Home Technical Talk

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

Replies

  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @AntBay If the manually placed edge loops need to wrap around the bottom of the shapes then the topology structure of the base mesh will need to be changed so it redirects the edge loop around all of the corners and down along the bottom edge. Below is an example of what this process could look like.


     
    Here's a simplified comparison of the two topology layouts. It's also worth looking at alternate ways of adding support loops and round over features with different tools. The last column in this example shows that a bevel / chamfer operation will create the same topology on both layouts and it also provides a more consistent result than manually placing individual support loops.


    The topology layout and modeling process can be further simplified by blocking out the basic shapes, rounding the corners with a bevel / chamfer operation, cutting in coarse support geometry and generating the final support loops with a bevel / chamfer modifier.



    This modeling strategy also works if the bottom of the shape needs a specific profile. Just add the shape profile and let the bevel / chamfer modifier handle the rest of the support loops on the sharper areas.



    Recap:
    Adjust the topology flow at the most basic level (during the block out) and add the complex surface details onto that. Look for alternate tools and modeling strategies that will reduce the amount of time spent manually placing support loops while also increasing the overall consistency of the support loop edge width.

  • AntBay
    @FrankPolygon Wow, thanks for the amazing response. The images are greatly appreciated. I will experiment and try out the different methods.
    Thanks once again!
  • MegaSofteae
    Hello everyone, I hadn't noticed this discussion post until now! I'm struggling with 3d modeling in general, I'm not sure where I should be improving. I wanna model but when I'm on it, there are a lot of questions that come  into my mind. I do experiment, but there are more especific questions such as: Once i've added the support loops, is it safe to unwrap the model, I feel like the wireframe is a mess. Support loops that interfere with smoothing, making areas which should be smooth, really sharp.

    I wanna model a hori hori knife, but I think that's too much complexity for now. For now I'm trying to model a knife

    Reference:



    For now I'm focusing in the blade. it will go inside the handle so I don't have to worry about the handle part yet. 



    I'm having trouble with the top part, since the top part has holes these loops wil have to end somewhere, so I'm thinking in putting them in the blade part, but these will intervene with the sharp subdivision that the blade needs. This is before using a shell modifier. Subdividing this, I think it would interfere with the sharpness of the blade. I might have done this wrong, how should I approach this?





    Model Obj download

    Thanks in advance! Cheers!
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @MegaSofteae Creating a detailed outline of an object's profile can seem like a logical place to start but in most cases this limits the way you think about the shapes by constraining it to a 2D space. This is why it's generally considered best practice to block out all of the major features of an object in 3D before divining in and adding all of the support loops and minor details.

    Observation is a significant portion of many different art processes and learning to recognize and replicate the basic shapes that make up the prominent features of an object is a skill that takes time to build. After gathering reference images, begin by studying the object's shape and try to organize them into distinct groups based on how they interact with each other.

    Below is an example that highlights the three main shapes. The surfaces highlighted in red are completely flat, with an even thickness, that makes up the bulk of the shape. Blue is a uniform beveled surface that follows the curve of the blade's profile. Yellow is a partially beveled surface that follows the compound curves along the top profile of the blade. Most of this area has a uniform bevel but the shape is interrupted by serrations and it tapers off near the tang where the grinding tool rolled out and off of the part.



    The previous observations and analysis will help guide the modeling strategy, order of operations and rough proportions used to develop the primary shapes of the model. Blocking out the model is done to establish the primary forms, proportions and topology flow.

    There's a significant amount of interplay between these elements so they all effect one another and changes to one will generally require changes to the others. This is why it's important to keep things reasonably simple and fairly accurate.

    Using the minimum amount of geometry to effectively hold the shapes, while providing support for secondary details, means it will be easier to adjust the shapes and proportions of the mesh without having to constantly make manual adjustments to the support loops and fine details.

    There's a number of different ways to approach modeling the block out and which approach makes the most sense depends on what tools are available and what the model will be used for. Here's one example of what this process could look like for a base mesh that will be used to develop a high poly model and a low poly model for a game asset.

    Start by outlining the major forms according to the shape analysis. [In this example the basic shape profiles were established using curves because they provided an efficient and infinitely adjustable way to generate mathematically precise geometry with consistent spacing. The geometry density of the loops (around each individual surface profile) was adjusted until the number of segments was equal.]

    Convert the curves to mesh and use automated fill operations to generate the geometry connecting the matching edge loops that define each surface.

    Other methods could be used to develop the basic shapes but manual vertex extrusion processes tend to be comparatively slower, imprecise and inconsistent. An alternate option that would be comparatively quick and accurate would be to start with a solid, coarse topology flow block out and use other modeling tools to generate the curves along the shape's profile.

    Add depth to the object by extruding the outer edges of the shape's profile and setup a mirror modifier to reduce the amount of manual editing required. After these few basic steps the primary shape, proportion and topology of the object should be ready for additional details and support loops.

    Use a bevel / chamfer modifier with edge weights and grouping to automatically generate support loops that are infinitely adjustable. Add a subdivision modifier and preview the mesh. If the proportions and topology are good then there shouldn't be any major smoothing artifacts and it's time to add the smaller secondary details.

    Planning ahead for the amount of geometry required to support these additional details can save a lot of headache so be mindful of topology routing and geometry density early in the block out. Simplify areas that have excessive geometry and excess edge loops and increase the mesh density in areas that need to support finer details.

    In this example the back portion of the top bevel profile needed some additional geometry to support the serrations. The geometry was properly routed during the block out so increasing the geometry density here was as simple as selecting the outer edge loops around the area and smooth subdividing the geometry segments.

    Sometimes it's possible to manually cut in additional loops but doing this on compound curves and other complex surfaces will often disturb the segment spacing and produce undesirable smoothing behavior.

    It is possible to manually adjust the positioning of support geometry to compensate for smoothing artifacts but this often requires a significant time investment and can produce undesirable smoothing behavior elsewhere. This is why it's important to maintain a relatively consistent geometry density along curves during the block out phase and to rely on automated tools that generate mathematically accurate geometry.

