@Raphael_Bouch I dont know how it looks from reference images however when it comes to vehicles its best to have them in seprate parts as in real life unless you are doing super tight low poly with limited vert count budget.
Hi, sorry forgot to add the refs ^^' ; Well a simple google search and there you have it
You see aside from the "hood" at the rear and a few other parts all shapes seems to flow into each other seamlessly. I'm making a high poly mesh so I'd like to avoid any "smash that geo inside and no one will notice". Would be much more easier if it was a fire arm since those are made of hundreds simpler separate parts bolted together lol.
Hi folks, I've been practicing the ol' cutting a cube out of a cylinder, and thanks to @FrankPolygon's tutorials I have arrived at a place I'm very happy at. I then wanted to try a cube from a sphere, and I have figured a way to make it work for me. Would you mind just looking over it to see if it is the most optimal way of doing it?
Started with a quad sphere and offset extruded the cube.
Bevelled the edges that were selected in the first picture, and set the bevel to 2 segments, but used radial mitering.
Corners ended up like this
I used multi cut
and then deleted the centre of the two tris
smooths pretty well.
Would you do the same, or is there an easier approach? Many thanks in advance.
@christrom There's a bit of subtle pinching caused by the extra loop between the base of the square and the loop path on the surface of the sphere. Dissolving the loop and merging the corner vertices of the triangular quads should resolve the smoothing artifacts while also simplifying the mesh. (See the third and fourth examples below.)
If a high poly mesh subdivides cleanly and provides the desired level of surface quality then it's generally passable but whether or not a topology layout is optimal depends entirely on what the constraints are.
Subdivision modeling is often a balance of accuracy, efficiency, and quality. When something needs to be done quickly then it may make sense to sacrifice some accuracy or quality by using some less than ideal topology routing to connect new shapes to the existing geometry. In cases where accuracy or quality is more important then it often makes sense to spend a bit more time matching the shapes in the references and a bit more effort generating consistent quad grid topology.
The goal should be to find a sweet spot in the workflow that matches the desired level of shape accuracy and surface quality while also using the minimal amount of resources like time, geometry, etc.
Unlike basic cylinders, quad spheres produce a consistent grid topology. An easier approach would be to line up the edges in the grid with the intersecting square shape then extrude off the surface of the sphere and bevel the edges around the feature to create the support loops. This approach is quick but it does require adjusting the geometry density. Which can be somewhat limiting if the intersecting shape is in an odd location or is a different size that doesn't fit the existing grid.
If the quads are too large then it can be difficult to support the shapes when subdivision is applied. Especially when terminating the corner loops into a quad triangle to prevent the support loops from running out onto the surface of the sphere. It can be tempting to try and resolve this by connecting the corner vertex across the adjacent quad but this moves the pole further away from the supported area at the base of the intersection. Which tends to produce visible smoothing artifacts. Sometimes these smoothing artifacts can be minimized by softening the profile of the support loops but it's often better to just adjust the mesh density or topology to support the area around the shape intersection.
When the new shape falls between the existing edges of the quad grid then it creates a natural support loop path around the base of the intersection. It is possible to use the same bevel / chamfer operation to generate the support loops but there needs to be enough space between the center and outside edges of the loops. If the added support loop disrupts the segment spacing around the shape intersection then it can cause overlapping geometry or pinching artifacts.
These types of smoothing artifacts can generally be resolved by over and under scaling the intersecting shapes to provide more room between the base of the intersection and the outside support loop. It may also be possible to minimize the smoothing artifacts by softening the profile of the support loops but this generally won't work if there isn't enough room between the loops.
Which is why it often makes more sense to either place the inside support loop on it's own or remove the extra outside support loop manually with a loop dissolve. The existing geometry in the larger shape will act as it's own support for the outside of the shape intersection. With some careful planning and experimentation it should be possible to solve most of these basic topology flow issues during the block out phase.
Increasing the amount of geometry will make it possible to carry support loops across the surface of the sphere without causing artifacts. This is mostly useful when certain types of corner topology are a requirement but there can also be situations where a high quality surface is required and in those cases it does make sense to increase the amount of starting geometry. With this type of topology layout it's generally possible to sharpen the corners by sliding the outer edges closer to the corner. Just be careful to avoid causing any unintended surface deformation. Which can cause it's own type of smoothing artifact.
There are also some situations where alternate modeling workflows that use floaters may be an option. This thread has a few previous discussions that cover different uses for floaters and examples of different topology layouts for panel lines and through holes. Definitely worth the time to do a quick search and skim through a few of those examples.
