Tutorial: Getting the most from PBR and why you should be using it in TRS19

Quite a few Trainzers are really really cash strapped. It's a hobby that you can have for very little outlay. Even being able to add a GTX 1060 to an existing machine may not be possible. Laptops for example, some of the small form factor PCs with small none standard power supplies. I think N3V did a survey of hardware some time ago and the average GPU was much lower than I expected. So the target market for PBR is going to be the top 10-20% of the current users it certainly isn't for everyone and to me that means niche.

I'm afraid after reading your posts I'm not much clearer on what PBR is. I use photos quite often as textures but I've no idea if I can do that with PBR. What I call the heavies seem to be using some rather expensive software to create the textures. I even picked up a couple and after playing with them I'm still none the wiser.

No doubt PBR will come in time but I don't think it is absolutely critical at the moment. It will take time to understand it and come up with tutorials. Recently someone asked what software should they use for creation. Probably Blender but Blender 2.8 tutorials for trainz are thin on the ground. Blender 2.8 with PBR well yes but for a beginner. I don't know but I suspect it is not ideal especially judging from some of the comments in this thread. I'm not sure the infrastructure is in place.

TS19, it also has fewer bugs in it, and doesn't wonder off into database rebuilds quite as often as TANE besides the PBR stuff.

Whilst I understand in the US Military service is held by many to be marvelous it sometimes can make difficulties for an exmilitary person who is used to command to soften their approach when dealing with civilians and that is when you maybe misunderstood and be misunderstood as arrogant.

Cheerio John

I get it John.

(a) You don't have a very capable computer system.

(b) Your system cannot handle the graphic demands of PBR with the Shader set to Ultra.

(c) You cannot afford to change the above.

(d) There are others like you out there.

In the above case, obviously nothing I can say about the merits of PBR will change things since you cannot run the Shader set to Ultra anyway.

I would have thought common sense would have prevailed for anyone reading my tutorial, but for those who need everything spelled out, I have added the following to my original tutorial post to clarify matters:

Caveat: This tutorial is directed towards those of you who have a computer system capable of running TRS19 with the PBR Shader set to Ultra, but may not be doing so at present for some reason. For more on the minimum TRS19 hardware requirements for various levels of computer graphics see the following link:

http://www.trainzportal.com/blog/view/trs2019-hardware-requirements

BTW here is a better January 2020 survey from Steam on graphics card usage out there based on a much larger gaming community then just Trainz users:

https://store.steampowered.com/hwsurvey/videocard/

I really want to move the conversation in this thread back to those who meet this caveat and can use PBR with the shader set to Ultra but may not be doing so at present.

Bob
 
1890-1940 UK items please.

Thanks John
I second that, but make it US as well , another factor that most people doesn't seem to consider is the dearth of older high quality content items that are specifically made for 2019 and even TANE , those of us who model the steam era are starved of content , possibly because many of us are older as we grew up with steam . It's pretty easy to work making routes in 2019 if you model contemporary railroads, but anything older than 1950 , then you are struggling . As older creators such as Ben Dorsey pass away , it's going to be harder to make these older routes unless new blood steps in to take their place . for instance, I could NOT make my Uintah route to take advantage of PBS tech as most of the content was created for trains pre TANE , including the rolling stock and locos . In addition , I would have to recreate all of the thousands of textures and splines I've made that are specifically designed for the route , it would literally take another few years to get it to work in game so it looked as good as the new technology would allow . Is it worth it to postpone its release , or is it better to make it available sooner , even if it doesn't look as classy as it could using PBR and Turf FX ?

