You just need to take your support loops a vert/edge further. When you terminate it too close to the edge of the geometry like that, you'll get some gnarly pinching. Don't delete the original green lines, as that is very necessary to maintain your shape. Also you could add another holding edge to further help sub-d your mesh (blue lines)
For general subdivision modeling: if a mesh holds its shape, is easy to work with and subdivides cleanly then it should be fine. One way to validate topology and mesh flow strategies is to test it out using subdivision preview and a reflective material. Rotate around the shape in the view port and look for smoothing artifacts like pinching and stretching.
Like Eric mentioned: it is possible to over optimize a mesh and that can result in smoothing artifacts like pinching in the corners. Here's an example of how this optimization strategy doesn't work well in one situation but is passable in another. Smoothing artifacts and stress patterns are clearly visible in the subdivision preview. Probably best to stick with the one that doesn't pinch and is easier to add loops to.
Working through different shapes and topology strategies is a great way to learn but be sure to compare each strategy and analyze where and how it breaks down. Flat surfaces are pretty resilient to triangles and n-gons but there is a limit to how much you can get away with near shape intersections. Another thing to consider when comparing strategies is the trade-offs between time, effort and results.
Maintaining clean edge flow and optimizing the mesh density are both good practices but there's also a time aspect to every project. That's where you have to decide what result is acceptable for the amount of effort required. For small parts on a background prop it's probably not worth the effort and time required to perfect and polish the mesh topology. As long as it subdivides and bakes cleanly then it's passable.
There's also different requirements for different types of projects. A high poly mesh for baking is going to have slightly different requirements from a high poly mesh for product visualization, etc. Check out the wireframes posted by Matthias Develtere on ArtStation. His recent work on Wolfenstien: Youngblood demonstrates one way of balancing production speed and quality while using only subdivision modeling.
FrankPolygon, Thanks for the tips. Tell me some tutorials like yours on the gumroad "Cylinder Intersections". It is desirable about the topology of different angles
Hey all, trying to figure out how to make harem pants similar to these. Baggy pants with excessive cloth in the crotch for maneuverability and elastic bands on the waist and ankles. Any help?
edit: After reading the thread title I haven't achieved any results of my own, I was just asking while it was a WIP. I presume that something like a cotton material would automatically create the stretchy cloth physics, but I'm not too sure about how to create the elastic bands, although for all intents and purposes they don't really need to stretch, they could just be elastic bands.
Hey guys, trying to get these 2 edges to have a more angular subdivision should i try creasing them or use support loops i tried using support loops and i get little pinching at the corners. Should I leave it as it is and just bake it to low poly ?
Hey guys, trying to get these 2 edges to have a more angular subdivision should i try creasing them or use support loops i tried using support loops and i get little pinching at the corners. Should I leave it as it is and just bake it to low poly?
Try selecting the entire loop (navy) and giving it a chamfer (yellow). This will be a complete support loop for that edge to be tightened up, however, it's possible you'll get an unwanted pinch/tight effect on the other end of that loop.
Right now you have some support loops that are terminating too soon, and also removing a critical edge in the middle which is resulting in that artifact
There is a bit of a flare out happening at the bottom of the mag well here. In the front, it's curving forwards, but in the back, it comes straight down before coming out into a rounded edge as well. I'm struggling to model it.
I don't know what operations need to be performed to get this to look right. The scale tool is struggling to make any of it look consistent and even around the beveled part.
I'm able to make it flat in the back, but if I do that, I can't get it to properly curve out how it does there.
I hope I don't interrupt anything here. I've a question / challange for all Maya diehards out there like myself.
Screenshot from an Instagram post from Gleb Alexandrov (standard Blender)
How would you model this exact transition in Maya?
I'm pretty experienced and I thought this would be pretty easy but I couldn't get this exact shape done. Tried smooth bridge from bevel to bevel with curve controls, no good result. I don't really have any other idea.