Recap: When a quad sphere's grid lines up with the outline of surface features it's often possible to extrude directly off the surface. If the intersecting shapes aren't aligned with the grid then make sure there's enough geometry to create a consistent loop path around the intersecting shapes. Try to avoid unnecessary loop complexity whenever possibly by relying on the existing geometry as part of the support loops. Solve the major topology flow issues early in the block out and avoid pushing corner poles out into unsupported areas of the mesh. These same strategies will also work with negative shapes like cut outs and through holes.
Thanks again Frank. There really should be a tip button on here as you are giving away such valuble information for free. Have you ever considered a YouTube channel? You'd get loads of subscribers. I've worked into the original mesh and got rid of the extra support and it holds fine. I should have really thought about using the existing geometry in your cylinder example, as the last screenshot on here is doing very much the same thing. As long as the geometry is dense enough it really doesn't affect the silhouette. Thanks again, you should be charging for this! I would gladly pay.
Hey guys! First time posting here so I'm sorry if this has been asked and answered before buut I'm having a real hard time producing a good looking result cutting details into curved surfaces like so:
.
I've tried a few different methods, and haven't really come up with a satisfactory solution. Shape I'm going for has a slight horizontal arch that fades and flattens out the further up the vertical arch it goes, and then cutting details into said surface.
so I made a very rough boolean opeartion, after that I did manual cleanup for edge flow on open area for engine section just to give you an idea.
There are number of approaches you can take though,
You can try spline cage method for modelling this as one highpoly mesh, most commonly used for modelling cars.
However I would make a proper blockout of main shapes, run a boolean and split parts, clean the split parts making sure the border edges dont move and add details. This way it wont be "smash that geo inside"
I am sure you are using more reference images instead of that blueprint you shared as its really hard to read. I would suggest looking into scale modelling groups on Facebook for more break down photos of this model. That will greatly help in modelling.
Since its symetriacal design so model just one half as this will not overwhelm you. Remember there are no shortcuts, you gotta keep pushing pulling verts all day :D
@Thanez Hello!! thank you so much for resolve my question, I know I´m little bit a picky person jeje.
jaja don´t worry I can wait, and yeah for sure! I´m interested.
So Thanez let me share you another problem it suppossed to be easy but still I´m having problems… first I´m aware that I can resolve it with te boolean process that u already teach me but I think it could be more faster with the align working pivot.
I goal is to keep the planar and avoid to all cost non-planar geo.
The problem is if I take the front faces and I rotate them I got a lot of Non planar faces
2. So I use the working pivot in one of the edges and move the vertex along that direction in order to preserve the planar faces:
So one part its fixed the main problem is that te direction of the bottom vertex are not matched with the direction of the top vertex (I mean the diagonal direction created from the working pivot)
So I´ve used the same process but I know that it will not be ACCURATE.
Please tell how can I do it accurate, along the diagonal edge from the top in order to have all my faces Planar.
I'm gonna try and do a video this time because I usually spend something like 5-15 minutes modeling something to show you, but 40-50 minutes to write it all up to get my thoughts in there correctly, in order to avoid the norwenglish you'll soon be witness to.
When It's done processing on youtube, you can see it here:
While that's uploading I can point out how that resource you posted is really good. Those are some NEATO tricks I use constantly. They do work really well for him because the edges he's aligning all exist on a flat plane on the XY axis: In essence, he's working in 2d. Your edges are not on a flat plane, you're in 3d. When you try and force polygon 1 to be planar, you're at the same time affecting the vertices that make up the polygons next to it, shifting those polygons out of planarity. It's an issue.
Aaaand this took me 25 minutes instead of 55 minutes. I think I'll do this more in the future if it's helpful.
Hello!! Glad to reply my result of your first tip :)
so here is my example with the awesome help that you provide me about the boolean operation:
I´ve had implement the facet shading and it helps me a lot.
I can tell that in your example the final part its scalated in Z axis soo in that part naturally do u have nonplanar geo right? because you say in your first explination that "When you squish a circle like that, you make the edges of the polygons non-planar, revealing the fact that all quads are made of triangles. Turning on facets shading mode illustrates the issue perfectly. Here on the right, I stole your oval shape and remade the big part but without squishing it."
@SignalFlare07 Welcome to Polycount. Consider checking out the forum information and introduction thread.
This reply is a bit late but you're on the right track. Try to match the segments of the intersecting shapes and their support loops with the existing geometry of the underlying curvature. This will help prevent unintended surface deformation caused by support loops disrupting the curve's segment spacing. Use an iterative block out process to find how much geometry the lofted section needs to be able to support the smaller shapes.
