My name is Paul Hobbs, and I am a creator for Trainz, pretty much exclusively of British locomotives and goods rolling stock (my one attempt at a passenger coach is still not finished years later).
I moved to Germany in 1989 to work in the Automotive industry, doing various projects for Audi, BMW and Rolls-Royce and am now living in Munich and happily retired since October 2021, although I found out that doesn’t necessarily mean more time for content creation.
I have always been interested in railways and I was also interested in ‘home computing’ as it was known then and purchased my first computer in 1981. From 1984 onwards I was involved in creating computer models of automotive parts for various car companies, so it was a logical step to combine the two interests. In fact, I was thinking about something like Trainz long before it became technically possible.
As soon as I heard about the release of the first version of Trainz, that must have been in September 2001. Especially attractive for me was that it was marketed as a model railway simulator as I was approaching content creation from that direction – it was just what I had been waiting for.
Even better was the possibility of creating custom models using GMax, so I started learning that even before it was possible to export anything. My first model was a Class 04 diesel shunter, cunningly chosen because the tramway skirts meant I didn’t need to animate anything. This was uploaded to the DLS on 31st March 2002 and was one of the first third party locos to be released.
It's so old I don’t think it is even possible to get it in to later versions of Trainz as it predates cdp files, but it is still listed.
This screen shot is of a later version, but the mesh is basically the same.
I note with some amazement that the animated version of this loco without the skirts was uploaded on 19th April 2002, so I must have worked out how to do that quite quickly. It was a very long time ago.
Once that problem was out of the way I was off and running and able to produce pretty much anything I wanted, which I proceeded to do. I suppose that if I had concentrated on one theme (such as the LMS in the 1930s), I could have released more models, but I like all British prototypes, so I just made whatever took my fancy.
Since 2007 I have been working solely in Blender, which is much more suitable than GMax and is in constant development. I didn’t find moving over to Blender much of a problem, possibly because of my background.
I think the model that I was most satisfied with when I first saw it in Trainz was the BR Britannia included with the original S&C package. This was complicated and to be honest, was at the limit of what I was able to achieve at the time, which must have been about 2005-6. I remember that the boiler took five attempts to get it acceptable and the mesh bears the scars of the struggle to this day.
It is also worth noting that the much-improved lighting and the shadows in TS19 make even earlier models like this look much better than they did in the Trainz version for which they were developed, even though it now lacks the shininess which it originally had. Despite this, and its faults, I’m still very pleased with how it turned out.
I think it would have to be solving a new problem, something that I haven’t had to do before. In fact, I’m much more interested in finding the best way to create a model than finishing it, which would explain the 60 or so locomotives I have 95% completed and most of which have never even been exported to Trainz.
Another reason for the lack of releases from me is my attempts to put down in writing what I have discovered during my experiences with Blender since 2007, in the form of various tutorials. I would rate these as more important than any models that I have released, as hopefully others will use them to create more assets than I possibly could. Plus, if I don’t document what I have found out I’d probably forget it.
That’s easy – cabs. I hate doing cabs, as there is a lot of modelling required and it is hard to find drawings and photos for reference. You are also very close to the mesh in the interior view and if there is not enough detail, or the textures are not that good it is obvious. I have visited the National Railway Museum in York a few times and when you can see what a real steam loco cab looked like close up you can’t help wondering if it will ever be possible to make something similar in Blender in a reasonable timeframe.
Another problem is that you can be looking at parts of the exterior mesh very closely as well, for example the front of the tender if you have a camera positioned outside of the cab looking backwards, and what looks ok in the normal exterior view tends to look crude. This means having to add more detail to the exterior mesh as well.
The amount of time to build a good cab is comparable to making the exterior model and I usually don’t drive in the interior view anyway.
My first major influence was David Dallaston (Pikkabird), whose Class 37 diesels and Jinty were a lot better than anything I’d produced up until then, especially the texturing and how he did the LOD meshes. I learned an awful lot from looking at these models.
I’d also mention the late Andi Smith (andi06), who showed a long time ago what Trainz was capable of and Ed Heaps (edh6), who has produced at impressive speed a long list of interesting locos that I would have liked to have built myself and now feature PBR materials making them look even better.