    Once the serrated details are added to the shape the edge weights and grouping that control the bevel / chamfer modifier are adjusted and the subdivision mesh is complete. Since all of the modifiers are still active the underlying base mesh is kept fairly simple which makes it easy to edit and basically provides a free low poly mesh that can be used with minimal adjustments. The bottom row shows the base mesh underneath the modifiers, the subdivision cage mesh generated by the base mesh + modifiers and the shaded subdivision preview.


    When properly supported, flat surfaces are generally very tolerant to substantial topology changes so they can be a good place to simplify the mesh by culling unnecessary edge loop propagation. Triangles and n-gons are also fine, so long as they aren't causing any major smoothing errors. Subdivision modeling is an approximate process so it's all about balancing shape accuracy with process efficiency.



    There's a lot of different modeling strategies and different ways to approach the topology layout. What makes the most sense often depends on the project's technical requirements and weighing the results versus the amount of time and effort required. It's important to strive for a high quality result but it's also important to learn where to put the quality because it doesn't make sense to spend a lot of time on something that doesn't really impact the quality of the results or waste time on things players won't ever see up close.

    This thread has a lot of great examples of how other artists have solved shape and topology issues so it's definitely worth the time to skim through, look at the images for similar shapes and read up on how / why specific things work or don't work. As an example: a few pages back Sacboi posted a list of posts that cover modeling grip wrappings which should be helpful when it comes to replicating the grip in the reference image.

    Recap:
    Study the reference images to identify and group shapes according to how they interact with each other. Use this shape analysis to block out the basic shapes and establish the topology flow early on in the modeling process. Add details and support loops later and if possible use modifiers to generate support loops and reduce the manual workload. Avoid unnecessary mesh complexity and cull unnecessary edge loops to avoid smoothing issues caused by edge loop propagation. Continue to research, practice and evaluate different topology layout and modeling strategies.
  • MegaSofteae
    @FrankPolygon
    Hello! I wanted to thank you! The careful explanation is really appreciated! There are a lot of things I didn't take into consideration and was confused!

    Once I'm done with the blade, I will check the grip!
    Thank you! :D
  • rogi92
    Offline / Send Message
    rogi92 polycounter lvl 3
    Hello everyone,
    excuse me if there's already something like this.
    How is the best method do model this piece?
    Thank you masters : )
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @rogi92 Which approach is best will often depend on the specifics of a project's technical requirements. There's a lot of ways to approach modeling this part but there's also a lot of different factors that can influence modeling strategy decisions. Broad, open ended questions lack the context required to provide meaningful feedback. As Wirrexx has already mentioned: it's important to show the results of the modeling attempt and explain what may have gone wrong or what areas need to be improved.

    A good place to start would be to analyze the reference image, break the part down into basic shapes and come up with a couple of topology routing strategies to connect the individual shapes. Once the form of the object is understood: it's generally considered best practice to start the modeling process by blocking out the primary shapes (matching geometry segments whenever possible) before adding minor surface details and support loops. The modeling process itself doesn't always require a set order of operations so it just comes down to learning how to use the tools and balancing shape accuracy with process efficiency.

    Below is an example of what a basic modeling process and topology layout could look like. All basic inset and extrude operations with support loops generated by a bevel / chamfer modifier. The topology layout in this example has separate loop paths around each through hole feature.

    This is just one example and there's a number of different ways to approach modeling the part in question. Whether or not one is better than another is something that can't really be answered without asking more specific questions or exploring different modeling and topology strategies and comparing the results.



    Here's an alternate topology layout with a single loop path around the outside of the part. Comparing the topology layouts of each base mesh: the one above provides a clear path around each through hole but doesn't provide a clean path around the perimeter of the combined shape and it doesn't provide a clear path across the join between the large cylinder and the small tab. (This minor issue could be solved by splitting the mesh across the middle of the tab and filling the gap.) The one below provides a clear path around the perimeter of the combined shapes and across the join between the large cylinder and the tab but doesn't provide a clean path around the through holes.

    Which layout makes more sense depends on which areas need to be sharpened and what tools are used to generate the support loops. Bevel / chamfer and inset operations can add support loops around both of these topology layouts without running into any major issues but individual loop cuts could run into potential routing problems that may require manual cleanup.



    There's also the issue of optimal geometry density: it's possible to reduce the number of segments in the larger cylinder to simplify the mesh but that would also decrease the overall shape accuracy and require more creative topology routing to connect the two shapes. Since the top surface is mostly flat this shouldn't cause any major smoothing issues but it would disrupt the edge flow around the shapes. The disruption of the edge flow (in the base mesh) isn't a major issue if the support loops are added with a bevel / chamfer operation but it could cause issues if the support loops needed to be added with other tools.

    So whether or not one is better than the other depends entirely on what factors are at play. It all comes down to project goals, resource constraints, technical constraints and personal preference.

    It's also important to invest in self development so it's definitely worth the time to research how other artists have solved similar problems, apply this information by experimenting with different topology strategies and asking for feedback on specific issues encountered along the way.
  • rogi92
    Offline / Send Message
    rogi92 polycounter lvl 3
    @rogi92 Which approach is best will often depend on the specifics of a project's technical requirements. There's a lot of ways to approach modeling this part but there's also a lot of different factors that can influence modeling strategy decisions. Broad, open ended questions lack the context required to provide meaningful feedback. As Wirrexx has already mentioned: it's important to show the results of the modeling attempt and explain what may have gone wrong or what areas need to be improved.

    A good place to start would be to analyze the reference image, break the part down into basic shapes and come up with a couple of topology routing strategies to connect the individual shapes. Once the form of the object is understood: it's generally considered best practice to start the modeling process by blocking out the primary shapes (matching geometry segments whenever possible) before adding minor surface details and support loops. The modeling process itself doesn't always require a set order of operations so it just comes down to learning how to use the tools and balancing shape accuracy with process efficiency.