There's subtle and not so subtle differences between each manufacturer's designs. Some of the slight variations are down to things like how the blank was created, what tooling was used and how the order of operations was setup. The more obvious variations are usually down to aesthetic or functional design decisions. Locating additional reference images of this particular model and multiple close ups from different angles will help you resolve shape issues in areas that are difficult to model.
Also: diving in and working on details without blocking out the major shapes
will lead to headaches latter on in the modeling process. Take the time
to build up an accurate block out of the model. Use the additional reference
images to work through problem shapes and verify that all of the major
shapes are accurate. Once you have an accurate block out model everything will become much easier.
Here's an example of one strategy for adding the round overs and fillets. Start by blocking out the primary shapes. Add the vertical round overs along major shape borders using a chamfer or bevel operation. Use reference images to identify key features where round over width changes. Add the horizontal fillet around the top of the well using a chamfer or bevel operation. Note the variation in fillet width in the inside pocket. This is likely the result of two cutter sizes being used in two overlapping operations. Replicate this by using a smaller fillet width on the inside corner and blend it into the larger fillet width. Add the final round overs to the top and bottom perimeter of the well.
This modeling strategy can be adjusted to create whatever shape is
needed. The last two examples are a comparison of the fillet and chamfer
transitions that were generated using this process. The actual steps
and tools used (Booleans, modifiers, etc.) will depend on what works
best for the shapes and what fits the desired workflow.
Recap:
Gather additional reference images for shapes that are difficult to model.
Block out the major shapes and resolve problem areas early in the modeling process.
Add details and shape transition geometry after the block out stage is complete.
Time spent working through the block out process is paid back latter in the process. A good base mesh saves a lot of time since it can be used to create a high poly model with traditional subdivision modeling or by sending it into ZBrush for re-meshing, edge polish and detail sculpting. With some minor optimizations the base mesh also becomes the low poly model. Starting from a good base mesh saves a lot of time that would be spent on re-work.
Looks like the order of operations is important: Bevel the horizontal segments first. Dissolve and bridge where necessary. Bevel the vertical segments last to add the final curve.
There's subtle and not so subtle differences between each manufacturer's designs. Some of the slight variations are down to things like how the blank was created, what tooling was used and how the order of operations was setup. The more obvious variations are usually down to aesthetic or functional design decisions. Locating additional reference images of this particular model and multiple close ups from different angles will help you resolve shape issues in areas that are difficult to model.
Also: diving in and working on details without blocking out the major shapes
will lead to headaches latter on in the modeling process. Take the time
to build up an accurate block out of the model. Use the additional reference
images to work through problem shapes and verify that all of the major
shapes are accurate. Once you have an accurate block out model everything will become much easier.
Here's an example of one strategy for adding the round overs and fillets. Start by blocking out the primary shapes. Add the vertical round overs along major shape borders using a chamfer or bevel operation. Use reference images to identify key features where round over width changes. Add the horizontal fillet around the top of the well using a chamfer or bevel operation. Note the variation in fillet width in the inside pocket. This is likely the result of two cutter sizes being used in two overlapping operations. Replicate this by using a smaller fillet width on the inside corner and blend it into the larger fillet width. Add the final round overs to the top and bottom perimeter of the well.
This modeling strategy can be adjusted to create whatever shape is
needed. The last two examples are a comparison of the fillet and chamfer
transitions that were generated using this process. The actual steps
and tools used (Booleans, modifiers, etc.) will depend on what works
best for the shapes and what fits the desired workflow.
Recap:
Gather additional reference images for shapes that are difficult to model.
Block out the major shapes and resolve problem areas early in the modeling process.
Add details and shape transition geometry after the block out stage is complete.
Time spent working through the block out process is paid back latter in the process. A good base mesh saves a lot of time since it can be used to create a high poly model with traditional subdivision modeling or by sending it into ZBrush for re-meshing, edge polish and detail sculpting. With some minor optimizations the base mesh also becomes the low poly model. Starting from a good base mesh saves a lot of time that would be spent on re-work.
