Jump to content
Sign in to follow this  
Isleofgough

Retopologizing a sculpted hand

Recommended Posts

I would like to get some advice for retopologizing a sculpted hand model. Specifically, I want to be able to texture/UVmap the model, add displacement  and rig with a fairly detailed rig that allows even a few degrees of metacarpal flexion for an animation.

 

There are several factors generally recommended for topology: all quads, minimally non planar polygons, even size and distribution of roughly square polygons, avoiding more than five edges per vertex/point and even for five edges, to put in areas that don't move much.. But especially important are following anatomic relationships and the ability to move the displacement of a Zbrush sculpt onto a lower poly model (which can cause some problems with using inner and regular extrudes/bevels for fingernails and knuckles.

 

Retopologizing with the TopoPen or equivalent causes some shrinking after SDS is applied and can give more non planar polygons. Modeling with isoline editing gives a good look, but might be a problem with weight maps, as the vertices in the non SDS model can be pretty jagged. (see Modo example)

 

Attached are some models I've looked at for inspiration: several of the highest rated from Turbosquid, the asset that comes with Modo, and various tutorials on the internet for box modeling or edge extrusions or sculpting. I did a basic box model in Houdini and then used its equivalent of a shrink wrap deformer (ray node) to get close.

 

For a project like this, would you model or sculpt the fingernails? Do you recommend downsampling or extrusions or avoiding points with five edges on moving parts? Is having higher detail in the fingers that are animated and lower on the back and palm of the hand an issue, as the polygon size will vary a lot. Here is what I have come up with and the references I've looked at.  The Turbosquid models are on the right. Thanks.

HandTopology.jpg

Share this post


Link to post
39 minutes ago, Isleofgough said:

Retopologizing with the TopoPen or equivalent causes some shrinking after SDS is applied

Well that only happens if there are not enough polys in our base form, which is why we usually subdivide a base model to at least 1 or 2 levels and then apply our sculpting displacement maps to that.... but even with a very low poly base mesh, if we know about SDS shrinkage in advance we can just move polys out along their normals a bit during the retopo process to account for it...

 

39 minutes ago, Isleofgough said:

For a project like this, would you model or sculpt the fingernails?

Well I'd model them into the hand, primarily because they are connected to it in the real world. And because I want to UV and weight them together with the hand because it's quicker / easier / less hassle than doing them separately, for no real disadvantage.

 

39 minutes ago, Isleofgough said:

...and can give more non planar polygons.

 

That only matters at the lowest res base mesh level; when something is subdivided nearly all polygons are non planar, but to a very minor degree.

 

39 minutes ago, Isleofgough said:

But especially important are following anatomic relationships and the ability to move the displacement of a Zbrush sculpt onto a lower poly model (which can cause some problems with using inner and regular extrudes/bevels for fingernails and knuckles.

...and that is why anatomically supporting edge flow is important even in the base model, so that when things subdivide that polyflow is enhanced, not interrupted.

 

39 minutes ago, Isleofgough said:

Is having higher detail in the fingers that are animated and lower on the back and palm of the hand an issue, as the polygon size will vary a lot.

Yes. If you look at some of your reference models (most notably this one)...

 

image.png.483130a1d47eb13d8cabbdbcdd395c88.png

 

...you should see that actually polygon size does not vary that much between the sections. Polys in the palm are roughly the same size as polys in the fingers by the time we are ready to add sculpt displacement maps... and that is what we want to work best with those maps. There might not be much movement detail in those areas, but it is not that we are addressing here - we want displacement maps to look right, and they need to displace by the same detail level everywhere on the model.

 

39 minutes ago, Isleofgough said:

Modeling with isoline editing gives a good look, but might be a problem with weight maps

We shouldn't be using any SDS weighting at all in base models that we know will be later used with disp maps, as it produces unpredictable / artefacted results with displacement. Or do you mean with weighting the mesh to the joints ?

 

 

CBR

 

Share this post


Link to post
  • Topic Author
  • Author of the topic Posted

     

    Retopologizing with the TopoPen or equivalent causes some shrinking after SDS is applied

    Quote

    Well that only happens if there are not enough polys in our base form, which is why we usually subdivide a base model to at least 1 or 2 levels and then apply our sculpting displacement maps to that.... but even with a very low poly base mesh, if we know about SDS shrinkage in advance we can just move polys out along their normals a bit during the retopo process to account for it...

    The shrinkage with the TopoPen is straight math from the Pythagorean theorem. The larger the polygons created, the more the subsequent SDS will shrink away from the original high poly model. When SDS is applied, basically the mid position of each polygon remains the same but the edges and vertices will be rounded away from  the points. The TopoPen snaps vertices to the surface. To avoid shrinkage, the face itself would have to snap to the surface, which it doesn't (nor does Modo or Blender or Houdini's equivalent). The shrinkage can be seen by just adding SDS to a cube. (See attachment below for shrinkage regardless of how high poly the original model is). The original corners move inward to be more spherical. Placing smaller polygons might help, but it is really easy to inadvertently create spiral topology or differences in the number of polys between fingers. 

     

    Regarding the fingernails: sculpt or model

    Quote

    Well I'd model them into the hand, primarily because they are connected to it in the real world. And because I want to UV and weight them together with the hand because it's quicker / easier / less hassle than doing them separately, for no real disadvantage.