    Below is an example of what a basic modeling process and topology layout could look like. All basic inset and extrude operations with support loops generated by a bevel / chamfer modifier. The topology layout in this example has separate loop paths around each through hole feature.

    This is just one example and there's a number of different ways to approach modeling the part in question. Whether or not one is better than another is something that can't really be answered without asking more specific questions or exploring different modeling and topology strategies and comparing the results.



    Here's an alternate topology layout with a single loop path around the outside of the part. Comparing the topology layouts of each base mesh: the one above provides a clear path around each through hole but doesn't provide a clean path around the perimeter of the combined shape and it doesn't provide a clear path across the join between the large cylinder and the small tab. (This minor issue could be solved by splitting the mesh across the middle of the tab and filling the gap.) The one below provides a clear path around the perimeter of the combined shapes and across the join between the large cylinder and the tab but doesn't provide a clean path around the through holes.

    Which layout makes more sense depends on which areas need to be sharpened and what tools are used to generate the support loops. Bevel / chamfer and inset operations can add support loops around both of these topology layouts without running into any major issues but individual loop cuts could run into potential routing problems that may require manual cleanup.



    There's also the issue of optimal geometry density: it's possible to reduce the number of segments in the larger cylinder to simplify the mesh but that would also decrease the overall shape accuracy and require more creative topology routing to connect the two shapes. Since the top surface is mostly flat this shouldn't cause any major smoothing issues but it would disrupt the edge flow around the shapes. The disruption of the edge flow (in the base mesh) isn't a major issue if the support loops are added with a bevel / chamfer operation but it could cause issues if the support loops needed to be added with other tools.

    So whether or not one is better than the other depends entirely on what factors are at play. It all comes down to project goals, resource constraints, technical constraints and personal preference.

    It's also important to invest in self development so it's definitely worth the time to research how other artists have solved similar problems, apply this information by experimenting with different topology strategies and asking for feedback on specific issues encountered along the way.
    Thank you for your help and the information, I really appreciate it.
    It help me a lot.
    I will keep those information in mind.
    Again many thanks.
    Rob :)
  • JLmussi
    I thought this workflow might help some of you.  This is one of my favorite workflows for creating certain hard surface parts is using a little workflow called "Pattern Creation & Deformers"

    Let’s model the magazine of a machine gun with the following workflow:

    1) Break down the model into a repetable pattern, so the bulk of the work is done by symmetry.

    2) Create the model straight and use the help of a deformer to give it is final shape.  The bend deformer is from Maya but most 3d packages have a similar deformer and settings.

    Using symmetry will speed up the process and make sure our model is even. Using a deformer will allow us to create the bulk of our model with the ease of working in world space and simply just bend it into place towards the end of the modeling process. This workflow can be implemented to breakdown and create many complex hard surface models with relative ease.

    I created a free pdf guide with more workflows like this which can be downloaded here. 
    https://jlmussi.com/hsguide/


  • Daf57
    Offline / Send Message
    Daf57 polycounter lvl 9
    Hi,
    Is there any way to get a straight, even, consistent edge loop from this mesh? You can see the topo in the first image - the 2nd image is after removing the unwanted faces. It's a predictable outcome  - just wondering if there's a way to create a usable loop for extruding straight back (like cylinder). If not I guess I'll have to retopo, right?
    Thanks!
    Daf


  • Iaido
    Offline / Send Message
    Iaido polycounter lvl 11
    Daf57 said:
    Hi,
    Is there any way to get a straight, even, consistent edge loop from this mesh? You can see the topo in the first image - the 2nd image is after removing the unwanted faces. It's a predictable outcome  - just wondering if there's a way to create a usable loop for extruding straight back (like cylinder). If not I guess I'll have to retopo, right?
    Thanks!
    Daf


    Most tools have a knife tool. I would go into your side view and cut a line from the top of the head to the bottom. It should cut through, this will give you a straight edge loops, but be careful as this will also give you a lot of ngons and triangles. 
  • Daf57
    Offline / Send Message
    Daf57 polycounter lvl 9
    Hey thanks, Iaido! I did try it with the knife and it just made a big mess - ngons and tris like you said. I'll just do the retopo - it will be better anyway. :)
    Thanks again!
    Daf

  • aregvan
    Offline / Send Message
    aregvan polycounter lvl 7
    Hi all,
    I cannot for the life of me figure out how to have a perfect round silhouette and quads for subdividing, at the same time. Is there a simple way of doing this?


  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    aregvan said:
    Hi all,
    I cannot for the life of me figure out how to have a perfect round silhouette and quads for subdividing, at the same time. Is there a simple way of doing this?


    You could start with a 32 sided cylinder. Or model it flat and bend it with a modifier. Last tips is to use a proxy mesh and skinwrap  your object to the proxy mesh
  • Blaizer
    Offline / Send Message
    Blaizer interpolator
    @aregvan For holes in cylinders the rule is quite simple. The bigger the hole, the less polygons you need. For smaller details, you just need more geometry, that's all.



    And a with a cylinder of 16 sides, bigger hole:


  • aregvan
    Offline / Send Message
    aregvan polycounter lvl 7
    @wirrexx 
    @Blaizer

    Thanks for the help guys! Seems pentagons aren't so bad afterall.
  • AhmedAtheek
    hey,
    I'm trying to create this type of cut on a cylinder but I cant figure out how to get rid of the wierd pinching effect I'm getting and retain the same curvature of the cylinder.



  • Blaizer
    Offline / Send Message
    Blaizer interpolator
    @AhmedAtheek to get rid of pinching artifacts, you just need to use more geometry, like i pointed out in my previous reply to aregvan.
  • LoneRanger
    Offline / Send Message
    LoneRanger polycounter lvl 4
    How to connect it properly in my HP model? I've tried various ways but there is always some shading problems (surface is not smooth, there are little bumps etc.). When I add more horizontal edges to those vertices inside this big n-gon my cylinder is no round anymore.

