Home Technical Talk

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

Replies

  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    ronigeva said:
    The intention was to keep quads Without triangles or n-gons

    And why is that? it's a flat surface..


  • Scriptorium
    Offline / Send Message
    Scriptorium polycounter lvl 3
    Hey!

    How in the world do I model a rectangular(-ish) hole on a curved mesh without these shading artifacts/lumps?

    My attempts with and without a flow around the hole both produce terrible shading, I also tried redistributing density but to no avail, I think there's something wrong with the flow.


  • sacboi
    Offline / Send Message
    sacboi high dynamic range
    You're almost there!

    Establishing supporting geometry (uniformly spaced poly strips/edge loops) too control in this instance what appears to be a cut/hole topology upon a reflective surface possibly a 'vehicle'?!

    Is what I'd normally do when utising a similar workflow. Although not necessarily by hand, hence the overall uneven vertex positioning I see here, now if indeed the case I'd suggest a typical fix to smooth that curved region via the Shrinkwrap modifier:


    (Essentially, duplicate mesh as a parent, prior to those edited cuts)

    or

    Another option shared a few posts above yours - https://polycount.com/discussion/comment/2739935/#Comment_2739935
  • Scriptorium
    Offline / Send Message
    Scriptorium polycounter lvl 3
    @sacboi Thanks for you help and the video you linked! 

    I actually had several problems, most of which from the lack of understanding of the fact that I'm trying to do something on a surface that follows not one, but several curves. So the solution was to patch the hole, relax the topology, dissolve useless edges (that did not contribute to the desired curvature), and then achieve the desired curvature using curve interpolation (in this case it was MiraTool's Curve Stretch). 
    I probably should've made a duplicate to shrinkwrap to later, but I ended up just cutting the hole and Extrude -> Shrinking it. 

    The result still looks a bit freehand but I'm fine with it in this case. The 5 edge polygon on the convergence of the plates is intentional as I feel it helped the shading (and I think this is the use case of the 5 edge polygon method mentioned in Pixar modeling docs, but correct me if I'm making a horrible mistake.



    So, my key takeaways for this would be:
    1. Make sure you are aware of the curves of the surface you're trying to work with
    2. Cut holes in the surface after you finish modeling the surface itself
    3. Have less messy sculpts? 

    Am I still missing something obvious?

    EDIT: Forgot to mention I actual decided to go with a bit of a different shape for the hole itself, but nothing that would've altered the way I tried to resolve this, this was simply an artistic choice at the moment.
  • sacboi
    Offline / Send Message
    sacboi high dynamic range
    Nice solution, plus in my opinion it also depends upon prior planning in order to mitigate unforeseen problems.
    So, my key takeaways for this would be:
    1. Make sure you are aware of the curves of the surface you're trying to work with
    2. Cut holes in the surface after you finish modeling the surface itself
    3. Have less messy sculpts? 

    Am I still missing something obvious?
    Yes, I'd agree with your thoughts and no at the moment, nothing obvious comes too mind.
  • 99499
    Offline / Send Message
    99499 node
    I'm afraid to make a new thread about this because the last one did not go well, but does anyone have any resources for learning about "combining" 2 different objects with different vert counts? For example, I'm about to make a pair of shoes for my character who is low poly, but the shoes area an important part of the design and will be a higher poly. Not sure if I should just leave them as separate objects and parent them or what the common workflow for this is. 
  • LoneRanger
    Offline / Send Message
    LoneRanger polycounter lvl 4
    Hi, I'm creating game-ready model and I've got big problem with one part of it when trying convert to high poly for normal map baking.

    This is my low poly model with sharp edges marked:


    This is how it looks after applying Bevel (weight method) with 3 segments and then Subdivision surface modifiers:


    I've tried a lot of things - adding some support edges, various bevel options, cuts, etc. and nothing works. Only by marking sharp edge like on this image below, I can solve this one intersection, but then the other one is more broken.


    I don't know how to solve it. Even if I would like to start this model from scratch I can't imagine doing it in another way. 
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @LoneRanger Some of the smoothing artifacts may be the result of using an uneven number of bevel segments on areas where multiple surface angles converge. Try using the following bevel settings and see if that improves the smoothing behavior: 2 segments, 1.0 Shape, Arc Miter Outer.

    Increasing the mesh density tends to increase overall shape accuracy but subdivision modeling is something of an approximate process so there's almost always going to be some level of minor shape inaccuracy. Depending on various factors (player view distance, object size, texture density, etc.) minor smoothing artifacts on small parts may be a complete non-issue. Using the minimum amount of geometry to accurately hold the shapes balances shape accuracy with edibility.

    There's two common strategies for laying out the cage mesh topology on this type of shape transition: placing new support loop geometry over the existing edge segments that define the shapes or using the existing shape geometry as support loops and routing new support loop geometry between the existing edge segments.

    Working off the existing topology grid can be a quick and easy way to block out shapes and define sharper shape transitions but it also tends to create visible smoothing artifacts on the curved surfaces of low density cage meshes. Below is an example of what this process and topology can look like.



