Home Technical Talk

How many polygons should a rigged model for animation have?

node
Offline / Send Message
MeowMoo node
Hi guys, I am doing a 1:1 scale model for rigging in Zbrush and getting confuse of my workflow. I was going to retopo my model before I went in too deep but I don't know how many polygons should it have after retopo.

My question is,
1. how many polygons (approximately) should this model have after retopology?
2. What's the general poly-count do you have when making an animation-ready character?

I would be more than appreciated if someone could answer these stupid questions! 

Replies

  • oglu
    Offline / Send Message
    oglu polycount lvl 666
    As much as your system can handle, as much as needed.
    But most rigs does have multiple geo representations.
    For example the Dinos in KingKong des have 1mio for the anim/render mesh.
    But they had also fast rig versions for the animators.

    For a cinematic char i would aim for a 200k subD mesh + a fast version.

    Im sure you have seen some of those fast anim meshes.
    Its a lower version with deleted edgerings.

  • Obscura
    Offline / Send Message
    Obscura grand marshal polycounter
  • Alex_J
    Online / Send Message
    Alex_J grand marshal polycounter
    Obscura said:
    3
    technical, I think you only need two things, to know that one is moving.
  • Eric Chadwick
  • oglu
    Offline / Send Message
    oglu polycount lvl 666
    If im correct he is not asking about a game char polycount.?

  • MeowMoo
    Offline / Send Message
    MeowMoo node
    Thanks for answering my noob questions! 
    Just making sure I understand it right. So the the # of poly-count doesn't matter as long as my computer doesn't lag when animating? And the rigger will weight the skin twice for creating 2 version of the character (the high res for render and low res for animation)? 
  • oglu
    Offline / Send Message
    oglu polycount lvl 666
    Is the char for rendering or a game?

    If its for rendering yes do whats needed until your computer cant handle it.
    The second mesh is just parented not weighted.

    If its for a game you need to follow the pipeline rules.
  • MeowMoo
    Offline / Send Message
    MeowMoo node
    oglu said:
    As much as your system can handle, as much as needed.
    But most rigs does have multiple geo representations.
    For example the Dinos in KingKong des have 1mio for the anim/render mesh.
    But they had also fast rig versions for the animators.

    For a cinematic char i would aim for a 200k subD mesh + a fast version.

    Im sure you have seen some of those fast anim meshes.
    Its a lower version with deleted edgerings.

    Thanks for the information!
    Just making sure I understand it right. So the the # of poly-count doesn't matter as long as my computer doesn't lag when animating? And the rigger will weight the skin twice for creating 2 version of the character (the high res for render and low res for animation)? 
  • MeowMoo
    Offline / Send Message
    MeowMoo node
    oglu said:
    As much as your system can handle, as much as needed.
    But most rigs does have multiple geo representations.
    For example the Dinos in KingKong des have 1mio for the anim/render mesh.
    But they had also fast rig versions for the animators.

    For a cinematic char i would aim for a 200k subD mesh + a fast version.

    Im sure you have seen some of those fast anim meshes.
    Its a lower version with deleted edgerings.

    Thanks for the information!
    Just making sure I understand it right. So the the # of poly-count doesn't matter as long as my computer doesn't lag when animating? And the rigger will weight the skin twice for creating 2 version of the character (the high res for render and low res for animation)? 
  • Mark Dygert
    Poly count does matter, if you're going into a realtime game and even if you're rendering, you want to make sure your models have just enough polys to hold the detail but not an excessive amount that they cause other systems (rendering nodes and player machines) to chug. 

    There are other external factors at play most of the time, your character will be going into a scene or a game and while you can render things in isolation or crunch things down to get them to play nicely those steps are really just tinkering with the underlying problem of not carefully evaluating the project, not setting realistic tech budgets and not vetting the process through proper testing.

    You have to collaborate and balance things out not just on your machine but within the scope of the entire project. If you play the "well it ran fine on my machine" game, the project is going to suffer, especially if everyone pushes their one piece up to the limit of what their system can do. What happens when you bring all of those pieces together? Players pitch fits, testing burns to the ground and rendering budgets cost way too much.

    When people are first learning they like extreme rules "Always do this. Never do that." and sadly some teachers teach that way. But its really a balancing act of moderation, with a lot of different ways to tackle a wide range of issues. The best solution is often found somewhere in the grey middle area and might only be the best solution for that project under those circumstances. 

    (climbs off soapbox)

    The rigger typically won't weight things twice, at least not from scratch each time. Most of the time we weight one mesh, then we copy weights from that mesh to other meshes.

    In the case of proxy meshes, we can use copying settings that can help if you've already done a smooth blending version.  But I've seen a few people weight the proxies first then copy the weights from the proxy and then smooth those weights out. There are more than a few scripted workflows that follow that train of thought.

    Personally I don't like doing it that way, especially when geodesic voxel skinning gets so much of the default skinning, right. Doing a 1:1 skin on proxies isn't really a big deal since most of your time when skinning is painting and blending weights. Also a 1:1 skin can be automated fairly easily through a bit of Mel and Python or just through some of the default settings. Again there are a lot of different ways to do that and it really depends on how you set up your pipeline and how much you automate.

    A good rigger will be fairly well versed in scripting and will make the process as quick and flexible as possible.
    A bad rigger will relay solely on some automated tool that only really works for a 1-2 workflows and they're screwed if they have to do it another way. they may get the job done but it's going to be pretty crappy and probably eat up a bunch of time, or put constrains on the project that don't really need to be t 

    I use proxies not really for performance but for clarity. scenes get cluttered when there are a ton of controllers hovering around a mesh. Its easier to click on the piece rather than play find the floating thing that is probably clipping into something else.

    Also there are rigging tools that greatly help with proxy segmentation and doing the skinning on them. While I think rigging is a valuable skill to have, especially if you're going to be an animator, most of the time tools like Advance Skeleton, The Setup Machine or even Biped and CAT will be really helpful in getting the process moving, even if it's for the R&D phase if not the entire project. If you don't know how to rig manually I honestly think you shouldn't take on that responsibility for a project and have to rely solely on automation, its going to fall apart or make life miserable when you really need to rig in a particular way and can't.

    So it really comes down to use.
    How far down the rabbit hole do you want/need to go?
    What are the needs of the project and what will need to be deployed?

    If you start to answer some of those questions then you start to get a grasp on how to go about building it all out. But there isn't one workflow that everyone uses for everything. Really, you just have to dig in with the tools you have and make the best of it and realize that you're going to find better ways to work later on down the road and that's fine, a career is an iterative process and some of the best people I know aren't afraid to say "I'll have to think about that" or "we should research and test it". You're never really done learning and just because can do a specific task at a specific place it doesn't mean you'll be amazing at everything, everywhere. Which is why people ask some very specific questions in interviews. 

    TLDR: Use what you have available to you. Try to do as much research and vetting as possible, as quickly as possible.

    Would someone please take this thing away, I keep climbing on it.
    (kicks soapbox out of the way)
Sign In or Register to comment.