lowest LOD, and for leaf planes on some SpeedTrees up close. Unfortunately there's no option to use this outside of SpeedTrees. However camera facing planes can be extremely distracting on larger objects up close, as they all rotate to face toward the camera all the time (so as you move, the trees rotate/move in sync with you). Even on leaf planes on SpeedTrees it can sometimes be distracting IMO, but it is less so than the entire tree doing this. The 'JVC-BB' trees are an example of an entirely camera facing tree, and show mostly how this works (including issues with it rotating in the wrong axis).


Effectively this brings us to the fact that billboards are not natural or realistic objects, and in effect are going to work poorly the more realistic you make the rendering in the game. This leads to the need to move away from billboards, as has been occurring for the most part in Trainz. There are, of course, some object types that can benefit from billboards, in particular smaller bushes/shrubs/etc where the flatness is going to be less obvious, and of course they will be a necessity in the lower LODs of otherwise 3D objects.


Unfortunately this does mean that there needs to be a move toward 3D trees, either manually modeled (as some members have done over the last 11 years), or created with SpeedTree or other similar resources. Some tree generation tools may allow you to import a modelled/textured tree into Blender or 3DSMax, where you can then lower the poly count, and then create LODs for it.

That said, all is not lost. The trees that Tony mentioned are all TANE or TRS19 standard SpeedTrees, and work quite well from my experience. Not every tree will be suited to every route. And some creators have seen some good results mixing select billboard trees with SpeedTrees to fill out forests, especially beyond the initial tree line.

IMO it's probably also worth contacting creators to see if they will allow you to reskin their released SpeedTrees. Some may have a 'close enough' shape for a different type of tree by simply changing the bark textures and the leaf textures; you'd need to play with the colours in the billboard to match as well, but it doesn't need to be exact in most cases. Even making the leaf and/or bark textures lighter or darker can create new variants that may suit different environments better.

It's also not impossible to create seasonal variants from otherwise non seasonal trees, since seasons can be in their own individual folders. Although not as efficient, simply masking out the leaves in the alpha channel of a 'winter' variant of a tree can make a winter version without needing a new model.

Unfortunately it's not possible for N3V to make most of the content for Trainz. Art man hours are costly, and as such N3V does need to focus our time where it will get the most value. At times this has been foliage, but that will change depending on what is in development/planned/etc. For the most part, it is up to the community to development the majority of content, however N3V has worked with members of the community on various content projects, however unfortunately only a limited number have done this in relation to foaliage. However if members of the community were interested in working with N3V to develop further foaliage, they should contact the team to discuss this.

I know this isn't necessarily what people want to hear, but as a route builder and 3D artist, this isn't a simple issue with simple solutions. Trees, like I said, are one of the most difficult object types to get right in modern games.

Regards
Zec

P.S. for anyone interested in modelling trees, this page might be of some help :) http://wiki.polycount.com/wiki/Foliage
 
S301 said:
Hi All
Unfortunately trees are a major issue for most games developers, and a LOT of money goes into this area in most cases. This is why companies such as IDV/SpeedTree have become so extensively used, as their tools allow for extremely good tree generation relatively easily (compared to trying to hand make the same tree in a 3D modelling tool, or trying to make billboards look like 3D trees).

Unfortunately, in modern gaming environments, flat planes for trees are exceedingly hard to handle without making them stick out like a sore thumb. This is why they do change when changes are made to the rendering methods/systems in Trainz. The more realistic you make the game engine, the more they look like flat/flattish planes.

In particular, if any shadowing is applied to a billboard tree, it suddenly looks poor due to the flat-plane nature being highlighted by the shadows. Shadows can be disabled on specific meshes, but it isn't currently possible to prevent a shadow being cast onto a mesh (and doing so may not work visually as you would now have shadows behind that mesh but not on it!).

The same applies to shading (ie if the surface is lit or unlit), as your light vs dark has a hard line.

Twisted billboards, as in Clam1952's trees, can make big steps toward mitigating some of this, but they are still effectively flat in nature. Unfortunately this means that the shading still shows up the flatness.

There are some possible options to work around this, but these can introduce major problems as well. Either for the creator, or for the visuals, or for everyone...


One is to remove all directional lighting from these objects; which is how billboards were handled in earlier releases. This however introduces a lot of issues with matching them with the surrounding environment, as surrounding objects (including other trees) may be lit by the sun and yet the billboard tree is only being lit by ambient light. There are also issues with other areas with this, including limited or no normals map ability, and a potential to break the PBR rendering (since ambient lit surfaces have different reflections to direct lit surfaces).


