This seems like a simple shape, I would just model it flat and then apply a bend modifier on top of all modifiers. This is a feasible scheme, thank you, but I would like to know whether it can be made under the condition of already bending?
At a glance, it seems as if your model is based off a US design with tapered hydrant outlets.
So here's an optional simplified solution too your topology query plus keeping in mind when implementing a mostly manual editing approach, it's acknowledged best practice to use minimal geometry, otherwise higher complex polycounts equate to less control which usually results in mesh errors and in turn shading artifacts.
Workflow
- Cylinder segment counts:
Main body - 20
Large outlet - 16
Dual lateral outlets - 12
- Firstly, centrally locate the secondary shapes accordingly on the main cylinder via their axis whilst also taking care to match the segments close as possible. Then run boolean operations for each, to cut multiple relatively shaped intersecting voids.
- Next, cleanup should be a straightforward process since disparate segment ratio's for each cylinder, will allow near enough accurate alignment. Nonetheless stray vertices are typical when free forming a given model so a bit of welding may result.
- Lastly, subdvide and shade smooth alongside adding control / support loops where appropriate to provide soft transitional edge flow.
I've been trying to make this spiral/ helix thing, and I'm not really sure what would be the proper way to do it. Creating a spiral is not too difficult, but the problem is with the start/end parts of it. The reference I have for it doesn't seem to have a gap between the start of the spiral and the cube it needs to connect to., and I'm not quite sure how to achieve that.
I've been trying to make this spiral/ helix thing, and I'm not really sure what would be the proper way to do it. Creating a spiral is not too difficult, but the problem is with the start/end parts of it. The reference I have for it doesn't seem to have a gap between the start of the spiral and the cube it needs to connect to., and I'm not quite sure how to achieve that.
another trick for you.
1.a plane with segments
2.ffd2 modifier, to move the plane one square up on the right side.
and judging from this thread my polys are waaaaaaaaaaaay too low hence why some details appear to be missing from my meshes
This is a great thread Guess i should not be afraid to push it and this is the trap i keep on falling into
This is an example of what i want to create it's namely these side panels plus the door is where where i get stuck at but judging from this thread it appears for vehicles low poly is not the way to go it seems.
Now i am beginning to understand why high poly is absolutely required for some items
and judging from this thread my polys are waaaaaaaaaaaay too low hence why some details appear to be missing from my meshes
This is a great thread Guess i should not be afraid to push it and this is the trap i keep on falling into
This is an example of what i want to create it's namely these side panels plus the door is where where i get stuck at but judging from this thread it appears for vehicles low poly is not the way to go it seems.
Now i am beginning to understand why high poly is absolutely required for some items
show us what you've done and where you are stuck mate?
and judging from this thread my polys are waaaaaaaaaaaay too low hence why some details appear to be missing from my meshes
This is a great thread Guess i should not be afraid to push it and this is the trap i keep on falling into
This is an example of what i want to create it's namely these side panels plus the door is where where i get stuck at but judging from this thread it appears for vehicles low poly is not the way to go it seems.
Now i am beginning to understand why high poly is absolutely required for some items
show us what you've done and where you are stuck mate?
Basically an example of my problem is this crown vic i started working on as it's a less complex object
I cant seem to figure out how define the separations within the body panels but i can get the rough block out done really good also my edges as you can see can become really wonky as well
Essentially what ever i make usually comes out looking like a Stock car body as a monocoque chassies (try to imagine how a nascar stock car body looks and well you get the picture)
and judging from this thread my polys are waaaaaaaaaaaay too low hence why some details appear to be missing from my meshes
This is a great thread Guess i should not be afraid to push it and this is the trap i keep on falling into
This is an example of what i want to create it's namely these side panels plus the door is where where i get stuck at but judging from this thread it appears for vehicles low poly is not the way to go it seems.
Now i am beginning to understand why high poly is absolutely required for some items
show us what you've done and where you are stuck mate?