We seem to have a fair few creators who make older locomotives and rolling stock ( although not all of it to 2019 standards ) , but not too many who make the other items that are is essential to make the locos and rolling stock live in context.From what I can see .It's going to take many years for the assets that we have available pre 1950 to be replaced and updated to 2019 tech levels, as for the US narrow gauge items , I would be very surprised if it ever happens. Is there anyone who works in US narrow gauge making 2019 level freeware items apart from pencil42 ? I am not knocking new developments, but just pointing out it a hell of a lot of work to get the assets upgraded to take advantage of the new developments, especially if you work in niche areas such as pre war narrow gauge , or railways other then the US or UK. Now, if someone was prepared to do it for me, then I'd make the move, but not having the time or skills to do so, you can see why it's not the way forward for me, even though I DO have a system that could run the new tech to take most of it's advantages to the full ( I can't use turf fx as I have to use an AMD card on my Mac. )
 
Last edited:
...
I'm afraid after reading your posts I'm not much clearer on what PBR is. I use photos quite often as textures but I've no idea if I can do that with PBR. What I call the heavies seem to be using some rather expensive software to create the textures. I even picked up a couple and after playing with them I'm still none the wiser.

Try reading the first few pages of The PBR Guide - Part 1 by Wes McDermott (Allegorithmic). It talks about light interaction with a surface and helps to understand how the software emulates real life conditions. Wes is one of Allegorithmic video authors and very easy to listen to in his tutorials.

If your concern is about the packaging of Trainz PBR materials then perhaps this might help:

The albedo texture is much like the diffuse but without any specular or AO. It is simply the colours used by the model.

The normal texture is the same as previously except the alpha channel is now used for height information. For most models you don't need height information but they are commonly used in ground textures. Use sparely as height information looks odd at acute angles. This is not a Trainz issue but inherent in the technique.

The parameter texture is a package of four 8 bit greyscale channels: emission (R), roughness (G), AO (B) and metallic (A). Packaging this information this way may sound odd but it is common in the gaming industry, presumably to save image load times and memory. You need some software that will package those channels for you. GIMP will do it but it is clumsy. I've found a new Blender addon called Bake Wrangler that will do it as well and am currently working on a workflow that I may publish as a tutorial.

So, how do you create those those channels?

AO you can bake - nothing new there.

You can create the emissive, roughness and metallic using the original information in an image using Materialize.

Or, with the use of software such as Substance Painter, you can paint the colour, roughness and metallicity all at the same time. Just select the channels to be used and adjust the intensity as the same time. That way you can have a nice brand new like painted surface, and then add some brown/red rust as a layer with variation in the roughness and metallicity. Rust would be much rougher in feel and almost direlectric (no reflection).

You can also paint the emissive/emit value as well but unless you want your asset to glow then plain black is a good choice.
 
I use photos quite often as textures but I've no idea if I can do that with PBR. What I call the heavies seem to be using some rather expensive software to create the textures. I even picked up a couple and after playing with them I'm still none the wiser.

John, you might have a look at the Photos-to-PBR section in this thread.

https://forums.auran.com/trainz/showthread.php?156660-PBR-Texturing-Workflows&

The workflow might look a bit complicated but if you follow it step by step you can convert a photo or a diffuse image, i.e., a textured UV map, into the albedo, normal and parameter maps required for PBR texturing in TRS19. The workflow automatically combines the height map with the normal map to include parallax, and combines the four maps needed to produce the parameter map.

As for an explanation of PBR perhaps this might help.

http://www.doug56.net/PBRMaterials/


Cayden
 
John, you might have a look at the Photos-to-PBR section in this thread.

https://forums.auran.com/trainz/showthread.php?156660-PBR-Texturing-Workflows&

The workflow might look a bit complicated but if you follow it step by step you can convert a photo or a diffuse image, i.e., a textured UV map, into the albedo, normal and parameter maps required for PBR texturing in TRS19. The workflow automatically combines the height map with the normal map to include parallax, and combines the four maps needed to produce the parameter map.

As for an explanation of PBR perhaps this might help.

http://www.doug56.net/PBRMaterials/


Cayden


Thanks I think we'll get there but it needs the bits in the right place.

Cheerio John
 
Thanks I think we'll get there but it needs the bits in the right place.