Following on from this is to reduce the directional lighting by adjusting the object's normals. This can allow you to change how the object appears to be lit (ie does the object act as if it is simply pointing upwards, rather than horizontally), but going too far with this can cause issues, such as it appearing too bright or too dark compared to surrounding objects. Finding the balance can be hard, but can produce helpful results.
.

Another is camera facing planes. This is used for SpeedTrees for their 'billboard' lowest LOD, and for leaf planes on some SpeedTrees up close. Unfortunately there's no option to use this outside of SpeedTrees. However camera facing planes can be extremely distracting on larger objects up close, as they all rotate to face toward the camera all the time (so as you move, the trees rotate/move in sync with you). Even on leaf planes on SpeedTrees it can sometimes be distracting IMO, but it is less so than the entire tree doing this. The 'JVC-BB' trees are an example of an entirely camera facing tree, and show mostly how this works (including issues with it rotating in the wrong axis).


Effectively this brings us to the fact that billboards are not natural or realistic objects, and in effect are going to work poorly the more realistic you make the rendering in the game. This leads to the need to move away from billboards, as has been occurring for the most part in Trainz. There are, of course, some object types that can benefit from billboards, in particular smaller bushes/shrubs/etc where the flatness is going to be less obvious, and of course they will be a necessity in the lower LODs of otherwise 3D objects.


Unfortunately this does mean that there needs to be a move toward 3D trees, either manually modeled (as some members have done over the last 11 years), or created with SpeedTree or other similar resources. Some tree generation tools may allow you to import a modelled/textured tree into Blender or 3DSMax, where you can then lower the poly count, and then create LODs for it.

That said, all is not lost. The trees that Tony mentioned are all TANE or TRS19 standard SpeedTrees, and work quite well from my experience. Not every tree will be suited to every route. And some creators have seen some good results mixing select billboard trees with SpeedTrees to fill out forests, especially beyond the initial tree line.

IMO it's probably also worth contacting creators to see if they will allow you to reskin their released SpeedTrees. Some may have a 'close enough' shape for a different type of tree by simply changing the bark textures and the leaf textures; you'd need to play with the colours in the billboard to match as well, but it doesn't need to be exact in most cases. Even making the leaf and/or bark textures lighter or darker can create new variants that may suit different environments better.

It's also not impossible to create seasonal variants from otherwise non seasonal trees, since seasons can be in their own individual folders. Although not as efficient, simply masking out the leaves in the alpha channel of a 'winter' variant of a tree can make a winter version without needing a new model.

Unfortunately it's not possible for N3V to make most of the content for Trainz. Art man hours are costly, and as such N3V does need to focus our time where it will get the most value. At times this has been foliage, but that will change depending on what is in development/planned/etc. For the most part, it is up to the community to development the majority of content, however N3V has worked with members of the community on various content projects, however unfortunately only a limited number have done this in relation to foaliage. However if members of the community were interested in working with N3V to develop further foaliage, they should contact the team to discuss this.

I know this isn't necessarily what people want to hear, but as a route builder and 3D artist, this isn't a simple issue with simple solutions. Trees, like I said, are one of the most difficult object types to get right in modern games.

Regards
Zec

P.S. for anyone interested in modelling trees, this page might be of some help :) http://wiki.polycount.com/wiki/Foliage
Click to expand...
Thanks Zec, that's the sort of answer that I might have hoped for from Tony, it puts it more in perspective . I know how hard it is to do good 3D as I've dabbled with it over the years.. My real concern is the lack of people who are prepared to tackle the difficulties of making vegetation look good and the indifference of the company to the fact that fewer people are involved in producing these assets at this point in time.
If it were possible to edit the auran speedtrees to make variations that might be a help as they callus be tweaked to provide variety .
There needs to be financial incentives for a new player to make more vegetation for the future but how we go about it is an issue, as I very much doubt we can decide on priorities . I want an arid set of items for the west of the USA , but I might very well be in a minority and even if I did organise a kickstarter, I might very well end up with no assets that I wanted .However , even if I went out and produced a set of assets for myself to use, I then might find that a future software development would make my set look shabby. I'd then have to stay with an older version of the game and nv3 would lose a sale.

This isn't an easy problem to solve is it ?
 
pware said:
There are, I believe, over 1200 different species of eucalyptus trees in Australia, from the Alps to the edges of the deserts. Do we expect that all these different species should be available in all their different variations? Then what about all the other species, the bottle brushes, acacias, wattles, etc?