A couple pages back there's some detailed write-ups that cover blocking out similar hard surface shapes. Here's a few links to those posts.
Extraneous support loops on curved surfaces can either be terminated in a well supported area or carried across the existing edges of the curve. Which option makes the most sense depends on the complexity of the shapes and the amount of geometry available. Increasing the amount of geometry in a curve will make it easier to carry the support loops but it can also reduce the overall editability. Which is why it's generally considered best practice to try and match the segments of intersecting shapes and resolve the major topology flow issues before adding lots of support loops.
Here's a few links to some write-ups that discuss routing support loops on curved surfaces.
There's a lot of different ways to approach the order of operations but an iterative block out process will make it a lot easier to solve some of these topology flow issues. Below is just one example of what that block out process could look like.
Start by blocking out the proportions then add chamfers and curves. Modifiers can be used to generate these features non-destructively. Which will make it easier to go back and adjust the segment count, without having to re-model individual sections of the mesh.
Define the lofted surface profile with very basic geometry and smooth it using subdivision or a bevel / chamfer modifier. Boolean operations can be used to cut out the rest of the shapes. Adjust the number of segments in the lofted surface and the radiused cut outs until most of the edges are aligned. The segment matching doesn't need to be perfect. Close enough is usually good enough.
It's generally fine to leave a few triangles in the mesh if they are well supported and aren't causing any visible smoothing artifacts. If the mesh needs to be resolved to all quads for a specific technical reason then go back through the block out process and adjust the density of the curvature and the placement of the intersecting shapes.
Clean up any stray geometry and adjust the topology flow. Try to maintain consistent segment spacing along the curved surface. Use basic loop cut, join through and edge dissolve operations to re-route the topology. Keep everything co-planar by sliding the edges along the surface when making manual loop adjustments. Cut in some additional loop paths to define the chamfered feature on the bottom and slide the remaining vertex upwards to create the triangular notch above it.
Clean up any remaining loop routing issues then add support loops by beveling / chamfering the highlighted edges. Most of the corner support loops on the rectangular pockets can be terminated in a triangular quad or merged down into triangles that are anchored to the adjacent vertex on the curved surface.
Recap: Block out the basic shapes first then use an iterative block out process to solve the topology flow issues before adding the support loops. Let the shapes define the loop flow paths. Match the segments of the intersecting shapes. Extra support loops can either be terminated in well supported areas or carried across the surface.
Hello guys, I have been stuck here for days and I can't figure it out. The topology doesn't look good no matter what I try. I have tried many things, but none of them worked.
Beginner here, how do I connect a five sided polygon to a four sided in C4D? I'm studying to make a bottle, and the outline is the correct shape but I can't figure out for the life of me how to connect these edges and get rid of non planar polygons? A huge thanks in advance to the forum!
Beginner here, how do I connect a five sided polygon to a four sided in C4D? I'm studying to make a bottle, and the outline is the correct shape but I can't figure out for the life of me how to connect these edges and get rid of non planar polygons? A huge thanks in advance to the forum!
You can use boolean. Step by step in Blender: Bevel the edges > duplicate the triangle and separate to new obj > origin triangle to corner (use 3d Cursor is ok also) > rotate and scale triangle > solid and finally boolean with cube…
Notice here when you rotate, use transform orientations by normal (create new custom if u like) and set pivot point to active, from here you can rotate or use shear tool. Hope this help…
Beginner here, how do I connect a five sided polygon to a four sided in C4D? I'm studying to make a bottle, and the outline is the correct shape but I can't figure out for the life of me how to connect these edges and get rid of non planar polygons? A huge thanks in advance to the forum!
You can use boolean. Step by step in Blender: Bevel the edges > duplicate the triangle and separate to new obj > origin triangle to corner (use 3d Cursor is ok also) > rotate and scale triangle > solid and finally boolean with cube…
Notice here when you rotate, use transform orientations by normal (create new custom if u like) and set pivot point to active, from here you can rotate or use shear tool. Hope this help…
Never thought of it, I will give it a try - thanks a lot!
Hey guys I´ve a very important question for Blender, How can u see in realtime your shading with non planar geometry. In 3ds max you can see this with these options: This is the result: soo in Blender these geo is non planar but you can´t see that in the default shading: If I apply a triangulate modifier it works I would like to know if there is another way to see that problem.
Hope you can help me Cheers I´m really frustrated for this problem
@FrankPolygon Thank you very much for the reply! Out of curiosity, what is the tool you're using to create the first boolean cutter? I've been attempting it using lattices but it doesn't come out quite right. Using Maya 2024 by the way.