Cheerio John
Just had a look at the 42 minute tutorial to make pbr textures from photos, this is so much more complex than the present process , generally using photoshop and crazy bump one can make a decent texture in 20 minutes or so, this is infinitely more time consuming, involving picking up elements of blender , which is a non user friendly program if ever there was one. It's gonna freeze out many people who at the moment can make reasonable textures. The process needs considerable simplifying if you are going to take people who can manage to cope with something like photoshop, but who find the complexities of 3D beyond their skill set, or as in my case , just don't want to spend that amount of time to learn the basics of a 3D app, and then spend much more time to make something that looks a bit better, but not that much better than what they can already produce in half the time using much less complex software. There's only so much time , especially when one is in the sunset hours of life, one needs to prioritise what activities are more important, I'd rather make a route that looks a bit less realistic than spend my time creating some better looking items, only to snuff it before I complete the task.
 
Its been a long time since I was called arrogant by someone

Sorry if I caused offence, however it was the assertion my comment was aimed at, not the person making it.

As regards why I shelled out 55 quid for TRS19, well you could argue I was simply seduced by the N3V hype. As with every new release of Trainz, one hopes that certain features improve and TBH PBR textures wasn't even on my radar when I parted with the cash. In any event, if you hit the sweet spot with TRS19 post processing settings, then existing content can look much nicer without having to necessarily dabble in the high end stuff.

I will continue to build the routes I like to whatever standard I feel suited to my time and ability, with the content and terrain textures that do the job. At the end of the day it is for our own personal pleasure and if others want to condemn the end result in whatever sim, down to them.

My last word on this matter.
 
For some time now I have been tackling the use of PBR with structures. See this screenshot below:

PBR-Structure-Comparison.jpg


On the left is the original Ben Dorsey Gettysburg Station with the old 2D textures and a concrete base that doesn't work well with the new 3D PBR ground textures (they are poking through the concrete as you can see). On the right is my re-engineered version of the same station using 3D PBR textures and a concrete base that is thicker (Z=0.20 meters). The sheet metal roof part of the structure is much more realistic and better looking in 3D as you can see.

I am re-engineering a number of Ben Dorsey's structures to work in a 3D environment. One of the common problems I have run into is that the original base of the structure needs to be slightly thicker so that the 3D ground textures don't pop up through them. I have found that a base with the z dimension set to at least 0.20 meters works well for doing this.

Bob
 
dear bob ... as i appreciate your efforts for promoting pbr, i'm very interested in your tuts ...
dear forum ... can you please respect the original purpose of this thread as it is laid down in the title ...
dear mod ... as it was meant to be a tutorial : is it possible to separate the posts with the discussions and arguments into an independent (sub)thread ? ... please ... it makes this tutorial unnecessarily long and not to the point ...
dear master ... can we please move on with the class ... and thank you for your willing ...
grtz
daveric
 
dear bob ... as i appreciate your efforts for promoting pbr, i'm very interested in your tuts ...
dear forum ... can you please respect the original purpose of this thread as it is laid down in the title ...
dear mod ... as it was meant to be a tutorial : is it possible to separate the posts with the discussions and arguments into an independent (sub)thread ? ... please ... it makes this tutorial unnecessarily long and not to the point ...
dear master ... can we please move on with the class ... and thank you for your willing ...
grtz
daveric

Bedankt voor de zeer vriendelijke woorden!

Bob
 
Quite a few Trainzers are really really cash strapped. It's a hobby that you can have for very little outlay. Even being able to add a GTX 1060 to an existing machine may not be possible. Laptops for example, some of the small form factor PCs with small none standard power supplies. I think N3V did a survey of hardware some time ago and the average GPU was much lower than I expected. So the target market for PBR is going to be the top 10-20% of the current users it certainly isn't for everyone and to me that means niche.
Well, we could all wait 10 years until everybody has a graphic card that will handle PBR well, or we could start now. It's going to take 10 years to update everything with PBR anyway - I seem to recall somebody saying that N3V should have started again with all new models for T:ANE/TS19, so you should be careful what you wish for.
I'm afraid after reading your posts I'm not much clearer on what PBR is. I use photos quite often as textures but I've no idea if I can do that with PBR. What I call the heavies seem to be using some rather expensive software to create the textures. I even picked up a couple and after playing with them I'm still none the wiser.
That's not true at all. All you need for PBR is Blender, GIMP and the Wiki which are all free. There's a lot of talk about paid for Blender addons that should make things easier, but you really don't need them.
No doubt PBR will come in time but I don't think it is absolutely critical at the moment. It will take time to understand it and come up with tutorials. Recently someone asked what software should they use for creation. Probably Blender but Blender 2.8 tutorials for trainz are thin on the ground. Blender 2.8 with PBR well yes but for a beginner. I don't know but I suspect it is not ideal especially judging from some of the comments in this thread. I'm not sure the infrastructure is in place.
PBR is the future and it makes no sense if you have access to TS19 to use anything else - once you have seen what you can do with PBR there's no going back and it's not any harder than doing specular mapping in TS12.

