Samson 0-4-0 (low poly game asset)

This is something small I’ve been working on lately: a model of the Liverpool & Manchester Railway’s Samson. This was an early freight locomotive that was built in 1831. Slower than the Planet of the same era, but with a lot more grunt.

Since it’s a game asset, for an old game, there are fairly tight limitations on what I can achieve with it. Poly count is under 1,200 (mesh is still a little bit WIP). There is no point getting fancy with Blender materials and lighting, since in practice it only has to look good under game lighting. The lighting in Blender is just to give a good working environment will I knock it all into something coherent. It also has to be skinned with a dead basic DXT3. There are no normal maps or specular maps, and the maximum skin size is 1024x1024.

It’s about half skinned at the moment. A few bits are still wearing parts of one of the game’s default skins, just as a placeholder, and the new stuff I’ve done still needs more detailing. I think it’s shaping up well, but would appreciate any advice on possible improvements, bearing in mind the restrictions on what’s possible.

Got the mesh sorted. Came in at exactly 1,200 tris and I’m calling it done. Lots of use of alpha to keep the count down, but it’ll skin up ok.

Very nice - where can I get the game from? How easy/difficult is it to make models for the game? Mind you that would give me something else to play with and slow my other progress even more… :rolleyes:

Keep up the good work!

Cheers, Clock. :eyebrowlift:

Great polycount, he looks great. Maybe i would add some dirty decals textures on it :slight_smile:

1/ Lots of places. It’s the old Railroad Tycoon 3. Best to get the original CD version if you want to get into modding things. The Steam version has all sorts of DRM crap added to it and will bork any custom stuff. I think the GoG version isn’t quite as bad, but haven’t had anything to do with it myself.

Just be aware that it’s a train game, primarily an RTS, rather than a train simulator. The equations behind it (a lot of which we’ve reverse-engineered) are quite simple and not particularly realistic, and the model rigging and mechanics are also pretty simplistic. But some nutters are sort of addicted to it, a bit like other nutters are addicted to Morris dancing. The two pastimes are probably similarly daft. :smiley:

2/ Not particularly difficult, even for a beginner like myself. They’re quite simple by modern standards. The only difficult part is trying to make it look good within the limitations of a format that’s better suited to making things look like crap, and keeping the poly count down in the process.

I’ve been thinking that what I should do is find a way of reverse-engineering the (not that great) game lighting capabilities in a Blender lighting setup. It occurred to me that the way to approach it would probably be to skin some basic models with known single colour swatches, then screenshot them once they’re loaded into the game.

By checking the screenshots with a colour picker, and comparing the HSV values against those of the original swatches, and of course having the same swatches set up in Blender, with a bit of head scratching and a fair whack of luck and a whole lot of swearing I may be able to get the model lit up in Blender the same as it would be in the game. This would be handy for skinning, as it could eliminate the necessity to constantly re-load things live to check out colour balance, etc when trying to finalise a skin.

I did a Pennsy H3 recently. Out of necessity it’s a bit simplified here and there, and I can already see ways of improving the skinning, but it turned out pretty decent for what it is.