I was specifically trying to figure out a faster process of getting it to that slanted angle and having the faces extruded. The next parts of the model are not what I'm concerned about, because I know how to make it work. I'm looking for a simpler process for when I set the angle and add the extrusion, because right now, it warps the mesh and causes the vertices to be misaligned. I can fix that by hand, but I was wondering if there was a simpler operation.
For reference, here is the image again, and this is what I'm trying to remedy WITHOUT having to fix by hand.
A lot will depend on the available tool-set / plugins, preferred modeling operations and number of steps desired. Here's a few strategies for developing this shape with different tools. May need to mix and match to find something that fits your package and workflow. Definitely not the only way to do it so any ideas for using other tools, operations or plugins is worth sharing here.
Setting a precise angle:
Rotate the model to the desired angle and cut a horizontal edge loop across the shape.
Create a basic shape, rotate it to the desired angle and run a boolean difference on the main shape.
Use a skew / shear operation to generate the desired angle while keeping everything parallel.
Slice and other precision cutting operations are alternatives.
Extrude and scale without causing warping: Set the angle, add an edge loop, constrain subsequent operations to the top edge loop, extrude and scale up to adjust the width of the lip at the front and back, run an additional scale operation constrained to a single axis to adjust the width of the lip along the sides.
Using a uniform scale operation (with the origin set to the top loop) should prevent the mesh from warping (in most packages) but this first scale operation can only be used to adjust the width at the front and back. The second scale operation, constrained to a single axis, will push the sides out without effecting the rest of the shape.
Inset and skew / shear: Start with a flat bottom, add an edge loop, inset / extrude / solidify the edge loops to create the outer lip, use a skew / shear operation to push the geometry to the desired angle while maintaining parallelism. (Fastest but may not work in all packages.)
Rotate and extrude: Rotate the shape to the desired angle, cut in the angle, add an edge loop, extrude / inset / solidify the sides then extrude / inset / solidify the front and back. Rotate back into position.
Rotate and inset: Cut in the desired angle, add an edge loop, constrain the rotation to the verts in the top loop and rotate the front and back of the bottom loop forward the same amount as the initial slope, run an inset / solidify operation around the loop.
The rotation of the front and back of the bottom loop is what allows the inset / solidify / extrude operations to work without warping the mesh. This keeps everything parallel and co-planar but the rotation of the front and back needs to be taken out afterwords. This can be done by reversing the rotation process, by deleting the bottom loop and extruding the top loop down or by moving the bottom loop backwards to bring everything back to parallel. This can be a quick alternative to the shear setup but because of the extra steps it may not be ideal.
Rotate, scale, and extrude: Rotate to the desired angle, cut the angle in, extrude and scale the top of the lip in plane then rotate the shape back upright and extrude the lip downwards.
I'm trying to model the object seen in the pictures and I figured that modelling the intersection would be a smart move since it will define the number of edges needed for cylinder.
I modeled a part of it and added the supporting edge loops(no creasing) which would affect the cylinder curve and first tried matching cylinder sides with those of the intersection but it seemed to be too tedious and not that accurate. Had to use turbosmooth modifier as cylinder had to have 200+ sides.
My second thought was to expand the plane and see if I can do any optimization (don't mind the bad topology at the corner, it would be cut afterwards) on the edge count so the later joining process would be easier and it did seem to work although I'm not too sure as it would probably produce artifacts when turbosmoothed later.
I'm wondering if there's any better, more efficient or easier way of doing it, maybe I'm missing something.
If the mesh needs to be watertight then matching the segment counts is a good starting point. It looks like the plastic has a grainy texture so this could help hide
any minor smoothing errors. Sometimes close enough is good enough.
Instead of trying to match every support loop to a cylinder segment, try matching the cylinder's segment count to the major features without their support loops. Then use triangles or n-gons to catch the support loops where they run into the wall of the cylinder. This will require less geometry than trying to match every edge loop to a cylinder segment. Cut the support loops in the cylinder wall close to the shape intersection and try to limit the size of any smoothing errors.
Another (faster) option for a watertight high poly mesh would be a Boolean to ZBrush or QuadRemesher workflow.