Basically an example of my problem is this crown vic i started working on as it's a less complex object
I cant seem to figure out how define the separations within the body panels but i can get the rough block out done really good also my edges as you can see can become really wonky as well
Essentially what ever i make usually comes out looking like a Stock car body as a monocoque chassies (try to imagine how a nascar stock car body looks and well you get the picture)
So a rule of thumb is always to try to build separated parts separated. The car you are also modeling is separated, what you have in your blockout is basically one mesh. No car is build like that. Unless you go really low poly. Then I could see that it works. https://www.youtube.com/watch?v=NEDs5EFyVVk <---
Thanks for your tips wirrexx I saw a video of a guy modeling a tesla with the sub surface modifire i was impressed with the difference it made so i might try that next
he also used edge flow techniques to make the creases as well as booleans i thought that was neat too.
This has also convinced me to finally switch over to the latest version of blender it seems much easier to line your references up now as the old way of doing it just seems to distort them heavily.
Not sure if it's ok to just ask questions in this thread, but I'll give it a go. I'm trying to get into hard-surface modelling(mainly weapons) and I'm really struggling with it. Anything from getting the grip correct, scaling or details is just really hard to get right. Is there any tips for how to approach these types of complex shapes?
I have been checking out both ChamferZone and Eugene, but still have some problems with above.
Yeah, asking questions is fine but posting what you've generated thus far so people are able to offer more concise advice or even suggest particular workflow/s that might potentially be effective streamlining a certain aspect in your modeling process, would be much better
Not sure if it's ok to just ask questions in this thread, but I'll give it a go. I'm trying to get into hard-surface modelling(mainly weapons) and I'm really struggling with it. Anything from getting the grip correct, scaling or details is just really hard to get right. Is there any tips for how to approach these types of complex shapes?
I have been checking out both ChamferZone and Eugene, but still have some problems with above.
Hey friendo. In general, it's a good idea to gather as many reference images you can and really study the shapes of whatever you're modeling. Analyze how each part was made. If you understand the manufacturing process, most of the shapes model themselves. Some pieces will have to be made only by following the refs, so there'll be a bunch of guesswork, but some pieces, like the slide of a Glock 17 is easy to find the outer dimensions of. If you start with that which you know the dimensions of, you can line the rest up to fit. Block out everything and check widths and lengths to see that you're happy with it before starting to add compound curves and shit.
If you're REALLY struggling, I would suggest that maybe you've started at a too high difficulty level. If you're completely new to hard-surface stuff, there's a buuuuuuuuuuuuuuuuuuuuunch of stuff that you won't learn until you make a mistake and get to correct it. If you start with a very detailed piece, it's gonna give you all of the challenges at once and that will feel overwhelming. I suggest you start with something extremely simple, like a throwing knife. Model, bake and texture it and ask for help once you've tried and failed at something. When you're done, ask for crits in your own thread. Fix the stuff that's broken, and then step up to a fork. Then a shovel. Then a serrated knife. Then a pistol.
As for learning, the best tool at your disposal is this or any other large community. If you want thought out responses like this one, you're gonna have to put in the work first. This one is a freebie. If you spend 5 or 0 minutes modeling before asking for help, people will spend the same amount of time helping you. They have better things to do than try to help someone who hasn't shown they've put in the time yet.
Hi, I want to create a model of an old hammer in Blender and then texture it in Substance Painter. I'm learning to create game assets and aiming for realism. The model of the hammer is mostly ready. The hammer which I'm using as a reference has text and logo engraved on its head and I'm thinking how can I transfer this text and logo onto my model. Here it is:
It could've been a bit easier if there was only text, then I could find a similar font and put it on the height channel in Substance Painter. In fact I already did it, but I can't say that I'm happy with the result. Here it is:
Anyway, I also want to add a logo as on reference photo, so probably it's better to make text in some different way so I could reproduce the original font manually. So the question is how would you approach this problem? Maybe I should create an alpha (not sure if the terminology is right, but hope you understand what I mean) based on the photo and then use it as a brush in sculpting software or in Substance Painter?
If you know more than one way of doing that, for example "workflow to achieve best visual result" or "fastest workflow", I would love to hear about all of them 🙂 If you have somewhere in bookmarks a tutorial that will help me I would appreciate it.
How do i go about modeling this curve on the door?
My attempt
I've tried a few times but still cant get it quite right. What is also quite a challenge is how at the top the curve sort of fades out and reappears again.
You people are straight up magicians in here. I am way out of my league with what I am trying to do, but since I usually just pay others to model this stuff for me to use by my company, I wanted to be able to do it myself as well and understand how modeling works and what it takes.
Now I have made several things, but my topology is absolutely horrible and can be done better by a toddler, but I do have some liberties. There will be virtually no close-ups and if it looks right, it's good enough.