  • sacboi
    Offline / Send Message
    sacboi high dynamic range
    @LoneRanger
    Check out rule of thumb via intrusion shape cuts for curved surfaces, 4 posts above yours by @Blaizer - i.e. holes in cylinders.
  • Sims_doc
    Offline / Send Message
    Sims_doc polycounter lvl 7
    Hi, Folks! I am working on a "OTs-14 Groza" hybrid and I'm currently polishing the carry handle but there is a persistent problem that is hampering me. This is an example, I've WIP up that is what effective highlights the issue that i am having.


    The area that I've created next to the main rail is how, I think it looks on the reference itself however many of the other references that I've seen seem to lack this nook, which is the part that isn't rounded is what i've got so far. I think they may be airsoft.

    I'd just like to get someones opinion on how i should approach modeling this nook
  • sacboi
    Offline / Send Message
    sacboi high dynamic range
    Most OEM versions I had inspected online, seem to be manufactured with that particular notch which most likely is for additional attachments eg: telescopic or grenade launcher iron sights:



    Now, assuming you're using a Proboolean / Boolean + Dynamesh workflow, I'll suggest more segments too define that shape, then try again?

    EDIT:
    Also check out why the technique's author recommends high segment counts for both cylinders and curved area's, here.
  • Sims_doc
    Offline / Send Message
    Sims_doc polycounter lvl 7
    Hi @sacboi I do have a more define mesh as subdivision modeling. The example posted from earlier is effectively just an example and is just a simplified cutout of that shape. Regardless of the boolean operation itself. I believe this may give me what i'm looking for in-terms of desired outcome if i boolon it.
    I think that should do it, Just need some confirmation that the shape is indeed accurate to the reference in the sense that covers the area marked in red.


  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @Sims_doc The rail profile in those reference images has two distinct shapes: flat walls with radiused corners and tabs with angled surfaces on the underside where the mount can grip and index (center) onto the rail. That triangular notch is probably just a manufacturing artifact left behind by the dovetail cutter used to create the angled features on the underside of the mounting tabs.

    The real part could have been made without creating those triangular shapes but that would have required at least two additional setups that would have needed complex fixtures to hold the parts in the correct orientation. Every time a part is taken out of a machine and re-oriented it wastes time and creates an opportunity for  the operator to make significant mistakes. It's very likely it was made the way it was so all of the milling operations on the rail could be done in one setup.

    There is the vague possibility that there was some kind of planned accessory that would index off those triangular corner features but sometimes those odd little shapes that don't have a clear functional purpose aren't anything more than artifacts of the tooling choices made to simplify the manufacturing process.

    Sometimes it can be helpful to draw over the reference images to highlight the major shapes that make up more complex features. A good place to start is to identify the individual surface planes of each feature then figure out how they interact with and blend into the adjacent shapes.

    Looking at the reference images for this shape: the angled features appear to extend all the way down into the bottom corner where the flat walls and the flat bottom meets. The triangular shape at the front of the rail is radiused into the rounded corner because that's where the dovetail cutter rolled out of the slot created by the end mill.


  • Sims_doc
    Offline / Send Message
    Sims_doc polycounter lvl 7
    Thank you frank for breaking down that shape, I'd like to know what type of cone you used to get that shape as been trying to replicate it by myself before asking. With a boolean operation, I can get a similar result but nothing remotely close. This is what, I think your using as it's the close to a dovetail.

  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    @Sims_doc What you have is pretty close. Shapes in some of the reference images can be a little tricky to interpret because a lot of the corners were broken over so the edges of the real item aren't razor sharp. It's probably better to use the actual distances between the existing mesh features to figure out the size and angle of the boolean objects.

    As an example: the angle of the dovetail cut is equal to the distance from the top edge of the tab to the inside corner of the slot and the radius of the truncated cone is pretty much equal to the distance from the perimeter of the circle to the outside edge of the part.



    So start by establishing an angle that meets the top and bottom of the dovetails then line up the radiused corners of the boolean shape so they match the center of the cylindrical cut outs. This should create a shape that meets the point of the dovetails but leaves behind a little extra space above the triangular artifact in the cylindrical slot.
  • Sims_doc
    Offline / Send Message
    Sims_doc polycounter lvl 7
    Hi, Frank. Do you mind sharing that file with me? With some progress steps starting from nothing to where you are now as i'd really like to understand the progress behind it and replicate those results.

  • aregvan
    Offline / Send Message
    aregvan polycounter lvl 7
    Hi all. I am modeling an MP7 and cannot figure out how to move on from this part on the magazine.
    It is the bottom part where another piece attaches on the lip.




    Unable to move on from here.
    I also tried doing a boolean and building off the new faces, but that looked very bad in terms of polygon flow.

    Any help is appreciated!
  • wirrexx
    Offline / Send Message
    wirrexx ngon master
    aregvan said:
    Hi all. I am modeling an MP7 and cannot figure out how to move on from this part on the magazine.
    It is the bottom part where another piece attaches on the lip.




    Unable to move on from here.
    I also tried doing a boolean and building off the new faces, but that looked very bad in terms of polygon flow.

    Any help is appreciated!

    The issue is, that you're modeling two pieces that are separated as one piece.
    The bottom part is actually another mesh that's attached to it.
  • Sims_doc
    Offline / Send Message
    Sims_doc polycounter lvl 7
    @aregvan Hi, That doesn't look too hard to actually model, that teal lines i've added in are really all you'd need to do to create the shape that attaches to the magazine and from there you can just separate it and you've got yourself your two shapes. Alternatively you can just model it as one shape and extrude for the bits that need extruding but it will be a bit more complex to work with.
  • aregvan
    Offline / Send Message
    aregvan polycounter lvl 7
    Sims_doc said:
    @aregvan Hi, That doesn't look too hard to actually model, that teal lines i've added in are really all you'd need to do to create the shape that attaches to the magazine and from there you can just separate it and you've got yourself your two shapes. Alternatively you can just model it as one shape and extrude for the bits that need extruding but it will be a bit more complex to work with.