Looking back, the main influence on what I’ve been doing is seeing a model that made me think that I’ll have to raise my game several notches to compete. One of the first UK models that I saw that really showed off what PBR materials could look like was the SECR C Class from Camscott, and that meant I’d have to bite the bullet and learn how to do those materials, having resisted that step for a long time.
An official implementation of something like the late Andi Smith’s SuperScript library. This would save a lot of time when scripting an asset and stop creators from re-inventing the wheel. This should include texture/mesh swapping, headcode lamps and ACS couplers and pipes. SuperScript is no longer maintained as Andi is no longer with us unfortunately, and I believe the source code is not available.
One aspect of content creation that is (for me and I suspect a lot of others) exceedingly tedious is calculating where the LOD changeover distances should be. Trainz wants to have the changeover distances expressed as a percentage of on-screen real estate for each mesh instead of actual distances from the camera.
Changing this behaviour would probably be tricky, but the ‘Preview Asset’ tool could display the screen percentage value for each mesh for the current distance to the camera, as it must have this information internally. You would set the required distance to the asset and then read off the values that the *.lm.txt files require for a changeover at that distance.
I’ve already shown some of the models that are almost complete on my website http://www.44090digitalmodels.co.uk, but here is a shot of the LMS Crab 2-6-0 which is one of my latest models and has some big improvements over my earlier efforts.
One of the changes is to use Substance Painter to generate the textures, which allows for easy weathering, unevenness of the surfaces (very important if you want the model to look realistic) and subtle variations of glossiness for painted and bare metal parts. Substance Painter you must pay for, but there is the free Quixel Mixer which does much the same thing. Also, I’m now making much more use of baking normal maps for surface detailing such as rivets to save on the poly count.
I’m still learning the software, so the final version will probably look a bit different, but by using PBR materials and this software it is possible to get realistic metal surfaces which were very difficult to get right before the advent of PBR. To be honest, the appearance of all assets in versions of Trainz before T:ANE was, to me, quite disappointing with the flat lighting and the lack of shadows.
This model also has a lot more detail than I have included in the past as expectations have risen considerably in the last few years. There will need to be quite a bit more added to it though, having seen the screenshots of the latest DLC coming from N3V…
Here is an enlarged view of part of the tender, showing the sort of surface detail that is possible now. [A close-up of a machine Description automatically generated with medium confidence]
I should have added more soft edges to the axlebox guide castings and the springs…
There is always something extra to add to a model, another reason why I don’t release many of them.
I have also been working on updating the locos included with the Potteries Loop Line package to include more detail and PBR materials, such as the Garratt loco.
Another model that I am quite proud of is this APOC twin tanker, which was textured using the free Quixel Mixer. Railway modellers of a certain age will know that there was a model of this wagon available in the 1960s from Bob Wills and these seemed to crop up everywhere on layouts featured in the popular magazines, probably because it was so unusual.
In reality there was only one of these, which astoundingly survived until at least 1978. One of the great things about Trainz is that it is now possible to build models that you wanted to have over 50 years ago but which were unaffordable then (or so fiendishly complicated to build that you wouldn’t have been able to make a decent job of it even if you could have afforded it).
I sometimes wonder what the ‘greats’ of the model railway world – such as Peter Denny, Frank Dyer and many others whose names mean little to most people these days – would make of Trainz. I suspect they would embrace the possibilities rather more than we do.
Learning content creation for Trainz is like learning a second language (and I should know, having lived in Germany for over 30 years). Both require a lot of time and effort to master, but exactly like a second language you can get away with knowing very little at the start. Once you have the basics learned most of what you subsequently learn just makes things a bit easier.
The important thing I think is to stick at it and practice, the first models probably won’t be very good, but the beauty of Trainz is that anybody can have a go at content creation. There’s plenty of help available on the N3V Forum and Discord, and I’m not above asking for help when I get stuck – and that does happen even with the amount of experience I have.
I did see an interesting video on YouTube a while ago now that was about learning to draw which had a good tip – draw at least one line a day. Translating this to 3D modelling would be to create one face a day, the idea being to keep in practice, as of course you certainly won’t stop at one face a day…
So, don’t be afraid and give it a go, it’s very rewarding, there is always something new to learn and you can do all of this for free, no need to buy any software at all for content creation.
If you’ve made it to the end of this piece, then thank you very much for reading this far!