    Using existing geometry as support and placing new geometry between the edge segments that make up the curved surfaces can help reduce smoothing issues around shape transitions and tends to require less geometry but may produce slightly softer corners. Below is an example of what this process and topology can look like.



    Both of the previous examples start with a base mesh and most of the support loops are added with a bevel / chamfer operation. Adding additional shapes to the block out and adjusting the support loops is a relatively straightforward process. Loop select and dissolve edge operations are a quick way to clean up any excess geometry. Below is a comparison of some variations on the previously mentioned topology layouts.



    There's a few different ways to model this shape but here's an example that's fairly straightforward, easy to model and should be compatible with adding support loops via non-destructive modifiers.


    To recap:
    Determine whether or not minor smoothing artifacts will be visible to players before investing a significant amount of time in improving the results. Working off the existing geometry can be quick and easy but tends to create smoothing artifacts on low density meshes. Using the existing geometry as support and placing additional geometry between the existing edges of curved surfaces can help reduce smoothing issues around shape transitions.
  • ronigeva
    What can I do to avoid this pinch corners?

    i have "clean" topology and all the model is QUADS

  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @ronigeva Looks like the intersecting geometry falls directly on top of the cylinder wall edge segments and this is causing deformation when subdivision is applied. Placing intersecting geometry between the existing cylinder wall segments will allow the existing geometry to act as support (which provides room for additional geometry) and will also help preserve the concentricity of the inner cylinder wall.

    With enough geometry it is possible to place intersecting geometry directly in line with cylinder wall segments but it's more efficient to use the existing cylinder wall geometry as support and just match the segment count so the intersecting geometry resolves as all quads. Here's an example that shows the basic topology.


    Support loops can be added once the basic topology is established. How sharp the edges need to be will depend on the object but any difference in edge width can be taken up in the support loop that runs around the base of the intersecting geometry where it meets the cylinder wall geometry. Below is an example of what this could look like.



    Subdivision modeling tends to be an approximate process so there's almost always going to be minor shape imperfections. Increasing the mesh density tends to increase the shape accuracy but tends to decrease editability and efficiency. It's important to look at the size of the object and average view distance as minor smoothing artifacts may not be visible. What's acceptable will depend on how the model will be used so it's important to balance accuracy with efficiency.

    Below is a comparison of wide and tight support loops on the rest of the shape. It's possible to maintain relatively sharp edges with minimal geometry so use the appropriate amount of geometry to support the shapes. Often it's a case of close enough is good enough.


  • Blaizer
    Offline / Send Message
    Blaizer interpolator
    This method is achieved extruding the faces in the tube, and applying simple bevels. It's similar to what you have @ronigeva

    Using chamfers on the edges produces better results. But artifacts won't dissappear at all, because this kind of meshes needs more geometry. If you use pentagons, the mesh has less tension or shading artifacts.




    On the other hand, the top view is good enough, but it's better to use more geo for better results.

  • Rolf
    Offline / Send Message
    Rolf polycounter lvl 3
    Good morning every one.After long time i am return in maya for continue modeling my ship,main reason why i stop its lumpenes in my model.How i must to fix this problem (screen below).how to smooth out these lumpenes.if i use nurbs modeling technique this problem disappears,but there are areas that cannot be created using nurbs modeling.
  • LoneRanger
    Offline / Send Message
    LoneRanger polycounter lvl 4
    I'm starting to more and more understanding working with Subdiv but I encountered a problem with simple object. From the distance those artifacts will not be visible, but I really like to know how to solve intersections like this.
    This is my model with hard edges marked for bevel (I've tried method without it, just placing manual support loops but I got the same problems).


    This is what is happening:


    But when I connect those corners with hard edge below, corners then are looking fine, but my smooth shape is broken:


    I've tried placing loops in many places, also just around this problematic corners on second screenshot, but then there is a soft bump/stretch going on this circural smooth side face. 
  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    I'm starting to more and more understanding working with Subdiv but I encountered a problem with simple object. From the distance those artifacts will not be visible, but I really like to know how to solve intersections like this.
    This is my model with hard edges marked for bevel (I've tried method without it, just placing manual support loops but I got the same problems).


    This is what is happening:


    But when I connect those corners with hard edge below, corners then are looking fine, but my smooth shape is broken:


    I've tried placing loops in many places, also just around this problematic corners on second screenshot, but then there is a soft bump/stretch going on this circural smooth side face. 

    you do not have enough geo to support that extrude/intrude.


    And this was done with a 32 sided cylinder. The same method applied.
  • iacdxb
    Offline / Send Message
    iacdxb polycounter lvl 6
    Hi,

    Any idea how to make three triangles to quad....? Thanks.


  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    iacdxb said:
    Hi,

    Any idea how to make three triangles to quad....? Thanks.




  • iacdxb
    Offline / Send Message
    iacdxb polycounter lvl 6
    Thanks Wirrexx, I don't think I can do cut in middle bottom. see this...! Thanks.


  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    iacdxb said:
    Thanks Wirrexx, I don't think I can do cut in middle bottom. see this...! Thanks.



    ah another solution. It's not fully quaded, but two of your 3 tris are fixed.