    Hello.
    It seems natural to do this at first, but the issue is the front corner is beveled. If this was a straight corner, it would be simple to cut through and have two perfect separations.
    As the front corner bevels, I do not see how to make this cut even all the way through.
  • IronLover64
    Offline / Send Message
    IronLover64 polycounter lvl 2
    I'm trying to figure out how to bevel this cross section. Using the bevel itself creates subtle artifacts and the corners look weird. A massive number of other cross sections also have this problem. How do I fix this?
  • IronLover64
    Offline / Send Message
    IronLover64 polycounter lvl 2

    I also want to ask if anybody knows how to fix the artifacts shown on image 1. Because many of the looks are necessary for the front, I can't take all of them away (I tried). Does anyone know how to fix it? Thanks
  • Blaizer
    Offline / Send Message
    Blaizer interpolator
    @IronLover64 I would do this:



    Make sure the "pole" points are not in the edge bevel. btw, in a subdiv model you would need less details. Here's an example using pentagon topology in order to simplify the mesh and save effort (different bevel radius). Using bevels with 3 sides and support loops we can obtain the same result. 


    As you can see, the mesh is much cleaner and simpler. The third image is the model subdivided x1, pay attention to the pole location and topology.


    Triangles are also very useful.

    For the other issue, i think you just need to add more "cuts" and a few edge clean-ups. That type of n-gons in curved shapes, and more in those zones, are not very friendly. Avoid "edge propagation".

    hope it helps!
  • PancakeMSTR
    Offline / Send Message
    PancakeMSTR polycounter lvl 4
    Looking for a topology review. Would like to know what others might do differently : 

    No problem providing some more angles if anyone interested. 
  • franman
    Offline / Send Message
    franman polycounter lvl 13
    In Blender I have UV unwrapped a Sphere (which was originated from a cube) after defining my segments; however, in my UV editor the UV islands are orientated at an angle.

    How do I exactly align or orientate (w/o visually eyeballing it) those UV islands so they are no longer appear tilted?
  • IronLover64
    Offline / Send Message
    IronLover64 polycounter lvl 2
    Blaizer said:
    @IronLover64 I would do this:



    Make sure the "pole" points are not in the edge bevel. btw, in a subdiv model you would need less details. Here's an example using pentagon topology in order to simplify the mesh and save effort (different bevel radius). Using bevels with 3 sides and support loops we can obtain the same result. 


    As you can see, the mesh is much cleaner and simpler. The third image is the model subdivided x1, pay attention to the pole location and topology.


    Triangles are also very useful.

    For the other issue, i think you just need to add more "cuts" and a few edge clean-ups. That type of n-gons in curved shapes, and more in those zones, are not very friendly. Avoid "edge propagation".

    hope it helps!
    Thanks a lot!
  • IronLover64
    Offline / Send Message
    IronLover64 polycounter lvl 2
    Blaizer said:
    @IronLover64 I would do this:



    Make sure the "pole" points are not in the edge bevel. btw, in a subdiv model you would need less details. Here's an example using pentagon topology in order to simplify the mesh and save effort (different bevel radius). Using bevels with 3 sides and support loops we can obtain the same result. 


    As you can see, the mesh is much cleaner and simpler. The third image is the model subdivided x1, pay attention to the pole location and topology.


    Triangles are also very useful.

    For the other issue, i think you just need to add more "cuts" and a few edge clean-ups. That type of n-gons in curved shapes, and more in those zones, are not very friendly. Avoid "edge propagation".

    hope it helps!
    Do I have to edit each cross one vert at a time? The bevel modifier doesn't do that kind of corner
  • Blaizer
    Offline / Send Message
    Blaizer interpolator
    @IronLover64 Blender is quite basic making bevels, and has too many flaws, it can't do "extra mitering" like in modo for example. So you will need to work in one part of the mesh, and work with tiles and "arrays" in blender.


    With modo all this is quite simple an easy, but we still need to work on the topology.


    If you are doing it manually, beware because in this example, the bevel tool in edit mode, produces a bad result. Keep using the modifier, apply or "collapse", and later edit the tile mesh.


    Using the bevel modifier all is done quite fast. The rest is just an inset, like shown in the image. The bad thing is that we need to edit a lot, "disolving" edges and making some cuts. So tiles are the way to go + "arrays" + bridges.

    This is one of my main gripes with Blender... i'm heavily customed to do things in a faster way. 
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    To add to what's already been said: try to simplify the mesh by dissolving and merging some of the extra geometry. Straightening and rerouting the remaining loops into the existing geometry on the back strap should solve most of the smoothing issues. Flat surfaces that are properly supported are relatively impervious to abrupt changes in topology so it should be possible to delete a lot of those extra loops that are interfering with the curved shape transition.

    Another thing to consider is that angled surfaces that are flat are still flat surfaces. Ending some of these edge loops with n-gons shouldn't cause any major smoothing issues as long as the n-gons are away from the curved shape transition and further up on the angled flat. Below is an example of what this process could look like.

    Having a lot of loops that run off into adjacent shapes tends to generate a significant amount of extra geometry that can make the mesh difficult to work with. Blocking out the shapes in separate parts (to match the seams in the real object) and figuring out the topology flow on a simple base mesh should enable the use of bevel / chamfer modifiers to add most of the support loops. This approach should help simplify future projects and reduce the amount of time spent manually adding or editing support loops.



    If a mesh subdivides without any major smoothing artifacts then it's generally considered passable. A shiny matcap with a soft highlight roll-off is helpful for spotting subtle smoothing artifacts and surface undulations. It might be worth using this type of shader to check the sides of the shapes at glancing angles to see if there's any issues hidden by the current material settings.

    A bit more context on what the object is and what the model will be used for would also be helpful for those giving feedback. Are there any specific problem areas that stick out or other things you feel could be improved?