If the mesh doesn't need to be watertight then another quick option would be to use floating geometry. Here's a comparison of subdivision modeling (with triangles and n-gons) and floating geometry.
Also Frank if you don't mind, just an aside. Want to ask about that idea for cataloging all your subd modeling tips-n-tricks into a centralized ongoing paid resource and perhaps putting it on Gumroad as a nice little earner fer ya. So any progress thus far and/or info on it?
This seems like a simple thing to do, but I am having some trouble finding a clean way to do it. I want this window piece on the helicopter door to be separate geometry for reasons.
The model is a prop seen from a distance, so it can be pretty low poly. Still I want to get the rounded corners on this window. The obvious approach is just smoothing the faces and then tweaking the vertices. But this isn't very precise and is kind of slow. Is there a better way I can do this?
Update:
Somethign that worked ok was to model the window piece separate. That way I can control the bevels on the edges. Then use quad draw and live surface to conform to the shape of the hull, and then booleaned them together.
This seems like a simple thing to do, but I am having some trouble finding a clean way to do it. I want this window piece on the helicopter door to be separate geometry for reasons.
The model is a prop seen from a distance, so it can be pretty low poly. Still I want to get the rounded corners on this window. The obvious approach is just smoothing the faces and then tweaking the vertices. But this isn't very precise and is kind of slow. Is there a better way I can do this?
Update:
Somethign that worked ok was to model the window piece separate. That way I can control the bevels on the edges. Then use quad draw and live surface to conform to the shape of the hull, and then booleaned them together.
Hey guys. I want to get this form, but I encountered difficulties. The first problem is with these corners. Doesn't go out to make the cylinder round I understand why this is happening, but I can't figure out how to make a corner. And the second one. If me do a chamfer here, then how to deal with the corners? I get some kind of porridge There is an assumption that I initially made an irregular shape of the model and now the chamfers do not work. Push me in the right direction
@99499 The shape and cut of the pattern will play a role in how the loose fabric falls. If you're working in Marvelous designer: Try cutting a larger pattern with extra fabric in the side of the legs and adjust the cloth settings so it's saggy. Use a smaller pattern for the cuffs and try adding elastic settings so they stretch around the leg and gather the loose fabric. Here's a discussion on the Marvelous forums about a similar style of pants: https://www.marvelousdesigner.com/community/forum/ShowoffandLearn/4593
@sacboi Still working on it behind the scenes. Currently working on a few hard surface modeling exercises that will include a guide and some example files. Should be available in the near future and I'll find the appropriate place to post about it when it's published.
@iacdxb Would you be able to post what your mesh looks like and where you're running into problems on it? That way we can provide more accurate feedback. There's also several discussions about similar mesh grilles further back in this thread so it might be worth taking a look at those.
@ApachEsH The cylinder needs a few more segments to support the intersecting geometry and the intersecting geometry should land between the cylinder segments instead of on them. Block out the shapes and make sure there's enough supporting geometry before adding the rest of the support loops.
@FrankPolygon , you are a magician, thank you very much. And in the first option N-gons is it really normal?
Try not to follow norms unless you understand why they're set. Some times people get issues because they put N-gons in places where N-gons don't belong (usually on curved hard-surface areas) and so they ask for help. Other people then point out the N-gon as the issue. Then the norm is formed: N-GON BAD!
If you zoom closely enough in on Frank's mesh and analyze that curve to an insane accuracy, you'll find that the N-gon altered the cylindrical curve a bit. For super-accurate-close-up renders I'd avoid putting that N-gon on the edge, but for a part on an animated model or a small detail, it's too small to ever be noticed. How many minutes did Frank save by just saying "fuck it, this is good enough"? More than 3? N-gon good.
Here's a model of a leaf where I needed the leaf stalk to be a 4-sided cylinder in order to match with an array further down the model. I used an N-gon's intrinsic qualities to my benefit where the leaf and stalk meet. Ngon good!
@sera3D May I ask why you need to model steps and walls into one continuous mesh? You don't need to take the support edges all the way through the flat surfaces, just terminate them.
sera3D: Context is important - what is the mesh supposed to be? Do you have a reference image? The reason why I ask is to establish whether you actually need this to be one continuous mesh.