  • iacdxb
    Offline / Send Message
    iacdxb polycounter lvl 6
    Yes... Thanks. it works.
    ...
  • Doguib7
    Offline / Send Message
    Doguib7 polycounter lvl 6
    Hey there folks!
    I am looking for new ways or approaches to model a 3D cartoon character! I know this is pretty relative but generally, there are some useful solutions to get proper deformations and topology! The aim is a pretty standard 3D cartoon character!
    Please share with me whatever could be helpful!

    Thanks in advance!
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @99499 If the shoes are an important part of the character that the players will interact with and view up close then it probably makes sense to model them as separate objects and rig them to match the rest of the character. A lot will depend on what the shoes are supposed to do and what the rest of the workflow looks like.

    @Doguib7 Broad, open ended questions like that rarely provide satisfying answers. More specific questions, accompanied with images or examples of specific modeling or topology issues, tend to be more answerable. Some general information about character modeling and animation is available on the polycount wiki. ArtStation learning may also have some general resources and there's a lot of YouTube videos about character modeling, rigging and animation.

    http://wiki.polycount.com/wiki/Animation

    @Rolf The undulations in the subdivided mesh are likely caused by the diagonal twist in the edge loops that run up and down the hull.

    There's a few different ways to approach modeling this but try blocking out the shape of the hull with vertical segments that match the actual shape of the ship's frames. Placing these frames at regular intervals (matching the stations in the ship's plans) should make it easier to fair the lines of the hull. Here's an example of what this could look like.


  • Doguib7
    Offline / Send Message
    Doguib7 polycounter lvl 6
    @FrankPolygon
    Thank you very much, buddy! I'll take a look!
  • stuffinmyhead
    Offline / Send Message
    stuffinmyhead polycounter lvl 6
    Hello there.

    I am trying to model the FLIR camera of a seahawk helicopter. But i have problems with pitching and pretty much bad topology. The objective is to model this cylinder with all the flat cuts from these refrence images
    https://imgur.com/a/WkZprs6

    And here is my attempt at doing it. As you can see when i apply turbosmooth i get allot of bad pintching and artifacts. Could you please help me out here?
    https://imgur.com/a/Yy3GL9I

  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @stuffinmyhead It looks like the flat areas are cut out of the cylinder's shape rather than added to it. With this type of shape intersection it's important to keep the cylinder wall segments straight and keep the intersecting shapes perfectly flat.

    Start by blocking out the basic shape and use some planes or rectangles to determine the exact angle for each of the flat surfaces. Adjust the number of segments in the cylinder to support all of the intersecting geometry then subtract the flat areas from the cylinder. The existing cylinder wall geometry can be used as support by placing the intersecting geometry between the existing edge segments. How much geometry is required depends on how accurate the shapes need to be. If done correctly the subtracted areas will be perfectly flat and the perimeter of the new shapes will match the line drawings.

    Below is an example of what this process could look like.



    Triangles and n-gons are fine as long as they aren't causing any major smoothing issues. Minor smoothing issues may not be visible to players because of the matte surface finish and the relatively small scale of the object when compared to the rest of the vehicle.

    When deciding how accurate the shapes need to be, consider how closely and how often players will view the model. Use the appropriate amount of geometry to accurately hold the shapes and avoid over complicating the cage mesh.

    The last few pages cover a lot of other cylinder shape intersections and there's some additional information that may be helpful so it might be worth skimming through and reading about other cylinder shape intersections.
  • ConvexSurface
    @ConvexSurface Overall it looks like you have the right idea but sometimes connecting directly to a curve's existing polygon grid can cause a lot of smoothing issues. In these cases it's often better to place the intersecting geometry between the existing segments of the curved surface and use the existing curve geometry as support loops.

    Both of these topology strategies are legitimate methods for combining shapes but which topology layout is best will depend on the density of the cage mesh and the desired smoothing behavior around the shape transitions. The appropriate amount of geometry and the size of the supports loops will ultimately depend on the the desired edge sharpness and the intended use / view distance of the in-game model.

    When encountering minor shape inaccuracies and smoothing issues: it's always worth considering whether or not players will ever view something closely enough or often enough to notice. There's marginal returns on trying to improve something that won't be noticed by most players.

    There's a similar example of how to control the smoothing around a tab added to a curved surface on the previous page. One of the quicker ways to create this type of topology for an approximate copy of the part in the engineering drawing would be to inset the faces on the curved surface then extrude the profile of the tab shape and add the rounded features and select support loops with a series of bevel / chamfer operations. The rest of the interior support loops can be added with inset operations. Below is an example of what this modeling process could look like.



    One potential issue with this modeling strategy is the starting density of the curved shape can become a limiting factor for subsequent modeling operations. This makes is quite possible (if not extremely likely) that there will be situations where the initial segment count of the curve will interfere with changes to the surrounding geometry and may require significant manual re-work after a lot of time is already invested in the cage mesh.

    Resolving the majority of these shape accuracy and shape intersection issues during the block out will save a lot of time later on in the modeling process. Manually forcing geometry into position is generally both time consuming and relatively inaccurate so avoid it whenever possible and focus on finding the right tools and order of operations to develop the shapes quickly and efficiently.