Anyway, I am (trying) to model a Husqvarna 701/KTM 690 SMC-R engine to be used in renders, but it's just way too complicated for me to get close to a good result. The entire cilinder is almost one huge intricate casting and I have no idea how to tackle it. I just make shapes and boolean them together and try to clean up as much as I can, but it's becoming a mess and i'm just not getting anywhere near to a good result. Any idea on how to tackle it, or what modeling techniques I should utilise to make this? Especially the part in the center of the cilinder head where the tube sticks out.
I did find a 3D scan of the same engine (with a few small differences), but it's not making it any easier for me. I have made another engine previously for the KTM 390, but for some reason that one was way simpler for me.
How do i go about modeling this curve on the door?
I've tried a few times but still cant get it quite right. What is also quite a challenge is how at the top the curve sort of fades out and reappears again.
Fairly simple, just requires maintaining uniform edge spacing and sufficient geometry to provide a consistent smooth transitioning edge flow:
Workflow
- Sketch out the shape using vertex - edge slide mesh tools
- Delete applicable faces that form the vent aperture / opening
- To generate the longitudinal creased curvature for the intake shroud, utilize pivot point active element too shear the graduated shape, then add required control loops for the panel cuts
You people are straight up magicians in here. I am way out of my league with what I am trying to do, but since I usually just pay others to model this stuff for me to use by my company, I wanted to be able to do it myself as well and understand how modeling works and what it takes.
Now I have made several things, but my topology is absolutely horrible and can be done better by a toddler, but I do have some liberties. There will be virtually no close-ups and if it looks right, it's good enough.
Anyway, I am (trying) to model a Husqvarna 701/KTM 690 SMC-R engine to be used in renders, but it's just way too complicated for me to get close to a good result. The entire cilinder is almost one huge intricate casting and I have no idea how to tackle it. I just make shapes and boolean them together and try to clean up as much as I can, but it's becoming a mess and i'm just not getting anywhere near to a good result. Any idea on how to tackle it, or what modeling techniques I should utilise to make this? Especially the part in the center of the cilinder head where the tube sticks out.
I did find a 3D scan of the same engine (with a few small differences), but it's not making it any easier for me. I have made another engine previously for the KTM 390, but for some reason that one was way simpler for me.
*pics*
What exactly are you having issues with? The only thing I can see is that some unions are a bit sharp. This looks perfectly passable if you ask me. Maybe you'd have a better time if you planned out ALL the booleans before booleaning. That'll make cleanup easier, as you would have planned how many sides you want each cylinder to make the geometry match as well as it can.
Hey all! Was wondering if I could get some eyes on a project I'm noodling on. Goal is to model irl model kit pieces in Cinema 4D/Zbrush/Blender for some hardsurface practice. Am a relative newbie at this and am wondering what are some proper ways to go about modeling the following piece (top half only, no orange or bottom "jaw"). As you can see, I kind of hack-n-slashed my way through it using three splines, a loft, some symmetry and good ol' box modeling but my topology is bad and not edit friendly. Was wondering if there were better ways to go about this? Thanks! (^ Splines used in a loft to make base geo ^) (^ 360 Topology ^) (^ Current Render Results ^)
Hey all! Was wondering if I could get some eyes on a project I'm noodling on...
Hi. I'm a noob myself but maybe this will be helpful.
This should be quite similar to the shape you presented above, but with a bit cleaner topology. I've made it in Blender with help of the Bevel modifier. Here is a .blend file if you want to mess around with it.
The topology is not "all quads", but probably it's good enough, and I guess trying to make it all quads can make topology unreasonably more complex.
Hey all! Was wondering if I could get some eyes on a project I'm noodling on...
Hi. I'm a noob myself but maybe this will be helpful.
This should be quite similar to the shape you presented above, but with a bit cleaner topology. I've made it in Blender with help of the Bevel modifier. Here is a .blend file if you want to mess around with it.
The topology is not "all quads", but probably it's good enough, and I guess trying to make it all quads can make topology unreasonably more complex.
Much cleaner, thank you! Your method is much simpler, making the boxier form first and just beveling from there. I think I was getting hung up on the curve of the front portion and the bevel that transitions into the hard edge. No real need for all quads here, it's not going to be animated. Can absolutely work it from here, thank you again!!
