Home Technical Talk

How The F*#% Do I Model This? - Reply for help with specific shapes - (Post attempt before asking)

Replies

  • epistularum

    I'm trying to turbosmooth the object while retaining hardness on the inner beveled part and the outer parts like screenshot below where hard edges are in red. I always end up with pinching.

    Attempts:
    2) https://i.imgur.com/Bd8e02r.png https://i.imgur.com/ohsuNsP.png result:https://i.imgur.com/SwmYiQT.png



  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @epistularum There's no matching segment below the bevel to match the new support loops so it causes a pinching shading error. If the side is flat then just terminate the edge loops there. If there's a taper then it's going to need a little finessing.

    Here's something that could work for either. It's possible to reduce this into an Ngon or tri configuration. It all depends on the desired shape and the overall size of the object.





  • epistularum
    This still results into pinching, am I missing something in your explanation?

  • sacboi
    Offline / Send Message
    sacboi high dynamic range

    That surface is non-planar (not flat) hence the visual artefact (pinching) result.

    Try FrankPolygon's third *ngon* example and see if it makes any difference.

      

  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    If the ngon approach isn't ideal, try sliding these edges to the middle and that should soften the crease. Depending on the desired shape and corner tightness it may take a little finessing.


  • epistularum
    This is the n-gon approach:

    I tried your method but I still got pinching. But it got me thinking, I moved the edges to the extreme edges and welded them to the top/bottom while keeping a quad topology. This works but the topo seems horrendous, surely this is an issue? What do you guys think?

    Also another option that I found is simply trubosmoothing a bit once to have more topo to keep the curves and then extruding.
  • Klunk
    Offline / Send Message
    Klunk ngon master
    I'm trying to place the watch along this path. I don't want to use path deform because i dont want to deform the individual elements. I tried to use a path constraint but all the pieces lose their initial location and are all placed at the same place.
    I would love to know if somebody has a solution for this.

    yeah, you can use a spline based "chain" rig for stuff like this, works for chains, tank tracks or any thing where you need something to follow the next.

    the trick to the rig is to have helper objects act as the pivot points which are path constrained to the spline.... then each item in the chain is then position constrained to the pivot helper with a look at constraint to the next helper along. You can then add a spline IK modifier to the spline so it can be animated. As long as the spline length remains relatively constant throughout the animation it should look pretty convincing.




  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter

    The pinching in the first one can be explained by the support loops being too close in the middle. The second and third have similar topology in the middle but the second one pinches and the third one doesn't. This points to something else like the middle edge loop being marked as a crease.


    The second and third examples have similar topology around the middle loop and the pinching is too consistent between the first and second for it to be "just" a geometry issue. You mentioned sliding and welding edge loops on the third example and those operations may have resolved any obscure geometry issues but it seems more likely that it removed any crease settings that were causing pinching along the middle edge loop.

    I was able to replicate this type of pinching on the ngon topology by marking the middle loop as a crease.



    I'd be interested in taking a look at the actual geometry if you could upload an obj or fbx file.
  • sacboi
    Offline / Send Message
    sacboi high dynamic range

    epistularum said:

    I moved the edges to the extreme edges and welded them to the top/bottom while keeping a quad topology. This works but the topo seems horrendous, surely this is an issue? What do you guys think? 

    Good solution, also what matters is the smoothing result once the subD algorithm is applied not the underlying topology/Geo, whether aesthetically pleasing or not.

    EDIT: I'd recommend researching the subject further either on this thread or interweb, generally I think you'd be surprised with what you'll find, which basically will dispell misconceptions via those new too the discipline.  

  • MiguelGutierrez
    Offline / Send Message
    MiguelGutierrez polycounter lvl 5


    Anyone knows how to make the nose for the F- 22 raptor? I'm trying adding support edges even with crease and opensubdiv but I can get it. Any help will really appreciated!!

    Thanks!!
  • wirrexx
    Offline / Send Message
    wirrexx ngon master


    Anyone knows how to make the nose for the F- 22 raptor? I'm trying adding support edges even with crease and opensubdiv but I can get it. Any help will really appreciated!!