    Planning out how to use the minimum amount of geometry required to accurately hold all of the major shapes and matching the segments of adjacent shapes whenever possible will make it easier to edit the mesh and merge shape intersections. This is why it's generally considered best practice to block out the shapes of all the major features and figure out how most of the shape intersections will interact before adding a significant amount of support geometry or secondary details.

    Here's an example where the tab and the curved surface are developed as separate meshes and merged using a boolean operation. Intersecting shapes are placed between existing edges for support and additional support loops around the major shapes are added with a bevel / chamfer modifier. This leaves a few stray vertices that can be dissolved or snap merged into the adjacent topology. Since flat surfaces are largely immune the effects of triangles and n-gons there's no real reason to extend extra edge loops across the mesh unless there's smoothing artifacts.


    The engineering drawing seems to suggest that the tab has a fillet around the base where it joins to the part. The previous example uses a different modeling process but shares the same general topology layout with the first example. The loop that runs around the shape intersection is generally used to take up any difference between intersecting shape's geometry and the underlying curve's geometry but it can also be used to add a fillet transition between the shapes.

    Here's an example of what it looks like when a fillet is added by using a different value during the bevel / chamfer operation. Adding a fillet larger than the average width of the support loops is something that should probably be addressed separately and earlier in the block out.



    Another possibility is the curved section of the tab blends into the flat area just past the curved surface. In cases like this it's generally fine to connect directly to the underlying geometry since all of the support loops run perpendicular to the curved portion of the mesh. Below is an example of the same boolean union with bevel / chamfer support loops. This topology layout would also work with the extrusion modeling approach.



    Here's a comparison of the three topology layouts covered. It's possible to combine elements of each depending on how and where the tab joins the curved surface. The overall width of the support loops will control the edge sharpness and can be adjusted as necessary.

    Independently adjusting the number of segments in the curved surface becomes more important if very sharp edges are desired. For some shape combiniations (with significant differences in edge width like having sharp front transitions with fillets on each side) it may be necessary to run some of the bevel / chamfer operations separately.



    Below is an extra example that shows how solving shape intersection issues early in the block out stage makes it easier to add support loops and should also help reduce the amount of geometry used. All of the major shapes in these examples were created using tools that create consistent curves and maintain co-planar geometry. Finding the correct order of operations and relying on tools to make changes to the mesh should eliminate most of the manual vert pushing that tends to introduce shape inaccuracies.


    Razor sharp edges and CAD levels of perfection aren't exactly the strong points of subdivision modeling processes. With subdivision modeling there's almost always going to be some degree of shape inaccuracy and edge softness. This is actually very useful because controlling edge width / sharpness is an important part of controlling the visual read of baked normals and the approximate nature of subdivision modeling makes it easier to control complex shapes with a lot less geometry. It's also worth reiterating that depending on view distance, texture size and object scale these minor smoothing issues or other imperfections may be complete non-issues.

    Recap:
    Some of the pinching and webbing smoothing artifacts can be corrected by fully developing the shapes during the block out, placing the intersecting geometry between the existing geometry along the curved surface and maintaining a consistent edge width on the support loops. Connecting directly to the grid topology of the curve is fine wherever the intersecting geometry runs perpendicular to the curve. The tab in the engineering drawing appears to have a fillet around where it joins the rest of the part so depending on how large the object is and how players will interact with it the existing geometry may be adequate.
    Thank you very much for the detailed and prompt explanation Frank (and apologies for my late reply). There's one thing thing I haven't quite understood though: What operator(s) was used to perform the first step in the final example you posted? To make the connecting mesh between the tab and curve diagonal? Some kind of shear transform? I can't seem to reproduce it.
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @ConvexSurface You're welcome.

    Correct: The sloped surface was created by a shear operation with the pivot point set to the top edge of the shape. Depending on the shape it may be necessary to rotate everything into position so everything is in plane with direction of the shear operation. Afterwords everything can be rotated back into position. An alternate option would be to include the slope at the very beginning of the block out and use a boolean operation to join the shapes. It all comes down to where these changes happen in the order of operations.
  • stuffinmyhead
    Offline / Send Message
    stuffinmyhead polycounter lvl 6
    @stuffinmyhead It looks like the flat areas are cut out of the cylinder's shape rather than added to it. With this type of shape intersection it's important to keep the cylinder wall segments straight and keep the intersecting shapes perfectly flat.

    Start by blocking out the basic shape and use some planes or rectangles to determine the exact angle for each of the flat surfaces. Adjust the number of segments in the cylinder to support all of the intersecting geometry then subtract the flat areas from the cylinder. The existing cylinder wall geometry can be used as support by placing the intersecting geometry between the existing edge segments. How much geometry is required depends on how accurate the shapes need to be. If done correctly the subtracted areas will be perfectly flat and the perimeter of the new shapes will match the line drawings.

    Below is an example of what this process could look like.