Hey everyone, I'm having a really hard time figuring out how to model this curve on this PC case, It's an Alienware Aurora Ryzen R10 series PC Case. I'm at a loss for solutions on solving this compound curve underneath the front. It actually curves twice, not very familiar with tackling this challenge.
heres some reference:
I can't seem to get good shading around this area:
It's been frustrating me so I scrapped that attempt (mostly cause the side view isn't aligned with the front) and started a new model just using the front and on my second attempt I couldn't figure it out either, any help would be super!
heres the second attempt (different approach) It's got a sub smooth modifier
I modeled your object a little bit, the proportions are incorrect but I hope you find it useful.The basemesh was made from a box primitive. The polycount is kept as low as possible for easier editing. Most of the control loops are generated from a modifier equivalent to the bevel modifier in blender
I modeled your object a little bit, the proportions are incorrect but I hope you find it useful.The basemesh was made from a box primitive. The polycount is kept as low as possible for easier editing. Most of the control loops are generated from a modifier equivalent to the bevel modifier in blender
Hi everybody. I need to make hole on this shape. The density of vertices is uneven which makes difficulties for making holes. I'm creating verticies by dividing polygons, but that makes artifacts. Is there any suggestion for this problem ? Do i need more topology, or this is enough?
I`m trying to replicate the dots on this pipe. My attempt looks like this, but the final shape is more square-ish and there is a bit of extrusion at the front as demonstrated by this line. How do I do a better job?
@IronLover64 You need to make more edgeloops especially at the "corner" of your holes. Something like this, there are many tutorials and videos explaining this workflow. I suggest starting by looking up "Arrimus3D" on youtube.
EDIT; For the subdivision workflow you're going for, your base cylinder has way too few subdivision to begine with. I suggest starting over with a more dense cylinder (not too dense, maybe just double it), because if you're going to model detail into it later, you're going to run into problems, especially shading and smoothing artefacts.
@IronLover64 You need to make more edgeloops especially at the "corner" of your holes. Something like this, there are many tutorials and videos explaining this workflow. I suggest starting by looking up "Arrimus3D" on youtube.
EDIT; For the subdivision workflow you're going for, your base cylinder has way too few subdivision to begine with. I suggest starting over with a more dense cylinder (not too dense, maybe just double it), because if you're going to model detail into it later, you're going to run into problems, especially shading and smoothing artefacts.
Thanks. I followed the way Arrimus3D did it, but I still have the problems with the shape when seen from above or below as shown with this line.
Catmull–Clark subdivision averages between existing points and this creates a smoothing effect. Uncontrolled subdivision smoothing tends to deform shapes and shape transitions that lack supporting geometry.
Creating shapes that have crisp edges and minimal smoothing deformation often requires adding support loops on both sides of the edges that define shape profiles and shape transitions. Placing support loops on both sides of the defining edges will constrain the area that's being averaged out by the subdivision which tends to also reduce the associated smoothing effect.
Supporting geometry isn't limited to just support loops either. The distance between the existing geometry in the merged shapes also has an effect on subdivision smoothing. Increasing the distance between the segments (reducing geometry density) tends to amplify the smoothing effect while reducing the distance between the segments (increasing the geometry density) tends to reduce the the smoothing effect.
Increasing the amount of geometry in the starting primitives can make it easier to accurately merge the shapes and does tend to increase the quality of the subdivided mesh but arbitrarily adding more geometry to existing mesh doesn't guarantee an increase in overall shape accuracy.
This is why it's generally considered best practice to match the segments of the intersecting shapes whenever possible. Matching the segments of intersecting shapes tends to produce a mesh with a more consistent geometry distribution and makes it easier to maintain the accuracy of the basic shapes. More consistent geometry density will also tend to make it easier to use the minimum amount of geometry required to hold the shapes. Which helps reduce unnecessary complexity and makes the base mesh easier to work with.
Below is an example of what this process could look like for this particular shape. Start by evaluating the overall size of the object and the relative size of the smallest surface feature then select an appropriate number of segments for the smallest intersecting shape.
Use primitives to preview the base of the shape intersection. Since the larger cylinder isn't constrained by adjacent geometry and the cylinder wall segments will be disrupted by the smaller shape it makes sense to adjust the number of segments in the larger cylinder to match the existing geometry in the smaller cylinder. Keeping these cylinder wall edges parallel to each other and concentric with the rest of the large cylinder's geometry will help preserve shape accuracy and reduce unwanted smoothing deformation artifacts.