    @franman
    There's a few different way to approach this problem but a lot will depend on the current seam placement and unwrap. If some of the more popular UV tool addons are installed then check the documentation to figure out what tools or unwrap constraints would help keep the UV islands in the desired orientation. If it's just stock Blender then the simplest approach would be to unwrap the cube before turning it into a sphere. It's also possible to use cube unwrapping or some other projection method to start the unwrap in the correct orientation then use pin and unwrap to generate the rest of the UV islands but this is a bit more work than some of the other methods.

    The bevel topology issues on the grid pattern are a symptom of the problem and not the problem itself. This is caused by issues with the order of operations and the modifier settings. Blender's defualt bevel modifier settings apply the bevel operation to the entire mesh using sharp outside corners. It's these defualt settings that are generating the undesired topology.

    The limit method setting is used to control where the bevel operation is allowed to generate new geometry. By defualt this is set to none which means there's no limit to where the bevel operation can place new geometry. Changing the limit method to angle will only allow the bevel modifier to place new geometry on edges at or below the angle limit threshold. There's also other limit methods that can be used to control multiple bevel modifiers at the same time by using different geometry components.

    The miter outer setting is used to control the pattern used to fill in new geometry around the outside corners. By defualt this is set to sharp which tries to keep the outside corners as square as possible and this can generate less than ideal corner topology. Changing the miter outside to arc will allow the bevel modifier to add new support geometry around the outside of the corners.

    Below is an example that shows the starting mesh and compares the defualt settings to the setting recommended above.



    With the correct settings there shouldn't be any real need to apply the bevel modifier and manually adjust the topology of such a simple shape. Here's an example of the same modifier settings (limit angle 30° and miter outer arc) used in conjunction with simple modeling operations (subdivide, inset individual, extrude) to generate the desired topology in four steps.



    Manual editing is something that should be minimized in a contemporary subdivision modeling workflow. Some tool sets take different approaches and some applications may be more limited than others. Sometimes these limitations do make certain tools more sensitive to order of operations but that doesn't mean they can't be used efficiently.

    Every application is different so it's important to review the documentation and figure out what all the different settings do and how different tools can be combined to achieve the desired result in the most efficient way.

    Here's an example of how order of operations and optimizing the mesh for the bevel / chamfer modifier workflow creates a workable result that doesn't require any manual editing. A basic cube is inserted and subdivided then the outer corner is snap merged before the limited dissolve removes all unwanted geometry. The inside corner can be manually beveled without generating any unnecessary geometry then the bevel and subdivision modifiers are turned on.

    Order of operations are important for both the modeling and the modifier stack. This example has two bevel modifiers: the first adds the support loop and the second adds the rounded edge profile. Both remain live and editable throughout the entire modeling process and the subdivision modifier can be toggled on and off as well. Any changes to the base mesh will propagate through the stack and if everything is done correctly the support loops won't require any manual adjustments or editing.


    Going back in this thread a few years there's been several compelling arguments made for working on base mesh block outs with the bevel / chamfer modifier enabled. A few of those reasons are: it provides immediate feedback on how topology changes impact support loops, it can help guide decisions about detail scale and shape readability, it makes changing the base mesh easier since there's no need to fight with all the manually placed support loops, it means support loops and edge profiles can be changed at any time based on feedback and it means the base mesh can be used as low poly when the support loops and subdivision modifiers are turned off.

    Obviously this workflow isn't the be all end all and doesn't work for everyone in every situation but it does speed up hard surface subdivision modeling if the application supports it. There's still a place for manual editing. It's just that all of the more contemporary workflows like boolean / CAD re-meshing and to some degree the rounded edge shader workflows put a little more time pressure on subdivision modeling so every little bit of efficiency helps.

  • PancakeMSTR
    Offline / Send Message
    PancakeMSTR polycounter lvl 4

    If a mesh subdivides without any major smoothing artifacts then it's generally considered passable. A shiny matcap with a soft highlight roll-off is helpful for spotting subtle smoothing artifacts and surface undulations. It might be worth using this type of shader to check the sides of the shapes at glancing angles to see if there's any issues hidden by the current material settings.

    Yeah the mesh "works," just as a beginner I'm curious about what experts have to say. It has mild issues at glancing angles 

    But I don't think there is a particularly easy solution to that, except maybe at the bottom where the part starts curving upwards?


    P.S. it's a canopy for an RC helicopter

    Not really sure why I'm making it, for fun and practice I guess (trying to learn blender because I want to get away from 3ds max).
  • Blaizer
    Offline / Send Message
    Blaizer interpolator
    @PancakeMSTR In the frontal zone i would have used a triangular topology. Triangles can also be used in subdivion modeling in the first iterations (lvl0 or base mesh)

    The side of the helicopter seems to be curved.
  • count23
    Offline / Send Message
    count23 polycounter lvl 3
    Howdy, big fan, long time lurker, first time questioner.

    I'm trying to model this shape and having a particular issue with this part here.

    For reference. My low poly cage. I'm using the bevel modifier with a shape 1.0 and 2 segments to proceedurally generate control loops for easy of adjustment.

    I'm shrinkwrapping to this shape here, same deal with modifier control loops:


    My order of operations is currently Bevel+subd+shrinkwrap, with the expected outcome that when i shrinkwrap the base gave to the guide cage, the bevel modifier should have been applied and then subdivided twice to give it a nice rounded edge that overlaps with the guide mesh's edge (Both are in the same position AND are the same width. However, it looks like this:


    the control edges are dragged down onto the flat, despite the two cages being positioned identically to each other and the edge loops are 100% on top of each other with no deviation. What's odd is it seems the shape starts off lined up properly on the right, but then sinks in on the centre-line.

    When i change the order to SW+Bevel+Subd, the shape works but thanks to Catmull Clark, the entire shape works, but it's suffering the volume contraction that subd tends to:

    I tried using an offset and above/outside surface in SW and it just deformed the mesh without expanding it:

    I was hoping someone would be able to tel me if i'm approaching this correctly or if i'm mangling the shape and making it far too complicated. Or if there's some other topo issue i have wrong that explains why this is not working.
  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    Overall the topology routing around the major forms is well structured but the undulations in the shape is an issue. There's a lot of stray edge loops coming off some of those cut outs and this disrupts the segment spacing that controls the smoothing behavior of the curvature along the sides.