    Triangles and n-gons are fine as long as they aren't causing any major smoothing issues. Minor smoothing issues may not be visible to players because of the matte surface finish and the relatively small scale of the object when compared to the rest of the vehicle.

    When deciding how accurate the shapes need to be, consider how closely and how often players will view the model. Use the appropriate amount of geometry to accurately hold the shapes and avoid over complicating the cage mesh.

    The last few pages cover a lot of other cylinder shape intersections and there's some additional information that may be helpful so it might be worth skimming through and reading about other cylinder shape intersections.

    Thank you! :)
  • dan001
    Offline / Send Message
    dan001 polycounter lvl 6
    Hi everyone, I'm quite new to this. I wonder what would be your approach on this piece so the subdivision is correct.
  • Herbert
    Hi guys, i am trying to model this piece as a practice but i faced some problems on areas marked.
    In area A, i could not resolve the polygons to be all quads, so i tested with tris and n-gons and surprisingly does not cause any shading issues, however shading in area B suffer from bad topology i guess? So how would i solve the problem in this case. Also i am feeling the need to ask this question is there a way that i can achieve all quads in area A? just for the sake of doing all quads? 

    Thank you guys for your time!! 
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @dan001 In most cases there's room for improvement but broad or open ended questions rarely lead to specific, actionable feedback. What's best will often vary depending on the desired outcome and constraints of a specific use case.

    Getting good answers comes down to figuring out exactly what you're looking to improve and asking specific questions. A good place to start is to compare the expectations to the results and identify any specific issues or areas where the results could be improved.

    Images of the subdivision wire frames would be helpful and context is also important so be sure to include additional information about intended use, object size, player view distance, etc. This additional information will help the community provide more focused feedback.

    @Herbert Unless there's a specific technical requirement for all quad grid geometry there's often minimal benefit to manually re-working the mesh topology to create it. Triangles and n-gons are generally fine, as long as they aren't causing any major smoothing artifacts. This topic comes up often and there's a recent discussion about it on page 177.

    Increasing the geometry density of the cage mesh tends to increase the overall shape accuracy but starting with too much geometry and working on a small part of the mesh (without planing out the topology flow around the major shapes) can lead to situations where's it's difficult to cleanly merge additional surface features. Take the time to block out the basic shapes, match segment counts for adjacent shapes and figure out the topology routing before adding support loops.

    Try staring with less geometry and orienting the shapes so the intersecting geometry falls between the edges of the existing geometry. Block out all of the major features and adjust the segment counts to match the surrounding geometry. From there it's a fairly straightforward process of adding support loops with a bevel / chamfer operation and connecting a few stray edges to the surrounding geometry. Below is an example of what this process could look like when starting from a 24 segment sphere.


    One of the issues with a regular sphere is the number of segments in a given area tends to increase near the poles. Depending on the orientation and number of intersecting shapes it may be beneficial to use a quad sphere since the segment count is fairly even on all sides and the vertical segments near the center of the sphere tend to to be fairly straight. Below is an example of what this process could look like when starting with a 24 segment quad sphere.



    Recap:
    Both of these examples highlight the importance of blocking out the major features, using the appropriate amount of geometry and properly orienting the intersecting geometry to direct the overall flow of the mesh.
  • sacboi
    Offline / Send Message
    sacboi high dynamic range
    You've the patience of a saint :D
  • IronLover64
    Offline / Send Message
    IronLover64 polycounter lvl 2
    How do I model the handle? I tried doing it this way, but I couldn't get the intersection with the curve and triangle correctly.
  • IronLover64
    Offline / Send Message
    IronLover64 polycounter lvl 2
    Forgot to mention that I forgot to shrink the end of the triangle
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter

    Try to capture all of the major shapes in the block out before adding support loops and applying subdivision. It may also be helpful to break the grip down into separate parts and join the front and back together latter on in the modeling process.

    Evaluating the shapes in the reference images: The front part of the grip has a round over and the lower part of the frame has an angled profile. These two shapes are blended together with a radiused corner transition above the trigger.

    There's a lot of ways to model this shape but one of the simplest approaches would be to setup an edge loop path with the exact width of the round over and cut in perpendicular edges to act as stop points for the radiused corner transition area. Dissolve or merge the geometry along the front bottom of the frame to create the tapered triangular shape and make sure the corner geometry above the round over loop flows directly into this new shape. Select the outer edges (highlighted in white) and run a percentage based bevel / chamfer operation to add the round over and blended radius transition in a single operation.

    A similar process can be used on the back of the grip. Block out the basic shape, create and taper the end of a support loop path for the round over, use a bevel / chamfer operation to add the curved shape to the back strap and add the round over with a percentage based bevel / chamfer operation.

    From there it should be a relatively straightforward process of cleaning up any stray geometry (merge by distance, limited dissolve, edge dissolve, etc.) and adding additional support loops to create the final subdivision ready cage mesh. Below is an example of what this process could look like on the front and back of the grip.

  • JBurk
    Hi all, newish modeler. I have using blender for 8 months now. I am modeling a biplane and I am struggling with the metal cap that connects the cylinders and the propellor. These are the two versions I have tried. I would like to use only quads but I don't know how to model it so the portion where the two cylinders meet don't create a bump at the top of that triangular quad.