Ok sorry if wasn't specific: this is the whole thing. It is mostly a personal exercise so it's not going anywhere, I am just trying to make an high-poly, quads only as much as possible. As you can see I need to fix other other parts of the mesh but I was stuck specifically on that section.
If you absolutely must make control loops manually (btw this approach became obsolete about ten years ago and is extremely inefficient), they should terminate locally, not wrap around the entire mesh that way
What approach is obsolete? Control loops? Terminate them locally is exactly what I was trying to do.
@sera3D May I ask why you need to model steps and walls into one continuous mesh? You don't need to take the support edges all the way through the flat surfaces, just terminate them.
crude example
Thank you for your input! However that's going to create a lot of n-gons.
Thank you for your input! However that's going to create a lot of n-gons.
Thank you guys!
The "n" in n-gon stands for "not-your-enemy" Your mesh actually is a good example why the "all quads" philosophy fails, of course you can stick with it if you like.
Thank you for your input! However that's going to create a lot of n-gons.
Thank you guys!
The "n" in n-gon stands for "not-your-enemy" Your mesh actually is a good example why the "all quads" philosophy fails, of course you can stick with it if you like.
I just thought having as many quads as possible it was good practice, that's all. I am not scared of triangles and n-gons
I was trying to model the following object and ran into some issues when it came to the transition from one shape to another. I've tried extending the rectangular shape's border, spherifying it and lowering down to create the rising curvature but that introduced lots of distortion. I also tried booleans but it kept removing pieces of geometry from the rectangular shape. Some people suggested matching the poly count of rectangle with the separate circle and just go around cutting each intersection manually. I do want to try modelling it as clean as possible meaning no floating geometry. I'd prefer using supporting edge loops instead of creases too. I'm open for any suggestiong of a better/more efficient way of joining those shapes together.
I do not believe Per was talking using creasing. I believe he is talking about the difference between smacking in manual loops and using chamfer/inset tools to do the heavy lifting.
@sera3D Hey buddy, read my post on this page with the leaves in it. N-gons are not your enemy. N-gons on uniform curved surfaces can be your enemy. On supported flat surfaces, N-gons flat out don't matter, and will enable you to save a bunch of modeling time and render-time by letting you end your supportloops in one e-z neato place.
Here are some of the Tris and N-gons highlighted: Note, the bottom is even worse, An unholy mess inspired by the mad ramblings of satan himself.
Without the ability to end my support loops on the most immediate flat surface, I would FIGURATIVELY go insane.
I'll do you one better, here's a marmo view model of all the non-quads on just what's in the renders (4540 triangles worth): http://skins.thanez.net/oz9/ngons/ I would take N-gons out on a romantic dinner date. NGON: FRIEND! I could go on but
@Povilas here's how to do any model in 10 ez steps: 1. Always nail down the major shapes first. Realize that THAT is not a cone, it is a dome-shape. When doing elongated extrusions like this, it's better to start with a quadsphere, as it gives you long nice edges to use as supporting geometry. You get quadspheres by turbosmoothing a cube primitive. This is my squashed quadsphere. There are many like it, but this one is mine.
2. When doing intersecting geometry, match up the amount of sides first so you have somewhere to put your edgeloops. Close. Enough.
3. Modify some of the loops (constricted to edge or face) to encompass your extrusion.
4. Extend the cylinder into an elongated boi past the mid-point. Press B in to Boolean. Delete 75% of your model, you don't need the extra work.