@SignalFlare07 Glad the write-up was helpful. The lofted shape on the back of the receiver can be created with basic subdivision or with a series of bevel operations.
Here's what the subdivision approach could look like. Start with a basic cage mesh and refine the shape with subdivision preview enabled. Adjust the subdivision level up or down until the edges in the curve are roughly aligned with the rest of the boolean shapes. Using subdivision to create this surface profile is fast but there's limited control over where the segments fall. Making the cage mesh larger will spread out the edges and adding or moving edges in the cage mesh will bring the edges closer together but beyond that there's not a lot that can be done to accurately control the edge placement.
Here's what the bevel approach could look like. Start with the same basic cage mesh then bevel the center edge longitudinally to create the basic profile and segment spacing. Then select the perpendicular edge and bevel that to create the lofted curve and segment spacing. This approach provides a lot of control over the placement of the edge segments but does require a lot of trial and error when using destructive modeling tools.
It might also be possible to combine both approaches by using subdivision preview to create the basic cage mesh then turning the subdivision off and beveling the cage mesh to control the placement and shape of the lofted surface.
@FrankPolygon Holy crap you're awesome! Really appreciate you taking the time to do these writeups. Been getting more into hard surface modelling and the previous writeups you've done in the thread have been really helpful!
I'm looking for advice for Maya modeling - how the heck does one approach modeling a spool of cable?? Sweepmesh is great and all, but the whole modeling a helix, extracting a curve and then sweepmeshing thing results in a very clean artificial looking spool. I want some of strands to overlap and whatnot, but I've been researching for hours now and no results. I can't find any way to add some random noise to the curves or SOMETHING similar?
hey i have this problem i dont know how to fix it, please help me.
A few extra support loops should help a lot.
If you're going to use an n-gon heavy subdivision workflow, it really helps to have frequent and even loops and avoid dense bevels. Here's a twitter page to look at for inspiration if that's how you want to approach modelling,
I'm having a hard time getting this curved piece to look like my reference photo. I've worked on this one spot for about 3 hours, I've added support edges, creases and everything I can think of but I cant maintain my smooth curve
Hey, im tyring to make this droid from a concept art, can anyone help me with the base shape? This is what ive currently got, i dont know how to model the part where the "eye" is if anyone has advice thatd be great!
Looks like you have the primary forms. Just continue iterating on the block out to create the secondary forms. Below is an example of developing this shape with an iterative block out process.
Create the basic shape of the secondary feature using inset and extrude operations.
Use edge dissolve, edge slide, and shrink operations to develop the curved profile of the secondary feature.
Continue iterating and refining the geometry until the block out is completed. Support loops can then be generated using bevel / chamfer operations on the edges that define the shapes. Here's an example of what the basic topology flow could look like.
Recap: Keep the initial block out relatively simple and focus on developing the larger shapes first. Continue working through this process to generate the secondary and tertiary forms. If necessary, split the high poly mesh into individual components, as depicted in the reference images. Apply subdivision levels as required.
Adding parallel support loops down the middle of the guitar body will sharpen the inside corner of the recessed surface but will likely produce a smoothing artifact on the outside curve. There's a couple of different ways to handle topology routing around cut outs on curved surfaces but it's generally considered best practice to adjust the segment count along the curve to support the shape intersection.
It may be possible to cut in some additional curve segments and manually reroute the topology on the existing model but it's probably faster and easier to just redo the entire block out. Keep the topology relatively simple and focus on capturing the larger shapes first. Create curves with enough segments to support any intersecting shapes or cut outs. Consider adding the larger round overs as part of the block out geometry. This will make it easier to capture some of the shape transitions near the edge of the milled pocket for the guitar's neck.
Here's an example of what an iterative block out process could look like. Create the basic profile of the guitar body then add the round overs and use a boolean operation to cut out the neck pocket. Modifier based bevel / chamfer operations can make it easier to adjust the number of segments in the curves so the edges line up as support for the intersecting boolean meshes. Remember to use the existing segments of the curved surfaces as part of the support geometry around the shape intersection that defines the neck pocket.
Clean up around the boolean operation then cut in any additional support loops required for the smaller details. Try to solve most of the basic loop flow issues before adding the support loops and use the existing geometry in the shapes to help define the loop paths. This iterative approach to blocking out the shapes will generally make the modeling process a lot easier.
Final support loops can be added to the edges that define the shapes using a simple bevel / chamfer operation or modifier. The subdivision preview below shows how the existing geometry of the curve supports the intersecting shape and guides the topology flow around the mesh.