    Thanks!!

    12 sided cylinder with even out edges. The top section is a quad too (pointy)

  • MiguelGutierrez
    Offline / Send Message
    MiguelGutierrez polycounter lvl 5
    Awesome!! thanks I really appreciate your help. I can finally end this model... thank you so much!!
  • MiguelGutierrez
    Offline / Send Message
    MiguelGutierrez polycounter lvl 5
    I followed your instructions... is this the shape of the pointy nose in your model??



  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    yes it is @MiguelGutierrez collapse the top vertices
  • javcop
    Offline / Send Message
    javcop polycounter lvl 7
    Hey guys, any ideas on how to get this pinching to smooth out? Have them separating out to see if that would've helped, but it's no good. Tried some other stuff, but still struggling with this. Any help is super appreciated

    FBX: https://mega.nz/#!wZ5VyCxB!4iMbrOXxs17f7OVgr7A8BKS47oNK8htMwemXbLbykD0


  • javcop
    Offline / Send Message
    javcop polycounter lvl 7
    Managed to figure out my issue - aligned and smoothed out all my verts by using a plane as a live object. Could've used the same strat on the other corner, but achieved the same result and got rid of some extra geo by merging those edges


  • Mir76
    Offline / Send Message
    Mir76 polycounter lvl 3
    Hello.
    I wanted to do a test. I created this with Blender and with a little bevel on it.


    Is it the best way to do this in terms of modeling ?
    Thank you so much !


  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    "Good" is both relative and subjective. Depends on how big the object is, how it will be viewed, if it will be baked to a low poly mesh, etc.

    If the goal is SUB-D modeling then something like this is probably a little more traditional. How this all comes together depends on what you need the edges of the cube to do: square or round, soft or sharp. The big thing is you want to try and match the support loops on the cube with the segments on the cylinder.

    There's a lot of pages in this thread but if you dig you should find other examples of similar geometry. Hope this helps.





  • Mir76
    Offline / Send Message
    Mir76 polycounter lvl 3
    Thank you so much.
    I was trying to reproduce this same workflow :


    But if it's a plane with a cylinder I don't understand how to do with my bevel.

    I have already seen many pages but there is not this same bevel and this cylinder.






  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter

    You have the right idea: the geometry intersects between the vertical segments of the cylinder. If you're happy with how it subdivides and it bakes down fine then it's passable. It all depends on how clean it needs to be and what you're using it for.

    Where you can run in to problems: the two vertical loops on either side of the cube break off and flow into the corners. This can cause pinching artifacts. Without support loops around the intersection the cylinder can deform and loose shape. This will spread the smoothing errors but it won't hold a tight shape.

    Using less geometry in the round overs and fillets will make the object easier to edit. If the mesh has to look good with and without subdivision then that's a different use case.


    Here's what the process looks like when adding a fillet to the intersection. The supporting geometry and the edge flow around the intersection is the same for flat, chamfered and fillet intersections. The big thing is matching the cylinder's segments to the edge loops on the intersecting geometry. Use the existing vertical edge loops on the cylinder as support loops for the intersection.


    This geometry supports big and small fillets, chamfers and intersections with sharp, soft and rounded corners. It renders and bakes down well with subdivision and looks decent without.
  • Nonplus
    How would I bevel these highlighted edges so they remain tight when subdividing?



    A typical bevel with 2 segments produces this result:



    But this creates tris that mess up the edge flow:



    This is the best I have so far, but this still throws out some undesirable shading and makes the bevelled edges on the left lack a division compared to the rest:




    It's not perfect but it might have to do... unless anyone has a better solution?

  • sacboi
    Offline / Send Message
    sacboi high dynamic range
    Try vertex/normal weights or crease values to harden those edges.
    (I'm on a phone at work so can't provide an example)
  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    Nonplus said:
    How would I bevel these highlighted edges so they remain tight when subdividing?



