Rather ancient carriages I have made, looking for critiques

[DominusEdwardius]

The adhesion coefficient drops off very rapidly when the wheels slip so you get a ever increasing slip which spirals out of control very quickly if you aren't careful.

[TrabantDeLuxe]

Yep, thats what I've always understood. And seen on loads of videos.

[SteveP_trains]

Have to say I am gob-smacked over the level of quality of the cab shots.
Amazing texturing, attention to detail and realism of the 3D fittings
I am struggling to get realistic brass and copper textures of the fittings I am trying to build for a Ffestiniog Double Carlie so I take my hat off to you

Look forward to more screen shots

Reagrds
Steve

[TrabantDeLuxe]

Thanks. I've tried a lot of methods to get brass to work halfway decently. I've used the TrainUberShader.fx shader, which has the nice feature of seperating specular from environmental reflections. Here's a quick texture breakdown, I'm showing part of the original 1024 x 2048 map.



So. At the left we have the diffuse. It goes in it's own RGB texture. My workflow for making a diffuse is pretty standard.

• We start of with a base colour, being the colour of whatever material we're trying to make. In the case of brass, I pick a orangy-brown colour.
• Then, I go crazy with all kinds textures I've gotten over at cgtextures.com. Set the blending mode to 'overlay', 'soft light', or one of that category. Try to use levels adjustment and maybe a high pass filter to make the overlay texture (on average) a mid-gray. Experiment with opacity and even with the blend-if options. It might be wise to save the complete group of overlays so you can quickly insert them. Kinda build your own little library of textures.
• On top of all this, we throw a AO bake. Set the blending mode to multiply.
• Optional, but I think it helps bring out the edges: Throw the AO bake over all of this, high pass filter at 2.0 px setting, blending mode to overlay or one of it's cousins and pull the opacity way down.
• Again optional, but you can see I've tried to do some dirt and grime in places.

Now we have a diffuse, on the base of which we can make a specular texture. The basic premise is that anything that you want to catch highlights is white, and dull would be black. Where I come from brass is rather shiny, so we change the base colour to a very light gray. I then like to mess about with my texture overlays, increasing their contrast by a bit. Sometimes I make them darker, in the case of say rust pitting, or dirt and grime that doesn't shine. Finally - and have a look at the radiator in your room now - edges tend to catch highlights very dramatically, so we use the AO map for that. Either use the high pass, or pull it through the find edges filter and invert. In the latter case, use linear dodge as the blending mode and pull the opacity down a bit.

The environmental reflections map is another grayscale map that governs environmental reflections. Contrary to specular, the env reflections tend to show even when there's no direct light source illuminating them. Which is nice in a cab that's generally in the shadow. The downside however, is that too much env makes everything look like it's covered in gelatin. Quick and dirty way of generating the env map is to simply darken the spec map. In this case, I've also increased the contrast on the texture overlays.

The spec map, and env map go into what's known as a auxilary texture of the TrainUberShader shader. Trainubershader uses the red channel for AO (you could add AO in here), green for emissive (glowing, for radioactive injectors), blue for environmental reflections (so chuck the env map you made in here) and the alpha channel for spec (so throw spec in there). Sciff over at UKTS did a nice write up of trainubershader.fx. Trainubershader also supports a normals map, but that's for another time. Quickest way for a normal map would be cloud filter -> Nvidia normap map filter, done. Or go crazy and bake normals from a high poly.

And I just wanted to edit this in. Might be nothing new, and quite hacked for a lot of you, but me likes anyway:

[SteveP_trains]

Many thanks for the information on the way you have got the brass work to show so well. I have heard of that shader but never read into it in any detail so I will do so now.
Sounds as though it can make a big difference to details like the cab controls.

One other question, what software do you use for manipulating your textures?
I use a mixture of PSP and Paint.Net so are these suitable for the techniques you describe?

Much appreciated. Thanks

Steve

[TrabantDeLuxe]

I'm lucky enough to have acces to photoshop (although CS2 was available for free for some time), but any graphics package that has layers and that stuff will do. I believe Paint.net is to photoshop what blender is to 3dsmax, both are - as far as texturing goes, equally capable. Actually, I find you're not doing a whole lot of work in the editing software itself, it's mostly just adding layers on top of each other.