Recap: Focus on blocking out the shapes first and try to use the existing geometry of curved surfaces as part of the support loop paths around shape intersections.
What is the best way to model rectangular, sharp inserts on the wall of a curved cylinder? I tried retopo-ing the beveled edges to maintain quads but theres still artifacts? (You dont have to model all three rectangles, only 1 will suffice). Thanks. Wire:
There's a couple of different ways to approach the modeling process: do a rough block out for the primary and secondary shapes then adjust the segment count of the larger forms so there's enough geometry to accurately represent the smaller shapes or create an accurate block out of the larger primary forms then apply enough subdivision levels to support the smallest of the secondary forms. Deciding which iterative block out process to use really comes down to the available tool set and preferred modeling workflow. What's most important is to focus on creating accurate shapes and resolve most of the potential topology flow issues before adding all of the minor support loops.
Below is an example of some basic topology layouts for a square pocket on a truncated cone. The edge segments that make up the walls of the truncated cone are used as part of the support loop paths around the shape intersections and there's an offset between the existing edges and the edges of the intersecting shapes. This makes it easier to adjust the loop flow paths without generating unintended deformation.
There are some minor smoothing artifacts on the mesh with the lowest segment count but they are very subtle and mostly in areas where the intersecting geometry disrupts the segment spacing of the edges that make up the curved surface. Increasing the number of segments in the truncated cone does improve the surface quality but there's diminishing returns at higher density levels. Especially when it becomes difficult to balance the edge placement around the square pocket.
Here all of the support loops used to sharpen the corners run out into the adjacent face on the wall of the truncated cone. Since the n-gons aren't causing any visible smoothing artifacts there really isn't any major benefit to trying to resolve the mesh to all quads. If all quad geometry is required for specific technical reasons then there's a few different ways to handle extraneous loop termination but this will likely require using more segments or creating new edges across existing faces. Which complicates the base mesh. Triangular shaped quads can be used to get around some of this but they sometimes produce their own type of noticeable smoothing artifact or require manual editing to adjust the vertex placement along the existing surface.
So, as far as most hard surface game art goes, resolving every mesh to all quads is more of a tradeoff where efficiency is sacrificed to achieve technical purism. Traditional subdivision modeling workflows are slow enough as it is. Which is why it can be helpful to be a bit more pragmatic about mesh density, subtle smoothing artifacts, and leaving in a few triangles or n-gons that aren't causing visible smoothing artifacts.
Recap: Look for existing solutions to smoothing artifacts on similar shapes. Block out the shapes first then resolve major topology flow issues without creating unintended surface deformation. Try to use the existing geometry of the larger shapes as part of the support loop paths around smaller shape intersections. Adjust the segment count of the adjacent shapes so the edges are roughly aligned. Maintain the accuracy of each shape by constraining any significant changes in loop flow to the area between the outside support loops and the edges at the base of the intersecting shapes.
Hello, how can i model those armrests of office chair properly? (and quick) I just wanted to model a chair for game project, but stuck on those two, it's become my idée fixe which doesn't let me finish this without modeling armrests right (i hate it). I don't even know how to approach them, did multiple tries, but i didn't liked any of them. At first tried to do separated meshes - they're too visible Then tried merged, but it feels wrong, i don't like how topology come out Also tried to merge in ZBrush with dynamesh, but couldn't get smooth shading.
Hey im trying to get this base shape (circled blue) layer out but im not sure if im going about it in the right way. Something about what I currently have isnt right, there isnt a smooth transition between the large circle and the smaller circle. Perhaps instead of trying to make the circles, i should try modeling cylinders of the two shapes and then combining them together. Ive attached what I currently have as well as the ref and a link to other photos of my ref, I am only trying to get the base shape of the two circles together, im not worried about making all the Booleans and small details.
I think I figure out the shape (the small circle shape is still slightly off but ill keep working on it), I figured I could post my solution here unless anyone would like to see it in the future.
I am trying to match shape II in the picture below. Shape I is my attempt.
My method is to use a rectangle and make the lower face smaller. Then I put an edge loop between the top and bottom face, push it out aways, and bevel it.
However, the bevel only goes between the top face and the mid point. I need an even distribution between both extreme ends. Not sure why the bevel does not work that way. I am working in Maya 2022.
Steps:
1. Bisect a rectangle with an edge loop
2. select that middle edge loop and bevel it. But when adjusting the Fraction parameter, we only seem to adjust between top face and mid. I want to adjust between top and bottom faces evenly.
Secondary question: Can anybody tell me what this shape is called? It is an architectural shape... would you call it a cornice?