    Dissolving some of these extraneous edge loops and connecting the corners of the cut outs to the nearest geometry segment may resolve some of the shape issues along the bottom and near the tail boom. Any left over triangles or n-gons that aren't causing major smoothing issues should be fine and can be left as is.

    The basic shapes should generally be defined at a lower mesh density and the edge segments of the block out should be positioned to act as support loops for any intersecting shapes. Once the basic shape is correct it should be a fairly straightforward process of increasing the density of the base mesh to support any intersecting shapes.

    Below is a basic example of what this type of block out process can look like. Start with the basic shapes and establish the topology layout then use modifiers to add features and support loops while adjusting the base mesh to match the shapes in the reference images. Clean up the mesh and make any final adjustments to the major forms then apply a reasonable amount of subdivision to evenly increase the geometry to support adding additional surface features.


    Another option for the block out would be to outline the shapes with curves, match the segment counts and loft the geometry. There's a lot of different ways to approach modeling this kind of shape and the previous page has a few relevant discussions: Udjani shared a link to a video that covers adding details with a limited amount of cage mesh geometry and Sacboi shared a link to an excellent article about class A surfacing for vehicle models.

    To recap:
    The surface quality of the starting shapes tends to have a large impact on the mesh's overall appearance so it's important to have a solid base before adding details. It looks like the underlying mesh is good so try optimizing the geometry by dissolving some of the extraneous edge loops and see if that improves the smoothing behavior of the curvature along the sides of the model.

    The underlying shapes are fairly simple and the shrink wrapping process may be adding a lot of unnecessary complexity. Try blocking out the shapes and using the existing geometry as support for the cut out. Even a basic mesh like this can support some fairly tight edges.

    Here's a basic overview of what this process could look like. Trim and extrude a cube sphere to generate the starting shape. Add a support loop near the cut out. Use an inset operation to outline the shape and dissolve the last arc segment at the end of the shape then join the corner vertices to resolve the n-gon and add a support loop near the base of the corner. Use a bevel / chamfer modifier, controlled by edge weights, to generate the support loops. (Use arc miter to generate the proper support loop geometry in the corners.)



    This type of topology layout should support both wide and narrow edge widths. How sharp the shape transitions need to be will depend entirely on the player's view distance. A couple of pages back there's another discussion that covers a similar shape in greater detail. The same topology strategy can also be used when attaching curved segments to flat shapes.



    To recap:
    Player's average viewing distance should be used as a benchmark for determining the appropriate edge width. Offsetting shape intersections allows the underlying geometry to be used as support without introducing severe deformation on curved surfaces.
  • ANAFREE
    Hi guys. i tried to use FrankPolygon technique, some time it works as well but some time i got bad result even if the faces are flat!



    with subdivision modifier



    the problem appears more in  blender


    i don't know why, can you help me? if i need to add supporter edges can you tell me where i could add it ?

    The fxb file to check
    https://drive.google.com/file/d/1qkT9nm-syf7kf0PVcdZ1JhHcvb9gLR-9/view?usp=sharing







  • aregvan
    Offline / Send Message
    aregvan polycounter lvl 7
    @ANAFREE

    It would be best and less of a hassle to separate the handles from the cube mesh. I do not see a reason to keep them connected.

    As the subdivision tries to smooth out all the elongated topology, it creates overlaps which causes the z fighting.

    Separate the mesh, clean the messy face, and add a subdiv modifier.


  • ANAFREE
    aregvan said:
    @ANAFREE

    It would be best and less of a hassle to separate the handles from the cube mesh. I do not see a reason to keep them connected.

    As the subdivision tries to smooth out all the elongated topology, it creates overlaps which causes the z fighting.

    Separate the mesh, clean the messy face, and add a subdiv modifier.



    Thank you Aregvan
    i just trying to understand this type of modeling, What i can do and what can't :)

    What confuses me: Why do some people still use subdivision modeling for hard surface modeling / without curves?
    like this one
    https://www.youtube.com/watch?v=tXyZBefY_Ww

    did I miss something about Non-SubD ? 





  • FrankPolygon
    Offline / Send Message
    FrankPolygon grand marshal polycounter
    Aregvan is correct: the smoothing artifact is caused by overlapping geometry and modeling this type of surface detail as a separate piece of floating geometry will be more efficient than trying to blend everything together into a watertight mesh.

    There's an extra edge loop that runs between the primary support loops and there's also a couple of faces with close to zero surface area. This extra geometry shortcuts the primary support loops and generates overlapping geometry when the subdivision smoothing pulls everything inwards.

    Dissolving that extra loop and fixing some of the other topology issues by cutting a couple extra support loops near the corners resolves the smoothing artifacts.



    Modifier based workflows are about efficiency and flexibility.

    Taking a very high level view of everything: subdivision is just a process that adds geometry and smooths out the shapes. Most boolean re-meshing workflows use polygonal subdivision but the way the process is abstracted and automated makes it feel like something else to a lot of artists.

    Modifier based workflows are the same. They speed up the modeling process by automating repetitive tasks and generating consistent or repeatable geometry from simpler input meshes. Not too dissimilar from CAD workflows in the way shape structure and order of operations are important factors in producing clean results. These type of modeling processes all tend to be GIGO the whole way down.

    When done properly this type of workflow is FAR removed from traditional box modeling and fiddly manual edge loop management. It's a streamlined modeling process that takes simple input meshes and generates both the low poly and high poly models. Any changes to the base mesh are pushed through the modifier stack and are ready to go immediately. The base meshes can be sent out for detail sculpting and re-meshing once changes are final and everything is approved.

    Here's an example of what this process can look like. The base geometry is very simple and additional details like the wide chamfers and the support loops are all generated by modifiers that can be adjusted at any time. Areas that need additional support can have major support loops added in the base mesh. Shape perimeter edge loops are all generated and controlled by modifiers.

