you could try masking + polishing it out, or you could try projecting the original high res details back to that area by applying a soft mask selection and projecting
you could try masking + polishing it out, or you could try projecting the original high res details back to that area by applying a soft mask selection and projecting
That is some way but, I noticed that some of my friends do not receive this kind of bugs (or they are smaller) and I would be more interested in finding reason behind this than fight with this using smooths, masking, polishing etc. ^^
Ah, sorry. Couldn't understand that immediately. To be honest though. It is still hard to give an answer. What seems to have happened is that you had a lower poly mesh before this one. Which had an improper polyflow. And then you subdivided it with smoothing, causing it to pinch.
But this is quite speculative since we are only seeing the end-result.
Are you 100% that those bugs are not in the mid poly? Dynamesh should only mimic the surface to my knowledge. Having it create ripples is a bit odd.
Try the following: Change the mesh in the area where the bug is, dynamesh it, watch the end result, make an assumption. Recreate the scenario in another area, check whether your assumption is right. You should at least find something out doing this.
See that all the vertices are merged, the ones that are too close to each other because that will give you crapiness. Plus, you shouldn't divide triangulated mesh or even dynamesh.
Zbrush can Dynamesh triangulated mesh without any issue. If you're using Max to model the blockout mesh, try to convert it to editable mesh then back to editable poly, sometimes it does fix some weird stuff.
I don't think it's so much of a bug as it is just the result of how dynamesh uses the existing mesh's data to create new topology. Clearly it seems to take the existing vertices into account more directly than say Remesh, which just builds cubes around the volume instead.
Simple testing will show different dynamesh results depending on if a model is triangulated or not:
So I think it's good advice to try to avoid working on a purely triangulated mesh, even if the plan is to dynamesh it. If you'd rather use all triangles rather than clean it up, then maybe try jacking the dynamesh resolution as high as possible (this gives it more wiggle room to do whatever it wants with the topology, and the faces will be so small that any errors likely wont even be visible).
Hmm interesting, I did my test here as well between triangulated and quad mesh. Quickly blocked out the mesh with Zmodeler brush and this is my result.
The left side is triangulated and right is quad. Top showed the before dynamesh, middle after dynamesh, bottom after dynamesh without wires. Triangualted mesh does affect the poly distribution and slight inconsistency after dynamesh but no way creating the ripple effect like what OP asked before..
@cryrid why does your dynamesh has a lowpoly wires drawn ontop of the mesh?
@Revel I just photoshopped two screenshots together to show how the verts line up with the dynamesh changes in flow.
As for the initial ripple effect, I'm pretty sure topology is still to blame. I don't see any original wires for those, but they coincide with the hole shapes above them so I'd expect there to be some nasty boolean-like geometry in the pre-dynamesh stage (probably fanning).
Ok, I did a lot of experiments and talked with many people xD
When mesh is triangulated, between vertices occurs bugs. If I export mesh in quads, thera are less bugs, but they still occur since it is impossible to get rid of tris in 100%. My colleague gave me his model from modo and ... everything is fine in zbrush. Model was ofcourse in tris.
And, my main plottwist. I tried using Dynamesh from zbrush (not dynameshmaster) and ... everything is fine ... Really, zbrush's dynamesh is working well with this model 0 bugs, no matter tris or quads.
But, there is another problem (it occured in both methods but only now it became of my concern) if I choose more than 2kk poly, or more than 2048 "resloution" my model goes like this:
Haha yeah sometime too high resolution dynamesh cause this messy polygons, so far I don't think there's an explanation for it though unless I'm missing something..
I would suggest adding some support lines to help because it is a harsh angle. That's my best guess. I only suggest that because it works on normals maybe it'll work on Zbrush.
Replies
In some places zbrush create nice squares sitting next to each other and sometimes he is creating some weird roundish places.
That is some way but, I noticed that some of my friends do not receive this kind of bugs (or they are smaller) and I would be more interested in finding reason behind this than fight with this using smooths, masking, polishing etc. ^^
But this is quite speculative since we are only seeing the end-result.
Try the following:
Change the mesh in the area where the bug is, dynamesh it, watch the end result, make an assumption. Recreate the scenario in another area, check whether your assumption is right. You should at least find something out doing this.
I tried to divide this mid poly several times before dynameshing it, bugs are smaller but somehow more visible xd
Plus, you shouldn't divide triangulated mesh or even dynamesh.
Watch this btw, it's really useful in your example: https://www.youtube.com/watch?v=Fg6QPJ7XirE&t=2337s
I just did a little "overlay" and you are right. Those bugs are between vertexes. Question is why ?
Now I tried to just add more divisions and bugs are still occuring but there is more of them (since there is more divisions xd)
I also tested various scaling, and projection options inside dynamesh master but still, nothing
Anyone, anything ? I am willing to test even most crazy ideas xd
Simple testing will show different dynamesh results depending on if a model is triangulated or not:
So I think it's good advice to try to avoid working on a purely triangulated mesh, even if the plan is to dynamesh it. If you'd rather use all triangles rather than clean it up, then maybe try jacking the dynamesh resolution as high as possible (this gives it more wiggle room to do whatever it wants with the topology, and the faces will be so small that any errors likely wont even be visible).
The left side is triangulated and right is quad. Top showed the before dynamesh, middle after dynamesh, bottom after dynamesh without wires. Triangualted mesh does affect the poly distribution and slight inconsistency after dynamesh but no way creating the ripple effect like what OP asked before..
@cryrid why does your dynamesh has a lowpoly wires drawn ontop of the mesh?
As for the initial ripple effect, I'm pretty sure topology is still to blame. I don't see any original wires for those, but they coincide with the hole shapes above them so I'd expect there to be some nasty boolean-like geometry in the pre-dynamesh stage (probably fanning).
When mesh is triangulated, between vertices occurs bugs.
If I export mesh in quads, thera are less bugs, but they still occur since it is impossible to get rid of tris in 100%.
My colleague gave me his model from modo and ... everything is fine in zbrush. Model was ofcourse in tris.
And, my main plottwist. I tried using Dynamesh from zbrush (not dynameshmaster) and ... everything is fine ... Really, zbrush's dynamesh is working well with this model 0 bugs, no matter tris or quads.
But, there is another problem (it occured in both methods but only now it became of my concern) if I choose more than 2kk poly, or more than 2048 "resloution" my model goes like this:
Soooo, anyone know what is causing it ?