    What would be the proper way to model this with quads?
  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    JBurk said:
    Hi all, newish modeler. I have using blender for 8 months now. I am modeling a biplane and I am struggling with the metal cap that connects the cylinders and the propellor. These are the two versions I have tried. I would like to use only quads but I don't know how to model it so the portion where the two cylinders meet don't create a bump at the top of that triangular quad.


    What would be the proper way to model this with quads?
     
    It would help to add references to what you want to do! =)
  • JBurk
    oh right! so that big metal piece in the center is what I am aiming at. Im going stylized so I am ignoring the bolts that are in between each cylinder.
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @JBurk Overall you have the right idea and are pretty close with the second iteration. A few pages back there's a couple of relevant discussions about cylinder to cylinder intersections and multiple cylinder intersections close together. The basic idea is to block out all of the key features then adjust the segments of each part to match the number of segments in the adjacent geometry.

    For a radial engine model: It's probably best to start by establishing the size, number and location of all the cylinder jugs then add the body of the crank case and adjust those segments to match the intersecting geometry. Below is an example that shows what this process could look like.

    Use a tool like spin duplicate or a modifier like an object rotation offset array to copy and position the basic cylinder geometry into an exact division of a circle. Add a cylinder primitive for the crank case and adjust the number of segments until the edges along the cylinder wall of the crank case mesh lines up (reasonably close is good enough) with the intersecting geometry of the cylinder jug bases.

    Join the cylinder jug meshes to the crank case mesh with a boolean union operation and clean up any stray geometry. Since this is a stylized engine model any minor differences between the geometry of the crank case and cylinder jug bases can just be taken up directly at the root of the intersection.

    (An alternate method would be to add the support loops around the base of the shape intersection and dissolve any stray geometry so the offset between the shapes is taken up inside of the support loop. This process is discussed briefly in the topics linked to above.)

    With this model the shapes at the front of the crank case are added after the boolean operation but (depending on the complexity of the crank case) it may be necessary to adjust all of the shapes first then run the boolean operation to join the cylinder jug bases to the crank case. In which order the shaping and boolean operations takes place depends entirely on the complexity of the shapes. In this example it was only done in this order to simplify the process of joining the shapes. Adjust the order of operations as necessary.

    Select the edges around all of the major shapes and shape intersections then add the support loops with a bevel / chamfer operation.



    Here's another example of how the same process can be modified by using a different order of operations to create different shapes at the nose of the crank case. Radial engines from this era tend to have cast crank cases with machined bores so some undulations and bumpiness in the shapes may be acceptable. Deciding whether or not to copy some of these shapes from the reference image depends entirely on how abstract and simplified the features of the stylized model need to be.



    Recap:
    Block out the major features.
    Match the segments of intersecting shapes to the adjacent geometry.
    Include any complex shapes near the intersecting geometry before running any boolean operations.
  • JBurk
    @FrankPolygon Epic! Thank you so much! 
  • JBurk
    @FrankPolygon

    Here is what I got after blocking it out and using a boolean. I am attempting to get as perfect a spherical shape as possible and the only way I was able to achieve that was dropping the poly count a lot. with this model it is really hard to get a perfectly spherical cone with a hard transition into the cylinder jugs. This is totally acceptable for what I am doing but for the sake of experimentation, I would like to figure out if I can get this transition to be sharp and keep my perfectly cylindrical body.

    Would I just need a ton of geometry?

    And I am having trouble getting appropriate geometry because the cone tapers 
    On the left I can get it to line up but on the back end because the cylinder gets wider I cant match up the geometry. 

    This is all stuff I that I probably don't need but I would really like to be able to achieve.
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @JBurk Looks like you're on the right track and the process just need some minor adjustments.

    Once the block out is complete (and all the shape intersections have matching segment counts) just maintain the same segment count ratios when increasing the geometry density of the cage mesh. The appropriate amount of starting geometry really depends on how detailed the shapes need to be. Too much or too little starting geometry can be equally difficult to work with.

    Boolean operations tend to leave behind stray geometry and how this geometry is merged down will effect the support loop placement and subdivision smoothing behavior. Three ways to handle this stray geometry are: merging into the vertices at the base of the intersecting geometry, merging down into the vertices on the side of the larger shape or creating support loops around the shape intersection to take up the offset between the shapes and dissolving any left over geometry.

    The first two methods of merging down stray geometry into the existing shapes work well for quickly generating approximate shapes with softer shape transitions but placing support loops around the intersecting geometry and dissolving any left over geometry can provide a higher degree of shape accuracy.

    Controlling the subdivision smoothing behavior around the shape intersections requires placing support loops on both sides of the edge loops at the base of the shape intersection. The topology and width of these support loops will control how soft or sharp the shapes are when subdivision is applied. Wider support loops produce softer shape transitions and narrower support loops produce sharper shape transitions. Make sure the support loops run all the way around the shapes and shape intersections.

    Support loops can be added with either a bevel / chamfer operation or a series of loop cut operations. Both methods produce slightly different topology layouts and have slightly different smoothing behaviors.