5. Add 3 new edges that end in nowhere good. Move them inward to create your ribs. Fix up the intersection between your long boi and the kinda-dome. No one will notice the triangles this close to an edge, HEHEHE
6. Clean up the top by adding the only two selected edges that aren't edgeloops. Then create edgeloops.
7. Select these edges and connect them either by 1 or 2, depending on how sharp you want them.
8. I went for 1 connection since I don't want the edges too sharp.
9. Two symmetry modifiers (guess which axises) and a turbosmooth later, and we have this.
10. Cut the edge off the squashed dome and make the sides to fit. I am drunk and tired. Good night, sirs.
If you absolutely must make control loops manually (btw this approach became obsolete about ten years ago and is extremely inefficient), they should terminate locally, not wrap around the entire mesh that way
Lmao not gonna lie I spent like 2 hours today manually placing support loops on a simple mesh, which took maybe a couple minutes at most to just clean up and put a chamfer mod on it. I've only used chamfer mod on simple geo because I don't understand the requirements for getting optimal results on complex geo.
Got a few questions guys: 1. What if the number of cylinder segments can't be increased to be smoother for chamfer mod? I understand you gotta put the intended hp segments on a cylinder but what if you've committed to a lower count initially, then the mesh gets more complex and you need more segments where that cylinder was?
OR your blockout mesh has hard committed chamfers, which you'd have for low poly use, but when you wanna make a hp out of that you can't with chamfer again, cause you'll need more segments. A lot of the times throwing down a turbosmooth ontop of a chamfer mod means you still need to plan where your loops are going to support it right?
2. Don't crucify me for this but am I stupid with topology right infront of me or putting chamfer mods on intersections like these makes it useless? It won't produce similar results to manual loops like on the right if I throw a turbosmooth on it in terms of concave pinching;
3. When you make your blockout, which is the preferred topology as below to support chamfer mod?
4. putting an inset by sg or chamfermod still requires manual cleanup if you have some geo that is too small in comparison to the rest of the uniform chamfer no? Meaning it's destructive unless you put an edit poly mod on top. Or you can adjust it by weight, but again that's not ideal?
You're right, thanks perna. I'll have to keep experimenting and testing with more meshes to understand what works and what doesn't. I already understand manual support loops and it's been a lazy and time-wasting workflow to fall back on due to my own shortcomings. Do you ever manually tweak your insets/loops after the mods are on or there is never any need too even with complex meshes you think?
I wish you'd teach because that'd be a class I'd definitely pay for A lot of people are wasting their time using outdated or uninformed methods like myself. There's not exactly a bible out there on current up to date methods with what is obsolete or not, and I have to look on polycount to find out I'm an old dinosaur.
I have this half sphere dial that I'm trying to make, with a bit sticking out of it. After adding bevel for support, and trying to clean it up. There is some pinching in there that I can't get rid of. I'm not sure if I simply need to have more polygons, or if there is a trick that I'm missing.
Yogev, this is a common and fundamental issue/pitfall. You are extruding off the existing grid, as if you have to. But you don't have to. I know it looks tempting, and seems to make sense, but you need to plan for the control geometry, not add it as an afterthought. Think instead of how you can make the existing geo serve as the control geo instead of serving as foundation for the extrude.
Try to go off the text only, and experiment with it. It's better if you put the pieces together yourself instead of having it spelled out. The principle I describe is very important to understand.
edit: as a concrete hint: imagine what your geo will look like if you inset first, then extrude.
Thank you for replying Perna. I would prefer to go off the text only if possible. Maybe it's because English is not my first language, but I'm not sure that I understand what you mean, or maybe I'm just not familiar with the terms. The existing grid is edge flow of the sphere? By "control geo", do you mean like parenting the objects?
I think I MIGHT have understood what what you meant about the insert. Not sure if it is what you meant, but I think it looks better now. Certainly something I should've thought about first.
On a side note- Did you host your live-stream on YouTube already?
Already broke his head himself. There is such a model I saw a video tutorial on how to do a similar topology, it looked something like this I get some nonsense if me make a corner like that
Kanni3d To be honest I didn't understand you. But a couple of minutes of experiments and came to this decision I just lacked a vertical rib, thank you for the tip There are no artifacts and it looks normal. And this is the main thing. Thanks again
Hey guys, I am wondering which way should be better to model these windows into the walls. The windows are curved, so I am not quite sure if should I build a plane around them and insert that into the wall, or should I add loops to wall itself to fit them in. What should I do ?