    A typical bevel with 2 segments produces this result:



    But this creates tris that mess up the edge flow:



    This is the best I have so far, but this still throws out some undesirable shading and makes the bevelled edges on the left lack a division compared to the rest:




    It's not perfect but it might have to do... unless anyone has a better solution?


  • Nonplus
    wirrexx said:
    Nonplus said:
    How would I bevel these highlighted edges so they remain tight when subdividing?



    A typical bevel with 2 segments produces this result:



    But this creates tris that mess up the edge flow:



    This is the best I have so far, but this still throws out some undesirable shading and makes the bevelled edges on the left lack a division compared to the rest:




    It's not perfect but it might have to do... unless anyone has a better solution?


    Thanks for the response! However, doing this gives me the same issue I mentioned above:




    It's worth noting that this shape on the left here curves into the convergence. Perhaps this is what's causing the issue with the smoothed mesh + tris?
  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    Nonplus said:
    wirrexx said:
    Nonplus said:
    How would I bevel these highlighted edges so they remain tight when subdividing?



    A typical bevel with 2 segments produces this result:



    But this creates tris that mess up the edge flow:



    This is the best I have so far, but this still throws out some undesirable shading and makes the bevelled edges on the left lack a division compared to the rest:




    It's not perfect but it might have to do... unless anyone has a better solution?


    Thanks for the response! However, doing this gives me the same issue I mentioned above:




    It's worth noting that this shape on the left here curves into the convergence. Perhaps this is what's causing the issue with the smoothed mesh + tris?
    would you mind sending me the FBX (no obj, cant open them up in blender) I would like to see why they issue occur. Being a flat area. This seems weird.
  • mujuningaiza
    Offline / Send Message
    mujuningaiza polycounter lvl 6
    bcottage said:
    Why am i loosing sleep over this? i have had a few hours on this and cant get this simple shape :(

    Anyone care to help?
    you shoud read probolean in compound object in 3Ds MAX

    0UgPKpng
    bcottage said:
    Why am i loosing sleep over this? i have had a few hours on this and cant get this simple shape :(

    Anyone care to help?
    you shoud read probolean in compound object in 3Ds MAX

    0UgPKpng
    You better model that thing using nurbs, eg lofting curves. Then Convert nurb to polygon after doing so put it in place, delete some faces combine the mesh, add some loops to make nice poly flow, merge vertices and there you've it.

    Another method would be Extruding then merging some vertices, then beveling sharp edges.
  • Mir76
    Offline / Send Message
    Mir76 polycounter lvl 3
    FrankPolygon  -> Thank you so much.  I retain your remarks.

  • SnowInChina
  • Thanez
    Offline / Send Message
    Thanez interpolator
    @Nonplus Hi buddy. Unless you absolutely need the sharp edge to be an edge, you can have it be a strip of very thin quads instead, allowing you to do normal edgeloop workflow and avoiding that triangle that's giving you a hard time.
    Chamfer these badboys to get dat quad at the edge.

    Yay, weld these verts to the right.

    See that strip of quads replacing the triangle? YEHEHEEEEEBOIII.MOV

    Add edgeloops as tight as you want

    Squish these verts together

    PROFIT!





  • LVL45
    Offline / Send Message
    LVL45 polycounter lvl 7
    Hi guys, I need help here. I did boolean operation which leaves many unused vertices.
    Does anyone have suggestions how can I fast delete unwanted vertices and basically leave only those that forms angle and shape?
    Maybe some option or script that select vertices by certain angle?

  • Zoddo
    Offline / Send Message
    Zoddo polycounter lvl 5
    LVL45 said:
    Hi guys, I need help here. I did boolean operation which leaves many unused vertices.
    Does anyone have suggestions how can I fast delete unwanted vertices and basically leave only those that forms angle and shape?

    There's an option in your boolean settings in the advanced settings tab, i think it's "no edge removal". Usually that removes unwanted results.
  • UhhNope
    Looking for some help with this topology on this bicycle frame. I just brought together these three cylinders but now I'm struggling to clean up the topology. I'm not really sure how to eliminate the tris without distorting the straightness of the cylinders in the process.