    Here's an example that shows how the larger shape deforms when the stray geometry is merge down into the base of the smaller intersecting shapes. Included in this example is a comparison of support loops added with a bevel / chamfer operation and support loops added with a series of loop cut operations.


    Here's an example that shows how the smaller shape deforms when the stray geometry is merged down into the vertices created along the wall of the larger shape. Included in this example is a comparison of support loops added with a bevel / chamfer operation and support loops added with a series of loop cut operations.


    Adding additional geometry to the larger shape (before running the boolean operation) provides additional support around the shape intersection that can limit certain types of deformation and can increase shape accuracy.

    How much additional geometry is required depends on the shape, topology routing, smoothing behavior and desired level of shape accuracy. Placing the appropriate amount of geometry in the right spot will ensure there's enough support to take up any difference between intersecting shapes or can provide a place to merge down stray geometry without effecting the rest of the mesh.

    Below is an example that shows the effects of placing additional support geometry before running the boolean operation and adding the support loops with a bevel / chamfer operation. (Top is merged to the base of the intersecting geometry and bottom is merged to the larger shape.)

    Using a bevel / chamfer modifier (controlled by edge weights) is a flexible way to add adjustable support loops and maintain a highly editable base mesh but it does effect the smoothing between the intersecting cylinder shapes. Whether or not this acceptable depends on the overall goals for the project.


    Here's an additional example where the support geometry was placed before the boolean operation and the support loops were added with a series of manual loop cut operations. Note the minor difference in smoothing behavior between the intersecting cylinder shapes.



    The shapes at the back of the engine can be merged down into the crank case using the same methods described above and the geometry should line up without causing any major smoothing issues.

  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    @FrankPolygon always fun too see you do your stuff! =)
  • JBurk
    @FrankPolygon
    Woohoo! 
    Thank you so much for taking the time to give such a thorough explanation. You truly are a legend. 
    This geometry for joining the corners is a game changer. It helped my stretch the back of the cylinder jug geometry when it didn't quite line up with the edge the same way the front of the crank case did.

    Stoked!
  • Suosa
    Offline / Send Message
    Suosa polycounter lvl 4
    Hi guys, 

    First time posting here after lurking for a while! 
    I've been looking a round to see if I could find a case that is similar to mine but they all look sort of similar but a lot more complex and I just keep getting lost trying to get my head around it, being frustrated in finding a solution is not helping too..

    Basically I'm modelling a low polly stylized axe for a personal project of mine. I've finished the low poly axe but I now want to add supporting edge loops so I can bring it into zbrush to add the missing details and stuff.

    However, I decided to add some chips to the axe's blade (increasing the verts but very minimaly) so it would give it a bit more character. 

    The problem is I cannot, for the life of me, wrap my head around adding supporting edge loops so that I can have a charp blade and chipped parts. Here are some screenshots to give you an idea of what I am trying to say.

    Low poly version with chipped blade


    My main issue now is how do I keep the blade sharp and incorportate support edge loops around the chip so I can subdivide it properly when bringing it into Zbrush for details.


    Ive tried beveling the blade so I would look sharp but since the bevel goes through the chipped part then it sort of bleeds into it which is not ideal 




    What it looks like



    After this I've just been slowly losing my sanity trying to find ways to make it work. I've tried beveling and adding edge loops manually but it all feels like an unnaturally lonmg procedure.. or maybe I just need a reality check! 



    At this point I end up with ngons and having to merge vertices and losing edge flow (which I understand I don't need since I'm going to bake the details onto the low poly). The final result isnt that bad but I have 3 other chipped bits and the amount of manual corrections / edges I've had to make seems counterproductive at this point...

    I know this is painful to watch and apologise in advance.

    I'm just running out of patience and motivation to keep working on this as I've been stuck here for many hours.
    I would greatly appreciate any pointers on how to go about when it comes to details like these.

    Thanks!
  • IronLover64
    Offline / Send Message
    IronLover64 polycounter lvl 2
    How do I model the disk part as well as the screens?

    Since this mesh has an odd number of screens, I had to make a 14 sided cylinder in blender, keep two of the 14 faces, extrude the screens, and duplicate and rotate the faces 7 times. I want to add a bevel to the mesh because importing this mesh to 3D Coat for sculpting additional detail automatically makes the edges smooth. The topology isn't fit for beveling. How else should I model this?

  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @Suosa It looks like you're on the right track. If the mesh subdivides cleanly, without any major smoothing artifacts, then it's generally acceptable to use n-gons and triangles in hard surface subdivision models.

    Since this model is going through a sculpting pass it may make more sense to try and add these minor surface defects in ZBrush. A boolean re-meshing and detail sculpting workflow may be more efficient than trying to perfect all of these small details on the subdivision base mesh for the high poly sculpt. If including the larger surface defects in the subdivision mesh is a stylistic choice then just route the support loops around the major shapes and try to constrain any n-gons or triangles to the flat areas.

    Extremely narrow support loops tend to produce a tight edge highlight and although it can look good up close it may be difficult to read from regular view distances. Over sharpened edges can also cause baking issues so consider using a slightly wider and more consistent edge width across the entire shape.