Replies
You just need to take your support loops a vert/edge further. When you terminate it too close to the edge of the geometry like that, you'll get some gnarly pinching. Don't delete the original green lines, as that is very necessary to maintain your shape. Also you could add another holding edge to further help sub-d your mesh (blue lines)
edit: After reading the thread title I haven't achieved any results of my own, I was just asking while it was a WIP. I presume that something like a cotton material would automatically create the stretchy cloth physics, but I'm not too sure about how to create the elastic bands, although for all intents and purposes they don't really need to stretch, they could just be elastic bands.
Right now you have some support loops that are terminating too soon, and also removing a critical edge in the middle which is resulting in that artifact
I don't know what operations need to be performed to get this to look right. The scale tool is struggling to make any of it look consistent and even around the beveled part.
I'm able to make it flat in the back, but if I do that, I can't get it to properly curve out how it does there.
I hope I don't interrupt anything here.
I've a question / challange for all Maya diehards out there like myself.
Screenshot from an Instagram post from Gleb Alexandrov (standard Blender)
I was specifically trying to figure out a faster process of getting it to that slanted angle and having the faces extruded. The next parts of the model are not what I'm concerned about, because I know how to make it work. I'm looking for a simpler process for when I set the angle and add the extrusion, because right now, it warps the mesh and causes the vertices to be misaligned. I can fix that by hand, but I was wondering if there was a simpler operation.
For reference, here is the image again, and this is what I'm trying to remedy WITHOUT having to fix by hand.
You nailed it! Didn't think about beveling the vertical edges.
Thank you
I modeled a part of it and added the supporting edge loops(no creasing) which would affect the cylinder curve and first tried matching cylinder sides with those of the intersection but it seemed to be too tedious and not that accurate. Had to use turbosmooth modifier as cylinder had to have 200+ sides.
My second thought was to expand the plane and see if I can do any optimization (don't mind the bad topology at the corner, it would be cut afterwards) on the edge count so the later joining process would be easier and it did seem to work although I'm not too sure as it would probably produce artifacts when turbosmoothed later.
I'm wondering if there's any better, more efficient or easier way of doing it, maybe I'm missing something.
Anyone can help me with some basic steps.... how to make these kind of holes…?
Thanks.
The first problem is with these corners. Doesn't go out to make the cylinder round
I understand why this is happening, but I can't figure out how to make a corner.
And the second one. If me do a chamfer here, then how to deal with the corners? I get some kind of porridge
There is an assumption that I initially made an irregular shape of the model and now the chamfers do not work. Push me in the right direction
Some times people get issues because they put N-gons in places where N-gons don't belong (usually on curved hard-surface areas) and so they ask for help. Other people then point out the N-gon as the issue. Then the norm is formed: N-GON BAD!
If you zoom closely enough in on Frank's mesh and analyze that curve to an insane accuracy, you'll find that the N-gon altered the cylindrical curve a bit. For super-accurate-close-up renders I'd avoid putting that N-gon on the edge, but for a part on an animated model or a small detail, it's too small to ever be noticed. How many minutes did Frank save by just saying "fuck it, this is good enough"? More than 3? N-gon good.
Here's a model of a leaf where I needed the leaf stalk to be a 4-sided cylinder in order to match with an array further down the model. I used an N-gon's intrinsic qualities to my benefit where the leaf and stalk meet.
Ngon good!
I would like some help on adding supporting edges on this shape (the 'steps') :
This is my first attempt, of course other than being messy it affects the shape above which is not good:
I am stuck trying to break the topology of the steps in order not to go over the entire mesh.
Thank you for your help.
You don't need to take the support edges all the way through the flat surfaces, just terminate them.
crude example
What approach is obsolete? Control loops? Terminate them locally is exactly what I was trying to do.
Thank you for your input! However that's going to create a lot of n-gons.
Thank you guys!
Your mesh actually is a good example why the "all quads" philosophy fails, of course you can stick with it if you like.