Is there something besides a bevel I could use?
I tried a loft with curves and it gives good non-destructive editing but you end up with weird geo where the curve joins itself:
I'll provide my own answer here but certainly interested for any advice. It looks like lofted curves might be best thing in maya, though for this simple case it seems overkill. Here's what I've done though:
1. First I misunderstood the shape. Middle edge shouldnt be equal distance from extremes. Should be nearer to one side.
2. Bevel that
3. Then just move the top face down and it is done.
Job is accomplished but it seems imperfect because of having to tweak the initial shape. But whatever I guess its not end of the world.
My issue with curves is that I can't find a way to "smooth" a collection of CV joints, so the curve portion has to be made manually by eye, rather than letting computer do it.
Hmmm.. in blender you can do something like bevel edges and so...
Ah, works same way in maya after testing. I dont know why I had an idea to add a middle edge and bevel that instead. Made it more complicated than necessary.
Sometimes the mind is just on some worn out ways.. and if you stop and look up.. you think.. why did i wanted to go this way ? (applies to a lot of things )
I am a matter of hours away from purchasing my first proper computer, a nearly 3000€ machine built in large part thanks to the wealth of information regarding render requirements, functional hardware options, and suggested builds to be found on this website.
Further searches led me here, to this thread. a thread started over a decade ago, still going strong and active, with (seemingly) nothing but good-willed advice between experts, masters, and beginners like me.
I don't have a model to post here. not yet at least. I don't have any functional advice to share, for I am a beginner myself.
but I couldn't NOT point out the absolute bliss and pleasure it is to come across a forum like this. It's like I just came across a treasure in the depths of the ocean.
Hello I really need help on the issue of modeling a car wheel.
The thing is that the usual so say symmetrical wheel is not a huge challenge to model, but here is a wheel with unusual forms (attached photo orange wheel) I just do not understand how to make such a form. The third day already tortured.
I would be very grateful if you could tell me HOW to model it reasonably. It just makes me want to scream. - How The F*#% Do I Model This? Thank you very much in advance
Replies
@Raphael_Bouch I dont know how it looks from reference images however when it comes to vehicles its best to have them in seprate parts as in real life unless you are doing super tight low poly with limited vert count budget.
Hi, sorry forgot to add the refs ^^' ; Well a simple google search and there you have it
You see aside from the "hood" at the rear and a few other parts all shapes seems to flow into each other seamlessly. I'm making a high poly mesh so I'd like to avoid any "smash that geo inside and no one will notice". Would be much more easier if it was a fire arm since those are made of hundreds simpler separate parts bolted together lol.
Hi folks, I've been practicing the ol' cutting a cube out of a cylinder, and thanks to @FrankPolygon's tutorials I have arrived at a place I'm very happy at. I then wanted to try a cube from a sphere, and I have figured a way to make it work for me. Would you mind just looking over it to see if it is the most optimal way of doing it?
Started with a quad sphere and offset extruded the cube.
Bevelled the edges that were selected in the first picture, and set the bevel to 2 segments, but used radial mitering.
Corners ended up like this
I used multi cut
and then deleted the centre of the two tris
smooths pretty well.
Would you do the same, or is there an easier approach? Many thanks in advance.
@christrom There's a bit of subtle pinching caused by the extra loop between the base of the square and the loop path on the surface of the sphere. Dissolving the loop and merging the corner vertices of the triangular quads should resolve the smoothing artifacts while also simplifying the mesh. (See the third and fourth examples below.)
If a high poly mesh subdivides cleanly and provides the desired level of surface quality then it's generally passable but whether or not a topology layout is optimal depends entirely on what the constraints are.
Subdivision modeling is often a balance of accuracy, efficiency, and quality. When something needs to be done quickly then it may make sense to sacrifice some accuracy or quality by using some less than ideal topology routing to connect new shapes to the existing geometry. In cases where accuracy or quality is more important then it often makes sense to spend a bit more time matching the shapes in the references and a bit more effort generating consistent quad grid topology.
The goal should be to find a sweet spot in the workflow that matches the desired level of shape accuracy and surface quality while also using the minimal amount of resources like time, geometry, etc.
Unlike basic cylinders, quad spheres produce a consistent grid topology. An easier approach would be to line up the edges in the grid with the intersecting square shape then extrude off the surface of the sphere and bevel the edges around the feature to create the support loops. This approach is quick but it does require adjusting the geometry density. Which can be somewhat limiting if the intersecting shape is in an odd location or is a different size that doesn't fit the existing grid.
