Andi if you have used this work flow can you give me more details
I did not do much more actually.
I use the layers a lot - this array of buttons that switch visibility of objects on and off. Remembering what you put where is a mess but they are handy when things are combined with each other. In your case, you could, e.g., have the station sign without letters in layer 1, then the name of each station in layers 2, 3, 4, ..., each name in its own layer, all in their final place. Then you switch on layer 1 + 2 to export the first name plate, layer 1 + 3 for the second and so one. The bonus is that you can still work on the plain board in layer 1 and any improvement there will carry over to all stations, if go again through the exporting of each.
I find it important to always consider the difference of "copy linked" and "full copy" of objects. The names might be a bit different depending on Blender version and tutorial you read.
The difference is that in the linked version, both object instances share the same mesh. So if you change the mesh, the improvement is instantly carried over to all the instances.
The down side of linked copies is that sometimes you need little differences. If you created the copy as linked, just select "make standalone" or something like that. Then it has its own mesh copied over and you can modify it without influencing other things.
When I say mesh, I actually mean "mesh, UV map, material etc.". Basically, if you need some difference, you cannot use the linked version. End of differentiation.
If you feel like chancing the artistic side of things, you can try to make the texture look not so plain. I mostly failed in that but I am not the persevering type of person. Blender has this spray can feature. You can spray on the texture and also on the 3D shape.
Most important thing: Do it so gentle that you feel that it is not really worth the effort. Then it will be just about right. If you think (looking at the Blender model) that the time you spend on it clearly shows on the model, then people will say it is overdone. Exporting the model into RW and looking at it from "usual perspectives" is important, whatever perspectives are considered usual.
The key problem is that when you work on something in Blender, it occupies all your mind and it is easy to spend too much time on details that are irrelevant while missing something else that spoils the overall impression.
RW has some handy syncing feature whereby you do not need to leave the editor to see the freshly re-exported shape in-game. I forgot all about it as I dropped the artist role pretty fast.
When you feel that the detail you added just look stupid to be repeated on each panel, there is a bag of tricks to fight this impression. The cheapest ones are:
1) Mirror the UV map of the panel for a random selection of panels.
2) Move the UV map for a copy of the panel to somewhere else on the UV map so you can have two different, independent textures on your panels. Again randomly assign some panels to version 2. Together with item (1), we get 4 different looks on the panels which should do for 10 or so instances of the panel at any given platform.
Both tricks together keep you clear from tileable textures. These are an interesting thing to learn, too, but somewhat headache-intensive.
You could try the curve modifier one day. It will be great for curved platforms and retention walls and anything else that is not dead straight. Basically, you create Bezier curve as one object. Then you bring it in with the array modifier if you want stuff repeated along the line. If you apply the curve modifier alone to an object, it gets distorted along the curve, which can be right for your case, too. Sorry I forgot the rest. Too many years since and no time to Google for the details.
Love the Nerd info but the brain not computing today
I tried to answer this question for myself, long ago: What is so small that I really should not bother modelling it?
The simple answer is: Forget modelling stuff that is S wide and when you just see it from D metres away:
S = 1 cm ... D = 40 m
S = 1 inch ... D = 100 m
S = 1 foot ... D = 1200 m
The above is the upper bond of relevance. The lower bond is maybe 1/3 or 1/4 or so of this distance. This is really hard to establish. It not only depends on the graphics settings and display size, but also on how much it stands out.
Leaning out far, I could postulate: Do model details that are S wide when they can be seen from D metres away:
S = 1 cm ... D = 10 m
S = 1 inch ... D = 25 m
S = 1 foot ... D = 300 m
I would be glad to receive feedback from anyone who bothers as I know that this is a very theoretical approach. I really developed this back then to find an answer to the question "how big should a pixel an the texture be on the 3D model". It is answered along the above lines.
My conclusion back then was:
1 px for 1 mm object surface seen from 1 m away.3 px would be great on high-res displays but in other settings, even 1 px may be much. After all, 1 is a number you remember easily.
You can write the same rule as:
Pixels on texture = Object size in mm / viewing distance in m.Multiply by 2 or 3 for important details and/or to be "future proof".
Final nerd note: All the above is based on the default setting of the camera in RW. You can change the field of view. But I simply assume that people will carry the cost themselves. In other words, if you switch on telescopic view to detect anomalies in mid-range scenery, I cannot help you.
Not so nerdy observation: Since people can fly around freely, the whole distance computation boils down to what you consider reasonable. But in my view, the track location (or the head-out view on the near side) should set the standard.
The dramatic impact on station textures that TSW's rambler mode has is something we will worry about in 2019.
Edit: Viewing distance is a divisor, not a multiplier.