Article: Content creation for Trainz: A New Era

[AntonyVW]

Regarding "secrets" I can assure you it has not been our conscious intention in the past. However, our approach for T2 will be to explain more, and involve the community more from the very beginning - as you will see in a few short days with our project on Kickstarter.

The problem some of us are having though is that we dont feel the community is involved from the beginning. We keep being told to wait for Kickstart to us that is not in from the very beginning but in when the decisions have pretty much already been made.

 

[WindWalkr]

No doubt memory is no problem, even now, but won't the huge files just lead to very sluggish load times and hence even more of the stuttering which has become a Trainz hallmark? Or will SSD's be the minimum recommended storage device?

I don't think it's quite at the 'min spec' stage yet, but SSD and hybrids are certainly becoming common and we can assume that it will be the norm before E2 reaches middle-age.

Huge files don't necessarily increase loading time, because the game isn't forced to preload the entire file. Regardless, E2 will definitely improve on streaming performance- that's one of the big reasons behind moving to a 64-bit, pervasively-multithreaded engine.

chris

 

[andi06]

At least in the short term SSDs are many times faster than HDDs but they are also many times smaller so, unless you pay some attention to limiting the bloat, TRS won't be given space on everybody's SSD.

 

[sura]

All i am missing heavily right now is an up-to-date CCG with all possibilities opening up with the new engine. This complete manual was failed to release for TS12 eg. so most of our objects may not use all the built-in possibilities; of which LOD is only one - may be important - possibility. Me personally am waiting since more years for such a manual - i do not want to waste my time for creating objects being obsolete in a year or less.

Other point is the quality management of DLS objects. ATM only config.txt passes kinda controls before being released on DLS, although no actual manual; there is no visual control of childish/oversize/unnecessary textures, unnecessary 2 sided texturing, use of not visible polygons - which all costs performance and toy-looking assets.
Being so behind in your own manual release schedule makes me feel, you do not have enough time for that activity. May we trust you this new engine will be released with a valid Content Creator's Guide?

 

[Tony_Hilliam]

The problem some of us are having though is that we dont feel the community is involved from the beginning. We keep being told to wait for Kickstart to us that is not in from the very beginning but in when the decisions have pretty much already been made.

Antony (my name as well btw, with no 'h'), I think you will see why Kickstarter is the "beginning of a new era" once the project is live. In reality, engine research and development has been going on for years, as have our plans for "the next Trainz", but Kickstarter helps make it all a reality.

 

[davesnow]

Why does everything have to be so complicated?

 

[Kaur]

I am with the words of Tony, things have a process, without that process development equal would not be suited to all of us, that we are looking for perfection and something more.
From my point of view, the idea of Kickstarter is the most appropriate and depending on the demands of the new engine is ideal for that between all help to improve the taste of us is a great opportunity, that in these moments N3V has benefited from her community for that project, which support and tell Tony privately, my congratulations Tony.
(Always taste the colors a phrase made Spain). But I still believe it is our opportunity to do well and the taste of the consumer.

Take care.

Jordi (Kaur)

 

[Dinorius_Redundicus]

Yes, but, does your color map always need to be so large ?, is it required to be the size as your normal and spec ? why use a 1024 or 2048 color map if a 512x512 will suffice and you use the channels wisely.

Well, of course I use the smallest image size that still looks OK, that hardly needs saying. What I am saying is that in many cases - especially photo textures made from composite images, where all the surfaces are packed efficiently into one image - the larger format is necessary if the asset is to look any better than the fuzzy things we had back in the Trainz 1.3 or UTC days.

My point was that whatever size the diffuse image is, it will bloat by a factor of 4 or more if you add a channel for env reflection, then a normal/spec map, then double it by adding a snow mode. And that's not counting LOD meshes, nightmode, whatever.

I don't dispute that objects made with tbumptex or tbumpenv materials can look splendid, but there is a cost. It's probably justified on locos, rolling stock and trackside objects because they are likely to be viewed at close quarters, but I really question the wisdom of saying that almost everything needs these materials. This invitation to unnecessarily bloat your creations seems at odds with N3V's hard line campaign (justified in my view) against monochrome textures.

 

[Pendolino]

I don't dispute that objects made with tbumptex or tbumpenv materials can look splendid, but there is a cost. It's probably justified on locos, rolling stock and trackside objects because they are likely to be viewed at close quarters, but I really question the wisdom of saying that almost everything needs these materials. This invitation to unnecessarily bloat your creations seems at odds with N3V's hard line campaign (justified in my view) against monochrome textures.

I fully agree with Dinorius Redundicus: on one hand, we are told to make content as efficient as possible (no unnecessary polygons, LOD, as few textures as possible, no monochrome textures), and I agree with that. On the other hand, however, some of the new requirements for content creation - such as mandatory normal and reflection maps, look like a waste of resources, as nobody will ever see their magnificent look.

 

[norfolksouthern37]

some of the new requirements for content creation - such as mandatory normal and reflection maps, look like a waste of resources, as nobody will ever see their magnificent look.

how in the world did you get this out of anything?

 

[WindWalkr]

I don't dispute that objects made with tbumptex or tbumpenv materials can look splendid, but there is a cost. It's probably justified on locos, rolling stock and trackside objects because they are likely to be viewed at close quarters, but I really question the wisdom of saying that almost everything needs these materials.

