Friday, April 19, 2013

Gameloft Test


Hi, As I said in a previous post, I did two tests recently for two game companies. None of them were successful unfortunately. I'll talk here about the first one, done for Gameloft in 2 weeks.

It had to be a futuristic abandonned military base. And googling for these keywords, I found several artists doing the same test a few years ago with, of course slightly tighter constraints. In my case it had to be rendered within the 3ds max 2011 viewport, with special shaders they provided. The triangle count should stay under 35k for the whole scene amongst a lot of other constraints (texture packs, lightmap, etc.). I think it was by far the most technical test I had as for now.




They provided a concept art I could follow but gave the option of concepting myself the environment. I previous tests I didn't dare taking those opportunities but it seems using the "freedom" they offer you in such situation is a good point. At least it worked for some friends of mine. That's why I chose to go concepting this scene. I did very rough drawings on sheets of paper I've probably lost since then.


One of the parts I'm the most proud of is the multi-plan Background. It's the simplest part, but I had a few ideas for optimisation and they worked perfectly. Unfortunately it doesn't show much on screenshot. Indeed the effect benefits from motion into the pseudo level. Also, with tis lighting, my scene became entirely in the shadow, which became a problem when adding the lightmap. And I think it's one of the major issues with the work I done for this test. It would only look ok if you travel through the level, which is anyway annoying with 3ds max.

I did High Poly into low poly modelling and 3ds max shown very nice artifacts on the HP versions of the kind of containers. Here again, the result was not looking as good as I hoped due to the shaders, and absence of dynamic lighting that might have show normal maps up.

Overall I'm glad I was able to do it in only 2 weeks but not happy with the result. In terms of process I think working with shaders into max that can show you the final result of your scene is very appealing but in the end I prefer working a scene into a third party game engine. This way you avoid a lot of pitfalls. Even with a very low polygon count scene, the framerate ended never being over 10fps, which is a pain working with. Also, 3ds max is not meant to travel into a scene whereas a game engine does that perfectly well. And in the end, you don't have to bake the lightmap manually in most of game engines, which was very risky and time consuming with vray.

Now I wonder why they use this process and how they avoid the pitfalls I mentioned earlier. Anyway, there's a lot of points and decisions that I should have made differently.

Next time I'll talk about the second test I did for Cyanide. This time with UDK.