    Below is an example of what this could look like: block out the basic shape, add the larger surface defects and route major support geometry as necessary. Add support loops (highlighted) around the edges that control the major shapes using a bevel / chamfer operation.

    It should be possible to control the bevel / chamfer placement and behavior with groups or weights. Review the documentation for your application to find the correct settings combination that allows you to reliably place support loops with automated tools. Try to avoid having to manually place or adjust support loops.



    @IronLover64 Overall it looks like you have the right idea. Try to use a little less geometry when blocking out the shapes and use inset operations when creating the screens and desk spaces so there's room around the shapes to add bevels. Also try changing the bevel geometry settings from Sharp Miter Outer to Arc Miter outer.

    Here's an example of what this process could look like.

    Start with a seven segment circle and multiply the segment count using 7*N until there's sufficient geometry. Use a series of inset operations to create the truncated cone in the center of the table. Select a section of the truncated cone's wall and use an inset operation to create the screen's outline. Run another inset operation to create the depth and use the loop cut tool to add support geometry through the screen. Repeat this process to create the desktop on the table below the screen. Split the table into a single 1/7th section. Add bevel weight to the highlighted edges and add a bevel modifier. Merge in any additional details. Spin duplicate or object offset array duplicate the 1/7th section and merge vertices by distance. Add a subdivision modifier.

  • BRTT
    Hello! I have topology issues in some places. I've tried many options. The main thing is to check on a glossy surface. Show how to do it right please!
  • IronLover64
    Offline / Send Message
    IronLover64 polycounter lvl 2
    @Suosa It looks like you're on the right track. If the mesh subdivides cleanly, without any major smoothing artifacts, then it's generally acceptable to use n-gons and triangles in hard surface subdivision models.

    Since this model is going through a sculpting pass it may make more sense to try and add these minor surface defects in ZBrush. A boolean re-meshing and detail sculpting workflow may be more efficient than trying to perfect all of these small details on the subdivision base mesh for the high poly sculpt. If including the larger surface defects in the subdivision mesh is a stylistic choice then just route the support loops around the major shapes and try to constrain any n-gons or triangles to the flat areas.

    Extremely narrow support loops tend to produce a tight edge highlight and although it can look good up close it may be difficult to read from regular view distances. Over sharpened edges can also cause baking issues so consider using a slightly wider and more consistent edge width across the entire shape.

    Below is an example of what this could look like: block out the basic shape, add the larger surface defects and route major support geometry as necessary. Add support loops (highlighted) around the edges that control the major shapes using a bevel / chamfer operation.

    It should be possible to control the bevel / chamfer placement and behavior with groups or weights. Review the documentation for your application to find the correct settings combination that allows you to reliably place support loops with automated tools. Try to avoid having to manually place or adjust support loops.



    @IronLover64 Overall it looks like you have the right idea. Try to use a little less geometry when blocking out the shapes and use inset operations when creating the screens and desk spaces so there's room around the shapes to add bevels. Also try changing the bevel geometry settings from Sharp Miter Outer to Arc Miter outer.

    Here's an example of what this process could look like.

    Start with a seven segment circle and multiply the segment count using 7*N until there's sufficient geometry. Use a series of inset operations to create the truncated cone in the center of the table. Select a section of the truncated cone's wall and use an inset operation to create the screen's outline. Run another inset operation to create the depth and use the loop cut tool to add support geometry through the screen. Repeat this process to create the desktop on the table below the screen. Split the table into a single 1/7th section. Add bevel weight to the highlighted edges and add a bevel modifier. Merge in any additional details. Spin duplicate or object offset array duplicate the 1/7th section and merge vertices by distance. Add a subdivision modifier.

    thanks a lot!
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @BRTT Welcome to Polycount. Consider checking out the forum information and introduction thread.

    Some of the contributing factors that could be causing those subdivision smoothing artifacts are:

    • Shape inconsistencies and undulations in the base mesh.
    • Insufficient supporting geometry around shape intersections and cut outs.
    • Support loop routing and topology flow issues.

    A couple of pages back there's a discussion about routing support loops around cut outs on curved shapes and there's also been a few recent discussions about cylinders intersecting flat surfaces with rounded edges and shape intersections on curved surfaces. The information in these posts should be helpful for resolving the smoothing issues on that shape.

    The basic idea is to use the existing geometry as support for shape transitions and to place intersecting geometry between existing edge segments. This helps the curved geometry retain it's shape, provides support for adjacent shape intersections and leaves a place for additional support loops around the shapes. Additional information can be found in the links above and there's a lot of other great examples in this thread. It's definitely worth skimming through and looking for other solutions to similar shapes.

    There's a few different ways to approach modeling these shapes but here's a basic overview of what the modeling process and topology could look like.


  • Filip5
    Offline / Send Message
    Filip5 polycounter lvl 8
    Hello guys, I would like to ask how to model this mattress. 

    I have tried to do a highpoly of it but it looks so bad...



    Is there a easier way of doing this ?


Sign In or Register to comment.