If the quads are too large then it can be difficult to support the shapes when subdivision is applied. Especially when terminating the corner loops into a quad triangle to prevent the support loops from running out onto the surface of the sphere. It can be tempting to try and resolve this by connecting the corner vertex across the adjacent quad but this moves the pole further away from the supported area at the base of the intersection. Which tends to produce visible smoothing artifacts. Sometimes these smoothing artifacts can be minimized by softening the profile of the support loops but it's often better to just adjust the mesh density or topology to support the area around the shape intersection.
When the new shape falls between the existing edges of the quad grid then it creates a natural support loop path around the base of the intersection. It is possible to use the same bevel / chamfer operation to generate the support loops but there needs to be enough space between the center and outside edges of the loops. If the added support loop disrupts the segment spacing around the shape intersection then it can cause overlapping geometry or pinching artifacts.
These types of smoothing artifacts can generally be resolved by over and under scaling the intersecting shapes to provide more room between the base of the intersection and the outside support loop. It may also be possible to minimize the smoothing artifacts by softening the profile of the support loops but this generally won't work if there isn't enough room between the loops.
Which is why it often makes more sense to either place the inside support loop on it's own or remove the extra outside support loop manually with a loop dissolve. The existing geometry in the larger shape will act as it's own support for the outside of the shape intersection. With some careful planning and experimentation it should be possible to solve most of these basic topology flow issues during the block out phase.
Increasing the amount of geometry will make it possible to carry support loops across the surface of the sphere without causing artifacts. This is mostly useful when certain types of corner topology are a requirement but there can also be situations where a high quality surface is required and in those cases it does make sense to increase the amount of starting geometry. With this type of topology layout it's generally possible to sharpen the corners by sliding the outer edges closer to the corner. Just be careful to avoid causing any unintended surface deformation. Which can cause it's own type of smoothing artifact.
There are also some situations where alternate modeling workflows that use floaters may be an option. This thread has a few previous discussions that cover different uses for floaters and examples of different topology layouts for panel lines and through holes. Definitely worth the time to do a quick search and skim through a few of those examples.
Recap: When a quad sphere's grid lines up with the outline of surface features it's often possible to extrude directly off the surface. If the intersecting shapes aren't aligned with the grid then make sure there's enough geometry to create a consistent loop path around the intersecting shapes. Try to avoid unnecessary loop complexity whenever possibly by relying on the existing geometry as part of the support loops. Solve the major topology flow issues early in the block out and avoid pushing corner poles out into unsupported areas of the mesh. These same strategies will also work with negative shapes like cut outs and through holes.
Thanks again Frank. There really should be a tip button on here as you are giving away such valuble information for free. Have you ever considered a YouTube channel? You'd get loads of subscribers. I've worked into the original mesh and got rid of the extra support and it holds fine. I should have really thought about using the existing geometry in your cylinder example, as the last screenshot on here is doing very much the same thing. As long as the geometry is dense enough it really doesn't affect the silhouette. Thanks again, you should be charging for this! I would gladly pay.
Hey guys! First time posting here so I'm sorry if this has been asked and answered before buut I'm having a real hard time producing a good looking result cutting details into curved surfaces like so:
.
I've tried a few different methods, and haven't really come up with a satisfactory solution. Shape I'm going for has a slight horizontal arch that fades and flattens out the further up the vertical arch it goes, and then cutting details into said surface.
Homework done :) Thanks again Frank
Made a new attempt that I feel got a little closer but still feeling significant distortions around the corners of the inset details
so I made a very rough boolean opeartion, after that I did manual cleanup for edge flow on open area for engine section just to give you an idea.
There are number of approaches you can take though,
You can try spline cage method for modelling this as one highpoly mesh, most commonly used for modelling cars.
However I would make a proper blockout of main shapes, run a boolean and split parts, clean the split parts making sure the border edges dont move and add details. This way it wont be "smash that geo inside"
I am sure you are using more reference images instead of that blueprint you shared as its really hard to read. I would suggest looking into scale modelling groups on Facebook for more break down photos of this model. That will greatly help in modelling.
Since its symetriacal design so model just one half as this will not overwhelm you. Remember there are no shortcuts, you gotta keep pushing pulling verts all day :D
Hey, yeah that's what I ended up doing, thought you wouldn't reply ! :D
Thanks for the detailled explanation and taking your time to redo a model for me, +1 on you !
@Thanez Hello!! thank you so much for resolve my question, I know I´m little bit a picky person jeje.
jaja don´t worry I can wait, and yeah for sure! I´m interested.