  • Klunk
    Offline / Send Message
    Klunk ngon master
    Hi guys, I need help here. I did boolean operation which leaves many unused vertices.
    Does anyone have suggestions how can I fast delete unwanted vertices and basically leave only those that forms angle and shape?
    Maybe some option or script that select vertices by certain angle?


    if it's max then add a toPoly (Turn To Poly) modifier (remove mid-Edge vertices should be checked if not the check it) Which seems to be broken :/ doesn't work on open edges :/

  • UhhNope
    UhhNope said:
    Looking for some help with this topology on this bicycle frame. I just brought together these three cylinders but now I'm struggling to clean up the topology. I'm not really sure how to eliminate the tris without distorting the straightness of the cylinders in the process.

    ...

    I tried recreating the vertical cylinder to make it line up with the segments of the V-shaped poles more. I had to dissolve and slide some edges on the vertical cylinder though which didn't preserve the perfect roundness of the cylinder like I wanted.


  • esPhys
    Offline / Send Message
    esPhys polycounter lvl 7

    I wonder if you're trying to use too much geo. I'm not sure how tight you're trying to get the edges, but this works pretty well for me. That n-gon could be snapped down to the line below it without much difference, didn't think about that before saving the pic. 
    I feel like the more geometry you try to have when doing these kinds of shapes, you just end up with harsher pinching instead of spreading it out over larger polygons making it less of an issue.
    You didn't post a picture of the smoothed mesh so I'm not 100% sure of the issue you're having, but something that stands out is that you're vertical cylinder does not have consistent edges. It's possible to have a consistent cylinder with differently spaced geometry, but it's difficult and for something like this I'd recommend keeping the cylinders geometry perfect and planning ahead to match the edge counts for what you need.
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    Start with the basic shapes and match the segment count so the little tubes match each other and the big tube matches the intersection with the little tubes. The horizontal end of the intersection should land between the cylinder's vertical segments and not on them.

    You can use less geometry and rely on the subdivision to do the work: mock-up, boolean, cleanup (manual or merge by distance), bevel. From there it's just variations on a theme and it can be adjusted to match whatever gusset shape is needed.


    This is what it looks like with a middle loop. The intersection is sharper and requires a little more geometry.


    Same shape but different loop strategy that's a little more optimized.


    This is the above shape with some edge loops merged down in places to simplify the control loops. This geometry looks more like a welded intersection and could accommodate a secondary gusset piece.


    Here's the side wires.



    Some variations on the gussets based on adding geometry and stretching existing geometry.


    Could be further optimized down to something like this but after a certain point there's diminishing returns. Easy to edit and adjust though.


  • sacboi
    Offline / Send Message
    sacboi high dynamic range

    UhhNope said:

    UhhNope said:

    Looking for some help with this topology on this bicycle frame. I just brought together these three cylinders but now I'm struggling to clean up the topology. I'm not really sure how to eliminate the tris without distorting the straightness of the cylinders in the process.

    ...

    I tried recreating the vertical cylinder to make it line up with the segments of the V-shaped poles more. I had to dissolve and slide some edges on the vertical cylinder though which didn't preserve the perfect roundness of the cylinder like I wanted.

    You're almost there.

    Just learn too work with less geo especially when blocking out or defining form/s.

    EDIT: Plenty of examples populated throughout this thread, in fact the very first post illustrates a low poly approach.


  • ArcticTauntaun
    Hey guys, I was wondering what the best way to model this complex sphere intersection for a Revolver WITHOUT using booleans (and keeping it all quads)? I've tried a bunch of methods including aligning edges and using quad draw over a sphere but it just never seems to look right.



    This is the example of what I'm trying to model, Thanks guys!
  • esPhys
    Offline / Send Message
    esPhys polycounter lvl 7
    Hey guys, I was wondering what the best way to model this complex sphere intersection for a Revolver WITHOUT using booleans (and keeping it all quads)? I've tried a bunch of methods including aligning edges and using quad draw over a sphere but it just never seems to look right.