Paul
 
Last edited:
Right on Paul, to John photo textures are great but light in the game does not bounce off of a photo realistically. Things like rust and dirt and just general wear and tear are not easy to get with a photo.

Photo's get you

Untitled.png


PBR gets you

Annotation-2020-02-08-183248.jpg


And more. But hey if you want to play with boxes and plains with photographs stuck on em casting square shadows have at it. I think the future wants to play with stuff that looks realistic to real life or at least as close to it as we can get
 
Well, we could all wait 10 years until everybody has a graphic card that will handle PBR well, or we could start now. It's going to take 10 years anyway to update everything with PBR anyway - I seem to recall somebody saying that N3V should have started again with all new models for T:ANE/TS19, so you should be careful what you wish for.
That's not true at all. All you need for PBR is Blender, GIMP and the Wiki which are all free. There's a lot of talk about paid for Blender addons that should make things easier, but you really don't need them.
PBR is the future and it makes no sense if you have access to TS19 to use anything else - once you have seen what you can do with PBR there's no going back and it's not any harder than doing specular mapping in TS12.

Paul

I accept that PBR is the future but "and why you should be using it" puts my back up. It appears to be directive counseling which has it's place but rarely works well with volunteers. Currently we have a fair number of content creators dabbling. The .im format has a .im editor available and I've seen a few people make use of it. Suddenly that is no longer available. Yes trainz has a high end but it also has a lot of content creators who do simpler things such as reskins. To me the community side of trainz is important. Yes some content creators take a lot more time than you or I would to create something but it is something they have created themselves and they are happy with it.

Then you have expectations. A Hornby clockwork train set from 1920 may not look highly detailed compared with what in available today but it worked and was fun to play with. I think it has already been commented that to many the difference between .PBR and an older model with normal mapping isn't that much. For many the cost in content creation time and learning curve is substantial.

I accept that Blender, GIMP and the wiki are all free. However Blender 2.8 changes the interface so the old tutorials don't work in the same manner. Changing the process is never easy, it took three years before everyone starting using exchange to book a room. There was no technical reason why it couldn't be done before it just took that long to gain acceptance.

To me trainz is not a job. I don't need to complete something to a certain standard. If it stops being enjoyable then I'll stop doing it.

Note I'm not saying that PBR is not the way to go in the future but at the moment it doesn't appear to be simple and mainstream. I was looking at .FBX since I was interested in bringing in a .FBX file into TS19. I gave up on the grounds it wasn't worth the effort and that was after looking at a sample config.txt file.

So for me this is all sometime in the future.

Cheerio John
 
John if you do not think PBR is simple you need to take a look at substance painter. In an interface not that unsimilar to photoshop or any other image software that uses layers you can create realistic pbr textures right on your mesh. I mean there is a learning curve but there are also GREAT tutorial videos on youtube that cover it...

I am gonna take a guess here and guess you do not like change very much.

When I changed workflows from a diffuse, spec bump to PBR I had to take a month and go to youtube U and I am still no expert and learn stuff everyday but my content took a massive swing up in appearance with no real change to the way I modeled.

For example the bump Bob talks about with the alpha is created AUTOMATICALLY in substance. I use the same 3 slots in 3ds max, (diffuse, glossiness and bump) just with the 3 files substance creates and use m.pbrmetal or any of the other material extensions and it is the same process for the most part...

If it wasn't simple I would not be doing it LOL
 
Most of my structure effort is currently directed toward re-engineering Ben Dorsey's old meshes designed for 2D textures to use PBR.