    Surface details like the ladder rungs are split in half, instanced and repeated across the mesh. Changing he length of the pillar and adding additional rungs is as simple as changing a number in the modifier stack.

    It's also worth mentioning that using a rounded profile to add the support loops moves the outside edges of the shape and this tends to introduce it's own type of smoothing artifact. Square (2-4) or chamfered (3) support loop profiles tend to produce better results than rounded profiles.


    Without getting too far into the ideology and philosophy of it all: there's a lot of outdated or dogmatic subdivision modeling strategies that require a significant amount of manual editing. Compared to more contemporary re-meshing workflows they are objectively time inefficient for creating game content. That's not to say that there aren't edge cases in game art or other fields where this kind of typical old school, pure quad grid topology and all the rest of it is needed. It's just that, in the hands of your average Jr or Mid level artist, this type of old school process isn't competitive with newer processes in this particular field.

    The other side of this is that there are different inefficiencies in the other modeling processes. This is where being able to start with one basic input mesh and output whatever is needed really shines. There's also situations where stuff is going to be changed a lot and how many times can you run something back and forth between wholesale changes, polishing and detail passes before it sucks up a lot of time? Quality process should be in place to prevent this kind of situation but it still happens from time to time.

    Modifier based workflows bridge the gap and can be used to generate pretty much whatever base geometry is needed. It's not CAD but it can also have some degree of parametric accuracy that's missing from other approximation based strategies.

    Above all there's just a lot of misconceptions about how subdivision works and how broad the subdivision modeling process really is. It's a can't see the forest through the trees sort of situation. Most players won't ever see the high poly topology but it's very easy for beginners to get stuck on perfecting measurable technical aspects rather than focusing on the more abstract and subjective skills needed to improve the overall quality of the piece as art.
  • ANAFREE
    Aregvan is correct: the smoothing artifact is caused by overlapping geometry and modeling this type of surface detail as a separate piece of floating geometry will be more efficient than trying to blend everything together into a watertight mesh.

    There's an extra edge loop that runs between the primary support loops and there's also a couple of faces with close to zero surface area. This extra geometry shortcuts the primary support loops and generates overlapping geometry when the subdivision smoothing pulls everything inwards.

    Dissolving that extra loop and fixing some of the other topology issues by cutting a couple extra support loops near the corners resolves the smoothing artifacts.



    Modifier based workflows are about efficiency and flexibility.

    Taking a very high level view of everything: subdivision is just a process that adds geometry and smooths out the shapes. Most boolean re-meshing workflows use polygonal subdivision but the way the process is abstracted and automated makes it feel like something else to a lot of artists.

    Modifier based workflows are the same. They speed up the modeling process by automating repetitive tasks and generating consistent or repeatable geometry from simpler input meshes. Not too dissimilar from CAD workflows in the way shape structure and order of operations are important factors in producing clean results. These type of modeling processes all tend to be GIGO the whole way down.

    When done properly this type of workflow is FAR removed from traditional box modeling and fiddly manual edge loop management. It's a streamlined modeling process that takes simple input meshes and generates both the low poly and high poly models. Any changes to the base mesh are pushed through the modifier stack and are ready to go immediately. The base meshes can be sent out for detail sculpting and re-meshing once changes are final and everything is approved.

    Here's an example of what this process can look like. The base geometry is very simple and additional details like the wide chamfers and the support loops are all generated by modifiers that can be adjusted at any time. Areas that need additional support can have major support loops added in the base mesh. Shape perimeter edge loops are all generated and controlled by modifiers.

    Surface details like the ladder rungs are split in half, instanced and repeated across the mesh. Changing he length of the pillar and adding additional rungs is as simple as changing a number in the modifier stack.

    It's also worth mentioning that using a rounded profile to add the support loops moves the outside edges of the shape and this tends to introduce it's own type of smoothing artifact. Square (2-4) or chamfered (3) support loop profiles tend to produce better results than rounded profiles.


    Without getting too far into the ideology and philosophy of it all: there's a lot of outdated or dogmatic subdivision modeling strategies that require a significant amount of manual editing. Compared to more contemporary re-meshing workflows they are objectively time inefficient for creating game content. That's not to say that there aren't edge cases in game art or other fields where this kind of typical old school, pure quad grid topology and all the rest of it is needed. It's just that, in the hands of your average Jr or Mid level artist, this type of old school process isn't competitive with newer processes in this particular field.

    The other side of this is that there are different inefficiencies in the other modeling processes. This is where being able to start with one basic input mesh and output whatever is needed really shines. There's also situations where stuff is going to be changed a lot and how many times can you run something back and forth between wholesale changes, polishing and detail passes before it sucks up a lot of time? Quality process should be in place to prevent this kind of situation but it still happens from time to time.

    Modifier based workflows bridge the gap and can be used to generate pretty much whatever base geometry is needed. It's not CAD but it can also have some degree of parametric accuracy that's missing from other approximation based strategies.

    Above all there's just a lot of misconceptions about how subdivision works and how broad the subdivision modeling process really is. It's a can't see the forest through the trees sort of situation. Most players won't ever see the high poly topology but it's very easy for beginners to get stuck on perfecting measurable technical aspects rather than focusing on the more abstract and subjective skills needed to improve the overall quality of the piece as art.

    Thanks heaps for the detailed info Frank and for Aregvan too ❤
    You are my inspiration, i really appreciate your wonderful effort in detailing and explaining everything ❤👍



  • Rolf
    Offline / Send Message
    Rolf polycounter lvl 3
    recently I completely redid the ship's body, and now other difficulties have arisen - this is the connection of the two parts of the ship and the protrusion where the propeller is located, when connecting I violated all the rules of topology, can someone tell me how to connect so that it looks more accurate.then I also need to add a ledge under the central propeller, which can finally break the topology.below I will add a file with my model please rate the quality
Sign In or Register to comment.