    This is the example of what I'm trying to model, Thanks guys!
    Gotta post what you've tried to make so far.
    I think there should be some examples in the thread about similar shapes though. I know the thread is hard to parse because of how old and long it is with broken links etc. but you can also google 'hardsurface topology' and you'll get a lot of reference images.
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter

    Is there a specific reason why this shape can't be made with booleans and must be all quads?

    Yes it's possible to do this using old school techniques but it's all manual operations that mimic the results without the speed advantage of booleans. It's also possible to keep the mesh all quads but that requires planning the edge flow and matching line segments on every little detail.

    Both of these options are labor intensive and at the very best they won't be any better than using a contemporary boolean workflow. In most cases it will take longer, require more re-work and the results will probably be worse.

    If you're trying to learn subdivision modeling there are better ways. Using booleans to block out the shape, running an automated cleanup and doing some light Sub-D modeling work would be faster and it works just fine without all quads.

    A more contemporary workflow like Boolean + zBrush (or QuadRemesher) would be even faster and produce good results. The boolean + bevel / round edge shader workflow is also gaining traction but it has it's own limitations. Lot's of options to choose from.

    Unless there's some obscure technical limitation: newer workflows are more efficient and produce better results. I'm heavily biased towards Sub-D workflows but I use booleans all the time and in a production setting it comes down to speed and quality so the newer methods often win.
  • UhhNope
    sacboi said:

    You're almost there.

    Just learn too work with less geo especially when blocking out or defining form/s.

    EDIT: Plenty of examples populated throughout this thread, in fact the very first post illustrates a low poly approach.


    Thank you all for the replies. It sounds like I wasn't planning ahead enough when deciding how much geometry to put on each pipe length. It also sounds like I should look up some boolean modeling workflows for Blender.
  • armagon
    Offline / Send Message
    armagon polycounter lvl 11
    I was reading a few tuts from Thiago Klafke and I got curious about this shape:


    How does it look so good in-game? How can it be so round? HOW THE F*CK do I model that?


  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @armagon There's a few ways to do this but here's one:

    1. Block out the basic roof geometry.
    2. Chamfer the large corners with an appropriate number of segments.
    3. Add two lööps to the middle of the fascia.
    4. Inset the poly strip in the middle.
    5. Chamfer the top and bottom edges to create trim strips.
    6. Inset / solidify the trim strips.

    Clean up any edge loops that aren't being used by the trim sheets or roof textures.


  • ArcticTauntaun

    Is there a specific reason why this shape can't be made with booleans and must be all quads?

    Yes it's possible to do this using old school techniques but it's all manual operations that mimic the results without the speed advantage of booleans. It's also possible to keep the mesh all quads but that requires planning the edge flow and matching line segments on every little detail.

    Both of these options are labor intensive and at the very best they won't be any better than using a contemporary boolean workflow. In most cases it will take longer, require more re-work and the results will probably be worse.

    If you're trying to learn subdivision modeling there are better ways. Using booleans to block out the shape, running an automated cleanup and doing some light Sub-D modeling work would be faster and it works just fine without all quads.

    A more contemporary workflow like Boolean + zBrush (or QuadRemesher) would be even faster and produce good results. The boolean + bevel / round edge shader workflow is also gaining traction but it has it's own limitations. Lot's of options to choose from.

    Unless there's some obscure technical limitation: newer workflows are more efficient and produce better results. I'm heavily biased towards Sub-D workflows but I use booleans all the time and in a production setting it comes down to speed and quality so the newer methods often win.

    Hey man, thanks you so much for the detailed writeup!

    I think it's mainly because when I was learning 3D it was ingrained that (Booleans are bad, Quads are good), and to not have messy topology to be able to SubD. Should I be converting the boolean'd shape into all quads or should I just leave it?

    I've seen some ProBoolean 3DS -> ZBrush workflow guides, but do you know if there's something similar for Maya?