This isn't always easy to do as you have to accommodate the quirks of height in a texture. As an example I tried using a 3D brick texture with a height map on the meshes for the main structure and it didn't turn out well. The problem I saw seems to revolve around sharp angles of which many users have noted in a similar way with PBR ground textures applied to cliffs and other sharply angled features. You get a sliding effect as you look at the texture. I have also seen this in some cases with sharply angled roofs as well.

In some cases you may need to use a PBR texture minus the height map (ie; parallax) for certain areas of the structure that you are trying to retrofit from 2D to 3D PBR textures to make it work, unless you want to re-engineer the old structure, which may not be practical in many cases.

In my own experience so far most of the structure will not have a problem with PBR parallax textures. Just certain areas or maybe none at all. Every structure is different.

Bob
 
...
This isn't always easy to do as you have to accommodate the quirks of height in a texture. As an example I tried using a 3D brick texture with a height map on the meshes for the main structure and it didn't turn out well. The problem I saw seems to revolve around sharp angles of which many users have noted in a similar way with PBR ground textures applied to cliffs and other sharply angled features. You get a sliding effect as you look at the texture. I have also seen this in some cases with sharply angled roofs as well.

In some cases you may need to use a PBR texture minus the height map (ie; parallax) for certain areas of the structure that you are trying to retrofit from 2D to 3D PBR textures to make it work, unless you want to re-engineer the old structure, which may not be practical in many cases....

It's interesting to me you should bring this up. Yesterday I realised that Blender's Principled BSDF Shader node does not have a height input and so I cannot bake a height map with the current tool I am using. Delving further it seems that it is common in Blender to mix the height value with the normal value for a composite normal value. The net result might produce a reasonable result but is not what Trainz wants for parallax.

There are other ways of baking a height map in Blender but it seems to me that you would first have to bake a normal map, then add some height, if required, and then bake the height map separately. That's rather messy.


...

In my own experience so far most of the structure will not have a problem with PBR parallax textures. Just certain areas or maybe none at all. Every structure is different.

Bob

I recall trying height on a brick texture and deciding it wasn't worth the effort. Brick and rock walls are great for normals anyway.
 
John if you do not think PBR is simple you need to take a look at substance painter. In an interface not that unsimilar to photoshop or any other image software that uses layers you can create realistic pbr textures right on your mesh. I mean there is a learning curve but there are also GREAT tutorial videos on youtube that cover it

If it wasn't simple I would not be doing it LOL
trouble is, it's at least $20 a month subscription. You have to be making a lot of content to make it worth while and many can't afford that sort of commitment . I hate subscription formats for apps, fortunately I have the old adobe creative suite on disc so don't have to shell outer monthly shekels to the bloodsuckers.
Nice to see Bob is updating a lot of Bens old models , if enough content gets transferred I might jump on the PBR bandwagon in a few years, but only if it becomes easy to execute and cheap to achieve.
 
trouble is, it's at least $20 a month subscription. You have to be making a lot of content to make it worth while and many can't afford that sort of commitment . I hate subscription formats for apps, fortunately I have the old adobe creative suite on disc so don't have to shell outer monthly shekels to the bloodsuckers.
Nice to see Bob is updating a lot of Bens old models , if enough content gets transferred I might jump on the PBR bandwagon in a few years, but only if it becomes easy to execute and cheap to achieve.


Well IF you are taking any class at a local community college as long as you have a student ID you can get a student license yearly for free
 
Well IF you are taking any class at a local community college as long as you have a student ID you can get a student license yearly for free


I'm sure I read in the forum anything loaded up to the DLS had to be licensed for payware. Student licenses are normally restricted to non-commercial use, well at least in Wales they are. I'm not so sure about foreign parts.

Luv Karen
 
I'm sure I read in the forum anything loaded up to the DLS had to be licensed for payware. Student licenses are normally restricted to non-commercial use, well at least in Wales they are. I'm not so sure about foreign parts.

Luv Karen

Well there is PAYWARE and if you are creating that You are getting PAID and should be able to afford it. If you are creating FREEWARE with no cost to the end user that is considered NON-PROFESSIONAL at least here in the US...

Look I am just trying to help LOL I buy mine but sheesh I have spent more for lot less :)
 
Back
Top