There is always a cost. If you look at current-generation buildings vs locos, you will immediately notice that a typical loco has far greater mesh and texture detail than any building. So while you're correct that adding a normal map to a building will increase the disk cost of the building, you're incorrect in your assumption that this will make it anywhere near as expensive as a loco.

The nice thing about normal maps is that they can substantially improve the scene with relatively little effort on the creator's part, and relatively little runtime cost. You should get into the habit of creating a normal map for everything.

This invitation to unnecessarily bloat your creations seems at odds with N3V's hard line campaign (justified in my view) against monochrome textures.

Both of these have costs. The difference is that one of these has a potentially massive visual benefit, while the other literally has none.

chris

 

[Dinorius_Redundicus]

..you're incorrect in your assumption that this will make it anywhere near as expensive as a loco.

The nice thing about normal maps is that they can substantially improve the scene with relatively little effort on the creator's part, and relatively little runtime cost. You should get into the habit of creating a normal map for everything.

Both of these have costs. The difference is that one of these has a potentially massive visual benefit, while the other literally has none.

chris[/COLOR]

I didn't assume a building would be as expensive as a loco, only that it would be more expensive than it needs to be. In many cases adding a normal map does not give a 'massive visual benefit'.

I think you are wrong in saying a normal map requires relatively little effort from the creator. Yes, anyone can make a normal map quickly, but making an effective normal map, one that works visually the way it should, is often a nightmare when working with photo textures. You should try it sometime.

However, if that's what you want, I will do it.

 

[norfolksouthern37]

I didn't assume a building would be as expensive as a loco, only that it would be more expensive than it needs to be. In many cases adding a normal map does not give a 'massive visual benefit'.

I think you are wrong in saying a normal map requires relatively little effort from the creator. Yes, anyone can make a normal map quickly, but making an effective normal map, one that works visually the way it should, is often a nightmare when working with photo textures. You should try it sometime.

However, if that's what you want, I will do it.

Adding at least some normal info to even a photo texture is not really a nightmare. You should be able to discern what areas should be raised or not based on the texture, but the per pixel specular should at least be taken advantage of otherwise everything looks plain and plastic-like. Even where I have used photo texture, the normal map may contain little if any info, where the spec (alpha) is where the real benefit comes from. Ensure that your object can interact with the light better than just plain flatness or shininess. Study how real-world objects react to the light and how different materials you observe look with respect to shading and specular elements. It might sound silly, but once you start looking for that kind of thing you really notice it.

 

[WindWalkr]

Adding at least some normal info to even a photo texture is not really a nightmare. You should be able to discern what areas should be raised or not based on the texture, but the per pixel specular should at least be taken advantage of otherwise everything looks plain and plastic-like.

Well said.

To clarify my previous comment: there are certainly degrees of effort with content creation, and you can pour unlimited time into getting something "just right". I'm not intending to be dismissive of the people who are willing to go to great lengths. What I mean is that even a simple normal map can offer substantial benefits over a plain diffuse, in the same way that even a simple texture can offer substantial benefits over an untextured model.

This is all true right now, but you can reasonably expect that the normal map will become even more important with E2, simply because the lighting model is better and the default shaders are more detailed. There are a lot of rendering techniques that we can utilise that all draw their information from the normal map.

chris

 

[johnwhelan]

So he is saying "follow the guidelines" and if you want to do something "a bit different", ask questions and think about the ramifications before rushing off creating a brand new "feature".

Given the lead time it takes some of us to create content it might be nice to have some reasonable documentation and something to play with. With the present engine I've seen very little difference with normal mapping for example so haven't used it recently.

Thanks John

 

[amigacooke]

Why does everything have to be so complicated?

It involves people

 

[ATSF854]

Why does everything have to be so complicated?

If you take time (it takes various amounts of time for certain people) and work out everything over multiple projects it becomes standard to you and way less complicated.

Also, there should really be an emphasis on bump/spec maps Especially quality spec maps. With a spec map you can make sure that bump maps don't stand out too much and that objects are not shiny in shadowy areas like on models that have baked textures.

 

[pcas1986]

Given the lead time it takes some of us to create content it might be nice to have some reasonable documentation and something to play with. ...

Thanks John

At the risk of beating a worn out drum, I heartily support that. In fact my monetary support for the KickStarter thing will be significantly influenced by it. This is after spending a few hours today looking for some container information and finding a related thread complaining of the same thing three years ago!

On the issue of large images. I'm helping a young creator who has gone overboard with multiple images, all of which are very, very large. The commit time was so long that I deliberately went and resized all the images to a smaller size. I couldn't really detect much difference in quality and the mapping still worked fine.

Cheers
Paul

 

[Scitimar]

Why does everything have to be so complicated?

And to add to the comments by ATSF854 re: the importance of good spec maps there's more than a few tutorials available both free and pay that deal with this subject.

With the move to a new engine for Trainz ( DX11 based, thank the gods of your choice :udrool: ) there's no better time than now to look at what and how things are done in other games and freely available engines, it's not something you'll regret.We know it's not easy, but, with perseverance the results are worth the pain.

 

[grupovat]

I think that should have more tutorials available and open source, both for the treatment of textures, as the process for its use in the new E2. that way the community would appreciate it greatly.