I have no problems with the quality of the available built-in and DLS SpeedTree gum trees on offer, but then I am obviously not as picky. Scenery is meant to provide a backdrop to the railway, to be seen from a passing train, not studied in close detail to pick out its imperfections otherwise nothing would pass muster.

Good luck (seriously) with finding someone to take on the job of designing perfect SpeedTrees. If you succeed and it becomes payware then I will not be interested. I have no problems with session assets (e.g. rolling stock) that are payware but not route assets such as trees.

I am of the opinion that N3V have more important issues to resolve, and issues that are much more in demand, than trees.

We are simply going to have to disagree on this matter.
Click to expand...

We are indeed. You aren't interested in vegetation that is payware and this really epitomizes the problem for those who might make vegetation as payware ,it takes a very long time to make it . So people will pay for engines and rolling stock, but not so much for vegetation, its seen as an afterthought, a "backdrop to the railway " whereas if you really think about it, NV3 are pushing how realistic their software looks aren't they ? But to see a perfect engine with highly detailed track and rolling stock in a not very convincing backdrop full of transparent trees or trees that vaguely look like the species they are supposed to be is pretty strange isn't it ? The best railway modellers have rout that look great in all aspects , not just the 'railway " so why should trainz be any different ? . I think that's one reason i cannot warm to Trainz 2019 very much as so many of the payware routes are full of their trees, which I really do not find convincing, in fact , I'd rather have some of the better billboard trees than the ones that are featured in the game itself.

However I know it IS possible to do both , make the background look as good as the 'railway " as the work of Jango, Fithers and a few other masters shows, most of us cant get to that standard, but given the right tools we may well do. I'd just like to see some of the gaps filled .So I don't expect thousands of species to be created , but ONE bottle brush might be nice , just as ONE pretty realistic looking sagebrush would suit me as long as I could change its colors and size. I may just have to try and make them myself and stop building routes, which would be rather self defeating wouldn't it ?

BTW just have look at some of the trees that JVC was making before he stopped earlier in the year, it shows what can be done in this area by someone who is prepared to put in the time .https://forums.auran.com/trainz/showthread.php?156371-JVC-Payware-2020&highlight=jvc+trees
 
Seemed to me there was another thread, maybe under freeware? Of someone who was doing new trees. I'll see if I can find it....
 
Forester1 said:
Seemed to me there was another thread, maybe under freeware? Of someone who was doing new trees. I'll see if I can find it....
Click to expand...


That was Harold hminky, using jankvis's normals fix, unfortunately Harold had to give up due to serious health issues.
 
Forester1 said:
Seemed to me there was another thread, maybe under freeware? Of someone who was doing new trees. I'll see if I can find it....
Click to expand...

This might of been mentioned earlier in this thread but has anybody ever looked at the sapling add-on in Blender? It can make trees however I don't think anybody ever tested them in Trainz.

Anyway...I'm still waiting on them to fix the night lighting to something a bit more realistic.
 
clam1952 said:
That was Harold hminky, using jankvis's normals fix, unfortunately Harold had to give up due to serious health issues.
Click to expand...
He did post instructions I believe , I am going to look into it once I've finished this route, however it may limit the types of vegetation that can be created as it's mostly trees that are blueprints. I am also looking at treeit as a possibility and have contacted JVC to see if he could be persuaded to return to veg making if he had some financial support.
 
This discussion, as started by Dangavel and continued in Zec's very good post, does raise again an issue that others have made in the past. It is getting harder to create good assets for Trainz.

This is undoubtedly because the graphic display system is getting better and more demanding, along with our expectations. The assets made by creators in the early days of Trainz, when the CD contained a copy of GMax, are, for the most part, simply not good enough any more. Certainly the time involved in creating assets is growing and the required expertise is increasing. The changing standards makes it even harder for casual creators to keep up.

There could come a point when asset creation is in the hands of professionals only, and that almost certainly means increasing the cost to the end user. Crowd-sourcing and commissioning specific works will only go so far. We need another solution if we want to keep this hobby affordable.

My thoughts.
 
dangavel said:
He did post instructions I believe , I am going to look into it once I've finished this route, however it may limit the types of vegetation that can be created as it's mostly trees that are blueprints. I am also looking at treeit as a possibility and have contacted JVC to see if he could be persuaded to return to veg making if he had some financial support.
Click to expand...