I've tried extending the rectangular shape's border, spherifying it and lowering down to create the rising curvature but that introduced lots of distortion. I also tried booleans but it kept removing pieces of geometry from the rectangular shape. Some people suggested matching the poly count of rectangle with the separate circle and just go around cutting each intersection manually.
I do want to try modelling it as clean as possible meaning no floating geometry. I'd prefer using supporting edge loops instead of creases too. I'm open for any suggestiong of a better/more efficient way of joining those shapes together.
Here are some of the Tris and N-gons highlighted:
Note, the bottom is even worse, An unholy mess inspired by the mad ramblings of satan himself.
Without the ability to end my support loops on the most immediate flat surface, I would FIGURATIVELY go insane.
I'll do you one better, here's a marmo view model of all the non-quads on just what's in the renders (4540 triangles worth): http://skins.thanez.net/oz9/ngons/
I would take N-gons out on a romantic dinner date. NGON: FRIEND!
@Povilas here's how to do any model in 10 ez steps:
1. Always nail down the major shapes first. Realize that THAT is not a cone, it is a dome-shape.
When doing elongated extrusions like this, it's better to start with a quadsphere, as it gives you long nice edges to use as supporting geometry. You get quadspheres by turbosmoothing a cube primitive.
This is my squashed quadsphere. There are many like it, but this one is mine.
2. When doing intersecting geometry, match up the amount of sides first so you have somewhere to put your edgeloops.
Close. Enough.
3. Modify some of the loops (constricted to edge or face) to encompass your extrusion.
4. Extend the cylinder into an elongated boi past the mid-point. Press B in to Boolean. Delete 75% of your model, you don't need the extra work.
5. Add 3 new edges that end in nowhere good. Move them inward to create your ribs. Fix up the intersection between your long boi and the kinda-dome. No one will notice the triangles this close to an edge, HEHEHE
6. Clean up the top by adding the only two selected edges that aren't edgeloops. Then create edgeloops.
7. Select these edges and connect them either by 1 or 2, depending on how sharp you want them.
8. I went for 1 connection since I don't want the edges too sharp.
9. Two symmetry modifiers (guess which axises) and a turbosmooth later, and we have this.
10. Cut the edge off the squashed dome and make the sides to fit.
I am drunk and tired. Good night, sirs.
Got a few questions guys:
1. What if the number of cylinder segments can't be increased to be smoother for chamfer mod? I understand you gotta put the intended hp segments on a cylinder but what if you've committed to a lower count initially, then the mesh gets more complex and you need more segments where that cylinder was?
OR your blockout mesh has hard committed chamfers, which you'd have for low poly use, but when you wanna make a hp out of that you can't with chamfer again, cause you'll need more segments. A lot of the times throwing down a turbosmooth ontop of a chamfer mod means you still need to plan where your loops are going to support it right?
2. Don't crucify me for this but am I stupid with topology right infront of me or putting chamfer mods on intersections like these makes it useless? It won't produce similar results to manual loops like on the right if I throw a turbosmooth on it in terms of concave pinching;
3. When you make your blockout, which is the preferred topology as below to support chamfer mod?
4. putting an inset by sg or chamfermod still requires manual cleanup if you have some geo that is too small in comparison to the rest of the uniform chamfer no? Meaning it's destructive unless you put an edit poly mod on top. Or you can adjust it by weight, but again that's not ideal?
I wish you'd teach because that'd be a class I'd definitely pay for
This made my day!
I would prefer to go off the text only if possible. Maybe it's because English is not my first language, but I'm not sure that I understand what you mean, or maybe I'm just not familiar with the terms. The existing grid is edge flow of the sphere? By "control geo", do you mean like parenting the objects?
I think I MIGHT have understood what what you meant about the insert. Not sure if it is what you meant, but I think it looks better now. Certainly something I should've thought about first.
On a side note- Did you host your live-stream on YouTube already?
I saw a video tutorial on how to do a similar topology, it looked something like this
I get some nonsense if me make a corner like that
Try this, missing some critical edges to maintain that detail.
I just lacked a vertical rib, thank you for the tip
There are no artifacts and it looks normal. And this is the main thing. Thanks again