Home Technical Talk

3ds Max/Zbrush: Proboolean + Dynamesh hardsurface workflow tutorial

123457

Replies

  • guitarguy00
    Offline / Send Message
    guitarguy00 polycounter lvl 2
    rregula said:

    I am still getting the same problem. It is happening in Zbrush. This is the High Poly exported out of Zbrush and into Marmoset.

  • PolyHertz
    Offline / Send Message
    PolyHertz sublime tool
    It could also be caused by bit depth or texture conversion/compression issues. If you can post the files it would help quite a bit in figuring out the exact cause.
  • Kanni3d
    Offline / Send Message
    Kanni3d greentooth
    PolyHertz said:
    It could also be caused by bit depth or texture conversion/compression issues. If you can post the files it would help quite a bit in figuring out the exact cause.
    It ended up being exporting as .fbx and not enough sub-divisons. :astonished:

  • zachagreg
    Offline / Send Message
    zachagreg ngon master
    Noors said:
    Just passing by to say that claypolish with softness doesnt result in inflated "chamfers" like polish does (which was bothering me).



    The one issue with that is that for some reason clay polish fucks your geometry causing weird warping and then needs Zremeshing/Dynameshing after wards. If you don't intend to sculpt afterwards then this method works in a much cleaner way than traditional polish.
  • Kanni3d
    Offline / Send Message
    Kanni3d greentooth
    zachagreg said:
    Noors said:
    Just passing by to say that claypolish with softness doesnt result in inflated "chamfers" like polish does (which was bothering me).



    The one issue with that is that for some reason clay polish fucks your geometry causing weird warping and then needs Zremeshing/Dynameshing after wards. If you don't intend to sculpt afterwards then this method works in a much cleaner way than traditional polish.
    I find polish by crisp edges with hardness does great, no inflated chamfers either.
  • Noors
    Offline / Send Message
    Noors polycounter lvl 11
    Yep you're both right. Polish by crisp edges (with "empty" circle option) gives a similar result as claypolish smooth without affecting the topology.
    + a little relax to ease artefacts. Looks very fine to me.
    This dynamesh technic is really robust. Works every time where i'm not sure CAD fillet/meshfusion would.
  • garriola83
    Offline / Send Message
    garriola83 greentooth
    Still one of the best threads in PC ever. Thanks for everyone's contribution!
  • Kanni3d
    Offline / Send Message
    Kanni3d greentooth
    I didn't want to start a new thread, and figured I could get some insight in this thread. :smile: Apologies if this is a bit image heavy.

    I've been facing this irritating issue with dynamesh, albeit probably not a big deal in the grand scheme of things, since once it's baked down with textures applied + compressed in engine, it may not be all too noticeable. This issue however isn't unique just to me, it's simply a by-product of how dynamesh works, and others have possibly noticed this as well. I love and use this workflow extensively, and have always noticed that in certain rounded shapes/angles/transitions, you'll get this stepping artifacting shown below.  Even at really high dynamesh resolutions, resulting in about 2-3+ million points.



    The nature of this is just simply how the voxels are being projected onto the mesh, zooming in and looking at the topology, its the triangles and poles along this surface being the culprit, compounded by the fact we're trying to smooth it via polish tools.



    Above in the thread, I've seen some mentions of using relax etc. to relieve this artifacting, but it only does so minimally, and applies a regular polish which I personally do not like. Regular polish, regardless of the mode, adds an inflated effect to the edges, which probably isn't too noticeable in some cases, but I'd rather avoid it and just use crisp edges with a hollow circle.



    Something I thought of that would fix this issue altogether, is to just simply have all quads. So I thought of, "why not z-remesh my dynameshed model, project it, and polish the artifacts away?".
    Totally possible - so first, I grab my post boolean mesh, polygroup by angle (depending on the model, around 20-25 threshold, and this is just so that when I zremesh, it'll have at least some shape preservation with certain options ticked) and zremesh. Then at this point, I divide, project, and repeat until my new zremeshed model is about the same amount of resolution as my dynameshed model with artifacting.

    At this point, I can see the original artifacting has projected into the clean quads mesh, but that's fine because once i smooth, it's gone!

    BUT
    , there is a new problem I'm facing, which is that with more complex shapes, zremesh will tend to add poles/triangles randomly across the mesh, but worse, at the edges of hard edges/polygroups. This gives me new, different artifacts elsewhere on my model once smoothed... In the image below, it's not as bad as the previous attempts, but this is definitely the cause of the issue itself. I've tried exploring many different options and combinations of dynamesh, zremesh, different amounts of zremesh resolution, with/without polygroups, creasing, using QuadRemesher in Max and re-importing (which is supposedly the same, but improved algorithm) and no luck. :\



    I've even tried using zmodeler tools to inset this entire face with a clean loop all around, which helps, but also not too ideal because there could be some tight areas, which would have the inset collapse on itself, or be extremely dense.

    I really thought I had solved this pesky artifacting issue on my own with this method, but now after tons of researching and asking around, I've hit another brick wall and hope anyone else may have some sort of idea or insight to help me out here. Perhaps there is something I'm missing, some further or prior steps I can take to fix this, or an entire other tool in zBrush that i'm completely unaware of that may help as well?

    I apologize for going on a bit of a tangent on this small issue, but I hope I've illustrated this problem well enough, in case I've not, I've included a google drive link to a screen recording demoing an example mesh of this process to make things a bit clearer hopefully.

    https://drive.google.com/file/d/1ds36_BIONdQ8s50w2ofjv9Nd5_3wHGyV/view?usp=sharing
123457
Sign In or Register to comment.