Instructions? See this thread https://forums.auran.com/trainz/showthread.php?154750-JVC-BLUEPRINTS-(create-your-own-vegetation)

Wasn't aware Jan had to packed up, that's bad news.
 
grazlash said:
Speedtrees rock, because they use a library I can have 300,000 of the same tree on our route and still get over 30fps. Trainz uses v6 speedtree, where v8 or 9 is current. I'm not sure of the advantages of N3V moving to the latest version would be, but the disadvantage is that we would have no trees for a while, then have the same issue with incompatible v5 trees in TS12. Be careful what you wish for!
Try Plant - Sagebrush. It's a clutter item so you'll probably need to picklist it in CM and then select it with a double click from the picklist in surveyor. It may not show in the asset list but will still apply. It's variable height as well. Come to think of it, theres cactii as well.
cheers
Graeme
Click to expand...

I wish it would be so in general, but it isn't.
Excessive use of speed trees will slow down the system, 300.000 is not a large amount of trees.
Testing speed trees vs. card board trees did show a clear result on my map which is carrying a few million trees.

Speed trees are just good for close range, for large forest areas card boards are the far better solution.
 
Testsieger said:
I wish it would be so in general, but it isn't.
Excessive use of speed trees will slow down the system, 300.000 is not a large amount of trees.
Testing speed trees vs. card board trees did show a clear result on my map which is carrying a few million trees.

Speed trees are just good for close range, for large forest areas card boards are the far better solution.
Click to expand...

It depends upon who makes the "carboard trees", I like that name!

There are some carboard trees that are so overly done with so many double-sided textures that they drag the system down worse than a gazillion Speed Trees. There are also so many out there that don't have LOD that the pop up suddenly and cause awful stutters.

The problem with the cardboard trees is they rely on the CPU for 90% of their processing where as Speed Trees use the much faster GPU with its fast pipelines and specialized ALU (Arithmetic Processing Unit), and processor to quickly render the trees. The GPU in general is much, much faster than the CPU even today, and in fact NVidia has created specialized processing boards for use in proprietary engineering systems that work on major projects requiring a high number of calculations.

With the carboard trees, they need to transfers from disk, program, to a memory buffer. From the memory buffer, they go to the CPU for calculations, and then back to memory for transfer to the video card which then takes all that and sends out the data through its framebuffer to the display. These extra steps take time even in computer terms.

Speed Trees also use a special code library that contains primitives which are rendered very quickly in the GPU, and use DMA calls to talk directly with the GPU, which takes that extra path away from the system. It goes without saying, that having a lesser GPU will mean that having a lot of Speed Trees will slow things down. This is no different than having a lot of carboard trees, it's the fact that you can still have a lot of Speed Trees and not have trees disappear because the computer can't render them is a plus. With Speed Trees, there are adjustments to take care of performance issues such as lowering the quality level. Setting the quality level to "normal" is actually a good balance as it renders the trees cleanly and doesn't melt the computer.
 
JCitron said:
The problem with the cardboard trees is they rely on the CPU for 90% of their processing where as Speed Trees use the much faster GPU with its fast pipelines and specialized ALU (Arithmetic Processing Unit), and processor to quickly render the trees. The GPU in general is much, much faster than the CPU even today, and in fact NVidia has created specialized processing boards for use in proprietary engineering systems that work on major projects requiring a high number of calculations.
Click to expand...

John billboards are NOT rendered by the cpu all assets are rendered by the GPU, the CPU rendering was only applicable to TS12 and lower which rendered everything through the CPU. Think about it how the heck does the game engine tell it a tree and not a house or other scenery asset.

Can we put this false information to bed please.

I did a test with Speedtrees v billboards using my TS12 and TANE FR routes in TRS19,

Speedtrees 30% CPU
Billboards 29%CPU
 
clam1952 said:
That was Harold hminky, using jankvis's normals fix, unfortunately Harold had to give up due to serious health issues.
Click to expand...

clam1952 said:
Instructions? See this thread https://forums.auran.com/trainz/showthread.php?154750-JVC-BLUEPRINTS-(create-your-own-vegetation)

Wasn't aware Jan had to packed up, that's bad news.
Click to expand...
I wasn't aware of that either and I just heard about Harold. I wish Harold all speed in his recovery and my thanks to Jan for at least giving me some hope for non speed trees going forward.

Bob Pearson
 
Last edited:
You must log in or register to reply here.

Similar threads

D
Ok NV3 i'll ask again, any plans regarding vegetation ? .
3 4 5
Replies
81
Views
5K
dangavel
D
G
Help me restore TRS19
Replies
6
Views
324
Forester1
F
J
The whimsical physics of the Trainziverse
Replies
16
Views
579
chris2001trainz2010
C
M
  • Locked
Error posting threads to the Suggestion Box-car?
Replies
3
Views
125
oknotsen
O
A
Explanation about Trainz Forum's threads
Replies
7
Views
267
oknotsen
O
Back
Top