    I wasn't suggesting having the fingernails be a separate object from the hand. See attachment below for potential issues with modeling the fingernail and then trying to superimpose a displacement map on top of that. It is important that the polygon edges exactly line up with the fingernail edges, as I would add different reflectivity and roughness to the fingernails compared to the skin of the hand.

     

    RE non planar polygons

    Quote

     

    That only matters at the lowest res base mesh level; when something is subdivided nearly all polygons are non planar, but to a very minor degree.

    That is an argument made in the Modo forums for isoline editing. See attachment below for a Luxology created hand model with and without open SDS applied. You can see the points very jagged and non planar polygons. The argument made is that since SDS will be applied, it doesn't really matter if the points are jagged in the non SDS model. But that might be an issue if one wants to do any auto weighting of the points to a rig.

     

    Quote

    ...and that is why anatomically supporting edge flow is important even in the base model, so that when things subdivide that polyflow is enhanced, not interrupted.

     

    Yes. If you look at some of your reference models (most notably this one)...

     

    image.png.483130a1d47eb13d8cabbdbcdd395c88.png

     

    ...you should see that actually polygon size does not vary that much between the sections. Polys in the palm are roughly the same size as polys in the fingers by the time we are ready to add sculpt displacement maps... and that is what we want to work best with those maps. There might not be much movement detail in those areas, but it is not that we are addressing here - we want displacement maps to look right, and they need to displace by the same detail level everywhere on the model.

    That model is interesting, as it has several points connected to five edges even in areas where motion of the skin will occur. (What Mr. Vaughan calls "spiders").  It also would be a bear to map skin to bones, as it does not follow anatomic topology. The polygons do vary considerably between the palm of the hand, where they are basically square, and the forearm, where they are enlongated. Does the displacement map cause a problem if one doesn't stretch these with the UV map? If one created a grid with very dense mesh one one side and few polys on the other and then added a planar map, wouldn't a checkerboard displacement still work well?

     

    The issue of SDS weighting is basically editing with isolines. See discussion above.

    Quote

    We shouldn't be using any SDS weighting at all in base models that we know will be later used with disp maps, as it produces unpredictable / artefacted results with displacement. Or do you mean with weighting the mesh to the joints ?

    Thanks for the answers. I'm still a bit confused on the best way to proceed. I did add a small amount of polyextrude to the high definition mesh to prevent shrinking with subsequent SDS of the retopologized model, and that seems to have worked. Obviously, smaller polygons hug the high definition mesh closer than larger polygons, so I see the value of uniformity of polygon size in this issue. However, I'm not sure the additional five edged points and non anatomic loops wouldn't be a greater problem.

    Quote

     

     

    FingerModelvsSculpt.jpg

     

     

     

    ModoWithWithoutSDS.jpg

     

    RetopologyShrink.jpg

    Share this post


    Link to post
  • Topic Author
  • Author of the topic Posted

    I should add, here is a typical example of retopology edge loop issues with edge extrudes (not my model). Only one edge loop is selected.

     

    Also attached is a test of very unequal sized polygons with a checkerboard texture applied to the diffuse and displacement channel. I'm not seeing distortion. It would seem that the only way distortion would occur is if one changed the UV map after the high polygon model was converted to a displacement map in Zbrush and that even size of the polygons doesn't have much bearing in displacement.

     

    RetopologyEdgeLoopProblem.jpg

     

    PlanarCuts.jpg

    Share this post


    Link to post
  • Topic Author
  • Author of the topic Posted

    Thanks again. I discovered several things in this project: Box modeling in general gives much better anatomic edge loops than edge extrusions/retopo. Add and remove edge loops as needed rather than apply SDS. Don't do any complicated parts early, as they are a pain to edit later (fingernails). My criteria for a good model was: 1. all quads. 2. only 3 and 5 edge spiders and even these should be placed where minimal bending occurs. 3. For rigging and UV purposes, one should have an edge loop that splits the model in palmer and dorsal sides. 4. For rigging fine tuning (vertex weighting), your model should be able to be decimated so that parts can be parented to bones and the motion should look good. (this follows Tina O-Hailey's recommendation for rigging in Maya and the Houdini hand rigging tutorial series). See attachment for rigging by parenting and final result.

     

    RigRegions.jpg

    HandRetopo.jpg

     

    ModoUVmapHand.jpg

    Share this post


    Link to post

    Join the conversation

    You can post now and register later. If you have an account, sign in now to post with your account.

    Guest
    Reply to this topic...

    ×   Pasted as rich text.   Paste as plain text instead

      Only 75 emoji are allowed.

    ×   Your link has been automatically embedded.   Display as a link instead

    ×   Your previous content has been restored.   Clear editor

    ×   You cannot paste images directly. Upload or insert images from URL.

    Sign in to follow this  

    • Recently Browsing   0 members

      No registered users viewing this page.

    ×
    ×
    • Create New...

    FORUMS INFO:

    Dear members, we are aware of few more bugs that are still present withing the theme.We just wanted to let you know that we are working to fix them as soon as possible.
     

    NEW MEMBERS:

    Please be aware that we are manually approving all new registrations, due to spam prevention. Please be patient in case you cannot login right away, we will approve you within 12h or less if we decide you are not potential spammer. 

     

    Thanks for understanding! :cowboypistol: