Home Technical Talk

Hard surface workflow for games?

polycounter lvl 6
Offline / Send Message
DirtyBlueFL polycounter lvl 6
So I've been modeling for games in my undergrad for about 10 months now, and I've learned more from the internet than I have in school - go figure. But one thing I definitely want help on is my workflow for hardsurface modeling in Maya. Right now, I build the low poly model first, then I use that as a base to build the high poly model. I unwrap the low poly, create a cage from it, explode if needed, and bake.

I see a lot of people just dive right into the high poly model from the very beginning, but what I don't see is how they retopo to build a lowpoly for the final in-game asset. I am willing to bet that my workflow is actually the traditional or "correct" workflow just turned upside down. I'd really like to know how to do it the correct way, and end up with a contiguous (or close to it) mesh. A lot of the art coming out of the end of the Titanfall 2 NDA is contiguous hard surface, and it's where I want to aim my overton window. 

I understand that context and a point of reference is helpful, so to give you an idea of where I am at skill-wise, this is my artstation: 
https://www.artstation.com/artist/nicholashunter   

Replies

  • DirtyBlueFL
    Offline / Send Message
    DirtyBlueFL polycounter lvl 6
    For example:
    https://www.youtube.com/watch?v=da0Dgj7DBpc

    How the heck does one go about retopoing this? 
  • musashidan
    Offline / Send Message
    musashidan high dynamic range
    There's no right or wrong way, there's just the end result. And there are many ways to get there:

    Your current method - which is perfectly fine - model low>sub-D or chamfer(see below) for the high>use sub-D cage + topo optimise for the final asset.

    Boolean + Dynamesh - manual retop or repurpose the boolean mesh for the final asset.

    Chamfer using any number of tools found recently here in threads - repurpose the base mesh as final asset.

    Build the high in Zbrush or Fusion360 - manual retop the final asset.

    Build a mid-poly mesh as your final asset(becoming more and more popular due to generous on-card vert counts these days) - Use tools like NDO or substance to 'paint' normal details.(or mesh decals)

    Build a 'low' mid-poly and use FWVN + NDO/Substance or mesh decals

    I advise trying out all these methods and then you will have the knowledge/experience to make this decision, on future projects, on an asset by asset basis.
  • Avyie_G
    Offline / Send Message
    Avyie_G vertex
    musashidan is correct. There is no right or wrong way of modeling something for games, as long as everything's in quads and tris!

    Some people skip modeling the low poly in Maya because they know they're going to be doing all the details in zbrush. Knowing that, they can model the base mesh in zbrush and go straight from there. You might find yourself going this way if you become adept at making your base meshes in zbrush.
  • DirtyBlueFL
    Offline / Send Message
    DirtyBlueFL polycounter lvl 6
    Thank you, that's a relief to know. Fusion 360, Modo, or Max are next on my list of software to learn. The first 2 seem like they're more suited to rapid prototyping and high poly concept, but the ease of making complex topology make me look at maya and just feel so hamstrung. 
  • poopipe
    Online / Send Message
    poopipe grand marshal polycounter
    I always advise blocking out and testing properly before committing to anything even halfway resolved. After that you're best off building your highres so you know what the end result will look like - the low res bit is easy once that's all sorted. 


  • jozin
    Offline / Send Message
    jozin polycounter lvl 6
    So guys, its ok - there are many ways, but what is the fastest, IMO:
    the fastest way to work with hard surface is dynamesh/clip brushes/zmodeler -> combined fashion,  but there is always a huge slowdown when retopo this mess of polys;
    2) subd in zbrush using zmodeler - u could model everything using separate pieces (eq using overlaps) and combine lowpoly meshes where needed, then u can provide some bevels on the edges and ur low poly is finished, now to model high poly onlyest thing u really need - press the dynamic subdivision button

    So this way is more robust but u need to be aware of  vast amount of things and plan it in advance.

    What do u think about that? It's doable? or nope? I dont know, because of this me asking u,  in other case I would barely wrote something because I had known about the things, but this time, Im just not sure about, so please help, state ur opinion please if you want to.

    Thanks in advance, Greetings!!

    PS: and one interesting question, there is Pavlovich tutorial on handgun https://www.youtube.com/watch?v=MBqkPBh4Twg
    in the tutorial he just dynameshes all the pieces then decimates it all and uses this for low poly and texturing, is this reliable for modern games, must we ever retopo hard surface  pieses (rigid) to more nice looking topology, or we could just live it intact? Why his example with just decimation master is a bad example? what do u think?
Sign In or Register to comment.