[TrabantDeLuxe]

I've done character modelling. Skinning turns out to be rather difficult, but I can sort of pose this guy halfway decently. To be fair, the head and hands are basically carbon copies of the developer examples. Stats for nerds: 1183 polys, 2048 tris, and 512x512 diffuse.


Hello, I'm your driver. Next stop: Uncanny Valley.

[TrabantDeLuxe]

Right. I've done lots of reading on normal maps, and am definitely happy with the way these have turned out. Split your seams at the hard edges folks, it helps! It is lots of work, especially considering the fact I like to minimise seams for diffuse and spec, so I have given it 2 UV channels. You'll basically need to unwrap your model twice, but I think it is worth it.

Related Q: I am right in thinking that normal maps don't like to be mirrored, and that there is no trick to overcome this issue?



By the way, this got me thinking. Would there be interest in this as a 3d model for other devs? It has it's own textures, and I'm happy to supply the layered texture files and the source model. Perhaps it would be possible to compile a library of standard bits and bobs for those who are just starting. I do need to think about file formats and licensing.

[AndiS]

I better shut up about the normals since exploring them is still on my to-do list.

But I cannot help repeating that the freeware scene needs to free itself from being unpaid payware creators. I mean everyone works along alone in their closet, secretly hoping to become big and/or make money one day, or thinking that if others demand money for this, they cannot give away anything that could be reused for free.

I could well be that I am the only naive left, or the only one who does not hate his job bad enough to quit. Anyway, let me restate it carefully:

Each item shared consists of a theoretical part and a practical part. It starts with sharing historic quotes on steam physics and it does not stop at mirroring normals. There are quite theoretical modes of sharing, like passing on documents and quite practical ones like sharing a Blender file.

I consider all attempts at regulating the flow of information and keeping ownership on information failed, as far as train modelling is concerned. There may be a handful of examples where somewhere knows something and uses it and gains an advantage in the market from that and others don't go and copy it rather soon. But by and large, what one knows will be knows by those interested. On the other hand, there are tons of obvious knowledge out there that are not know to most and those keep asking the same thing on the forums.

If you share freely and openly, you increase the productivity of the receivers. You also attract wannabe receivers who do not have the capacity to benefit from what you give. However, their number is small in comparison with those who want you to supply the finished thing ready to use for them, for free.

Reuse of anything for any product that gets released in our scene is quite limited if you maintain the level of quality and plausibility common now. The company that produced your bogey may well have supplied it to operators outside the Netherlands, but those are likely to at least paint it differently. So for anyone serious about it, your bogey would be a starting point, not something to take and integrate.

If I were to turn your bogey into one that I can sell and I "am earlier to the market" because of that, the benefit of this act still depends on what the market gives me. Sticking to early steam era is your best guarantee against having anyone making big money from what you gave. That does not even touch legal aspects, just economic potential.

The Creative Commons went a few miles to work out the legal details, but we can even cut out that and use the framework Jim provided.

Enforcement of anything is always difficult when digital content is involved. Making people know your stuff, and that you are the owner, is the best way of having any abuse reported. And it is the only affordable measure.

I don't remember a single case of a RW creator being accused of stealing from another one. At the same time, there are tons of sites where you can download illegal copies of complete products. I.e., if someone is inclined to steal, he does it the simple way and if one is inclined to create, he makes sure his name is not tainted.

At the same time, checking how the others do 'it' (with 'it' being every little aspect of the game), will be daily routing for everyone who is seriously engaged here.


Caveat: Looking at the small number of freeware creators left and the buzzing payware industry, I must be missing something important. Don't listen to me.

[SteveP_trains]

Andy
I read your post with interest but have to admit I am not sure which camp you are advocating Let anyone who wants to, reuse bits of models or make sure you get some reimbursement & recognition? Possible both or neither?

From my point of view, since joining this forum, I realise how much skill I don't have or know about, in this modelling lark. But I am keen to learn from anyone who is willing to share and personally I learn quicker by having the source files & textures to work through rather than follow a lengthy text, no matter how informative it manages to be.
I, for one, would not try to simply re-use files but create my own, knowing that it can be done, once i work out what needs to be done

So I guess I am supporting your comments that we should not simply give away any of the hard worked IPR that someone will simply pick up and re-use but be willing to share the knowledge and files with others to help move our hobby forward.

I am hoping that the new Sim being talked about will not drive out the 3rd party hobbyists and be limited to the hard-core, business end of the providers. Hopefully still room for both?

Regards
Steve