So Thanez let me share you another problem it suppossed to be easy but still I´m having problems… first I´m aware that I can resolve it with te boolean process that u already teach me but I think it could be more faster with the align working pivot.
The problem is if I take the front faces and I rotate them I got a lot of Non planar faces
2. So I use the working pivot in one of the edges and move the vertex along that direction in order to preserve the planar faces:
So one part its fixed the main problem is that te direction of the bottom vertex are not matched with the direction of the top vertex (I mean the diagonal direction created from the working pivot)
So I´ve used the same process but I know that it will not be ACCURATE.
Please tell how can I do it accurate, along the diagonal edge from the top in order to have all my faces Planar.
This is my main resource
Files:
Thanx for the help.
cheers everyone
Hey buddy, glad you're back :)
I'm gonna try and do a video this time because I usually spend something like 5-15 minutes modeling something to show you, but 40-50 minutes to write it all up to get my thoughts in there correctly, in order to avoid the norwenglish you'll soon be witness to.
When It's done processing on youtube, you can see it here:
While that's uploading I can point out how that resource you posted is really good. Those are some NEATO tricks I use constantly. They do work really well for him because the edges he's aligning all exist on a flat plane on the XY axis: In essence, he's working in 2d. Your edges are not on a flat plane, you're in 3d. When you try and force polygon 1 to be planar, you're at the same time affecting the vertices that make up the polygons next to it, shifting those polygons out of planarity. It's an issue.
Aaaand this took me 25 minutes instead of 55 minutes. I think I'll do this more in the future if it's helpful.
Edit:
Here's that previous model I promised to upload: Octavio some sort of squished oval cone transition or something.max
And if you're at all interested, here's today's model:
Octavio planar problem del to.max
@Thanez
Hello!! Glad to reply my result of your first tip :)
so here is my example with the awesome help that you provide me about the boolean operation:
I´ve had implement the facet shading and it helps me a lot.
I can tell that in your example the final part its scalated in Z axis soo in that part naturally do u have nonplanar geo right? because you say in your first explination that "When you squish a circle like that, you make the edges of the polygons non-planar, revealing the fact that all quads are made of triangles. Turning on facets shading mode illustrates the issue perfectly. Here on the right, I stole your oval shape and remade the big part but without squishing it."
Cheers 🤗
Beginner here, how do I connect a five sided polygon to a four sided in C4D? I'm studying to make a bottle, and the outline is the correct shape but I can't figure out for the life of me how to connect these edges and get rid of non planar polygons? A huge thanks in advance to the forum!
This is the result:
soo in Blender these geo is non planar but you can´t see that in the default shading:
If I apply a triangulate modifier it works
I would like to know if there is another way to see that problem.
Hope you can help me
Cheers
I´m really frustrated for this problem
I've been attempting it using lattices but it doesn't come out quite right. Using Maya 2024 by the way.
It might also be possible to combine both approaches by using subdivision preview to create the basic cage mesh then turning the subdivision off and beveling the cage mesh to control the placement and shape of the lofted surface.
A few extra support loops should help a lot.
If you're going to use an n-gon heavy subdivision workflow, it really helps to have frequent and even loops and avoid dense bevels. Here's a twitter page to look at for inspiration if that's how you want to approach modelling,
Wire:
Subdiv:
I just wanted to model a chair for game project, but stuck on those two, it's become my idée fixe which doesn't let me finish this without modeling armrests right (i hate it). I don't even know how to approach them, did multiple tries, but i didn't liked any of them.
At first tried to do separated meshes - they're too visible
Then tried merged, but it feels wrong, i don't like how topology come out
Also tried to merge in ZBrush with dynamesh, but couldn't get smooth shading.
Perhaps instead of trying to make the circles, i should try modeling cylinders of the two shapes and then combining them together.
Ive attached what I currently have as well as the ref and a link to other photos of my ref, I am only trying to get the base shape of the two circles together, im not worried about making all the Booleans and small details.
I think I figure out the shape (the small circle shape is still slightly off but ill keep working on it), I figured I could post my solution here unless anyone would like to see it in the future.
Further searches led me here, to this thread. a thread started over a decade ago, still going strong and active, with (seemingly) nothing but good-willed advice between experts, masters, and beginners like me.
I don't have a model to post here. not yet at least.
I don't have any functional advice to share, for I am a beginner myself.
but I couldn't NOT point out the absolute bliss and pleasure it is to come across a forum like this. It's like I just came across a treasure in the depths of the ocean.
mega kudos to all of you.
It just makes me want to scream. - How The F*#% Do I Model This?
Thank you very much in advance