    Thanks!
  • miners
    Hey guys can you help me with this model i am trying to model this gun part but i just can't find the right way to model it any suggestions on how i can fix up my model. This is what i managed to do but can i cant get the smoothing just right.  =)

  • armagon
    Offline / Send Message
    armagon polycounter lvl 11
    @armagon There's a few ways to do this but here's one:

    1. Block out the basic roof geometry.
    2. Chamfer the large corners with an appropriate number of segments.
    3. Add two lööps to the middle of the fascia.
    4. Inset the poly strip in the middle.
    5. Chamfer the top and bottom edges to create trim strips.
    6. Inset / solidify the trim strips.

    Clean up any edge loops that aren't being used by the trim sheets or roof textures.

    Thank you so much for the detailed answer! It makes much more sense now!
  • Maya2025


    I'm trying to make something that looks like in the picture, repeating pattern with a round circle, but

    I have no idea how to do it because of my little knowledge about modeling.

    can anyone help me out with this ?  i really appreciate your guys' help






  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter

    It kind of looks like there's a step between some of the parts which suggest they're separate pieces. If that's the case then it would be easier to model the parts individually. Is it some kind of stamped metal shroud?

    For the mesh you already have: the simplest solution is to run an inset on the flat area, merge down any messy verts and connect the edge loops where necessary. The loops running into that area will end on a flat so it can be a little messy.


    If you were starting over from a clean block out you could do a chamfer along the hard edges and clean up any stray verts or pinching edges. On the actual shape there's a few spots where you can leverage existing geometry as edge loops. This example doesn't have those areas.

  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter

    It depends on a lot of things. Does the shape need to maintain those flat segments? Does the shape need to be completely round? Does the shape need to be made in zBrush?

    In a standard modeling package I'd try something like this:


    Start with a base shape. Cut in some edge loops. Create a single repeating part, copy it and merge it down. To create the single part: scale the outside verts along the edges. Inset. Inset. Add a loop for support. Copy and merge down the segments. This technique will still work if you need the hard segments. Adjust the starting shape's segments as needed. There's a few other ways you could cut in geometry if you need more accuracy. Again it all depends on the size of the object and how you're going to use it.




  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter

    That's understandable: the reality is it's more contextual than absolute. The key is learning when it's ok and when it's not. Sometimes it even comes down to something as simple as budget and object scale. Low budget, small stuff can't be polished to the Nth degree if you plan to make money off it.

    Using booleans to get to a base mesh can be a real time saver. Where you want to go from there depends on the workflow. The basic geometry in the example below was done in less than 15 minutes.

    Once the block out is approved, that basic geometry could be quickly built up into a sub-d mesh by chamfering the sharp edges, converting the n-gons to quads / tris and adding or adjusting the loops on the flats as needed.

    Working through a boolean to sub-d workflow requires all the same ideas regarding segment matching, edge position, loop layouts and so on. (The cylinders for the holes are behind the n-gons and will be merged in later once the outer edge loops are in place.) I left some triangles and n-gons in the mesh to show that it still works without being all quads. (Edited to add the sub-d image.)



    Below is a quick dynamesh and polish pass in ZBrush using the same boolean base. On the low to mid end of the skill range, I think it'd be tough to beat that result in under 20 minutes using traditional workflows. That's why the boolean to ZBrush workflow is so powerful.


    As for how clean that topology needs to be... If it's going through some re-meshing process like ZBrush or QuadRemesher then the answer is not very. Below you can see the triangulated mesh that went in and the result that came out of ZBrush.



    Unless there's some hard limit or specific reason why the mesh has to be all quads, the answer is it's ok to have some tris and n-gons in a traditional subdivision model if the mesh subdivides correctly without pinching, stretching, etc.

    I'm not a Maya user but there's a couple threads on here about people using a Maya to ZBrush workflow. If you search around a bit you might be able to find other resources. https://polycount.com/discussion/186757/help-maya-boolean-to-zbrush-dynamesh-workflow

    If you can get the boolean mesh to the state shown above then ZBrush should be able to handle it without any major problems. Even though it's for Blender, Victor Kam's quick guide illustrates the process. https://www.artstation.com/artwork/GXWBoQ

Sign In or Register to comment.