Perpendicular loops can be added to the cylinder walls to match the rest of the intersecting geometry. The segment matching doesn't have to be perfect. It just needs to be close enough that any difference between the two shapes can be taken up between the support loops around the shape intersection.
There's a number of different ways to approach merging the shapes and adding the support loops but if the goal is to maintain shape accuracy then try to avoid disrupting the position of the cylinder wall segments. It's also important to maintain relatively consistent segment spacing on the intersecting shape but if merging stray geometry requires making a compromise it's generally better to preserve the shape of whichever feature requires the greater degree of shape accuracy.
While there are edge cases where it makes sense to only add a support loop on one side of a shape intersection it's generally better to have support loops on both the outside and inside of most shape intersections. Placing support loops on both sides of the edges that define shape transitions will better define the area that's used to average out the smoothing and this helps constrain any deformation artifacts to smaller area.
These same basic concepts apply to shape intersections with tapered sides and complex compound curves.
There may also be situations where the size of the object, geometry from adjacent shapes or some other constraint makes adjusting the number of the segments in the shapes impractical. When using less than optimal amounts of starting geometry the next best option is to move or scale the geometry components along the surface of the the larger cylinder to counter act the edge stress caused by the subdivision smoothing.
The results won't be as accurate as segment matching but when done correctly there shouldn't be any major smoothing artifacts and the result is often better than omitting minor details or having to completely rebuild major sections of a model late in the modeling process.
Here is a comparison of the previous examples. Matching the geometry of the intersecting shape by increasing the number of segments in the larger cylinder does improve the quality of the surface but resolving the shape accuracy issues at a lower level of complexity and adding support loops to both sides of the shape intersection also help control the subdivision smoothing behavior.
Adjusting the position of over stressed geometry to counter the deformation caused by subdivision smoothing also produces usable results but does open up the possibility of inadvertently creating other types of artifacts and is really only a good option when segment matching isn't viable.
Segment matching is an important part of subdivision modeling and one of the keys to getting consistent results with complex shape intersections on curved surfaces. It's also important to preserve the attributes of the intersecting shapes and try to take up any differences between the intersecting geometry in the support loop around the outside of the shape intersection. This helps preserve the accuracy of the underlying shapes and will generally constrain any smoothing issues to a relative small area within the support loop.
Below is an example of how these concepts can be applied when matching the segments of the smaller intersecting shapes to the existing geometry of the larger shape.
Resolving smoothing artifacts caused by adding
surface features to a curved mesh is an often discussed topic so there's
a lot of prior art in this thread an elsewhere on the forums. The
previous page has a couple of good discussions about other methods of
adding holes to curved cylinders and there's been a recent discussion
about holes in curved surfaces and mesh complexity.
https://polycount.com/discussion/comment/2755033/#Comment_2755033
It's also worth mentioning that subdivision modeling is an inherently approximate process. The smoothing algorithm itself is averaging things out and behaves in a fairly consistent and predictable manner. It's art not purely mathematical CAD modeling. When it comes to using overly dense or complex base
meshes there's a definite point where the marginal returns start to have
steep time costs.
So there's always going to be tradeoffs between shape accuracy and modeling efficiency. This is why it's generally considered best practice to evaluate the overall size of the object, relative to the general view distance, and choose an appropriate level of mesh density based on that. Sometimes getting the desired results is about creating better geometry rather than using more geometry. https://polycount.com/discussion/comment/2746328/#Comment_2746328
Recap:
Match the edge segments around shape intersections. Use existing geometry as support for intersecting shapes whenever possible. Reduce unnecessary complexity by using the minimum amount of geometry required to hold the shapes.
Hello, guys. Sorry my English, it is so so. I studying blender, and now i working on my firt big project. I am learning to create models for game, and i faced a problem. While crearing high poly of machine gun, i received artifacts. I can not fix this and i need your help, pleeease. obj - https://disk.yandex.ua/d/TgHdsRvOLhulmg
This thread has a lot of great resources so it's probably worth taking some time to skim through the recent posts and look for examples of how other artists have solved similar problems on other shapes.
A couple of posts up there's a recent write up that covers the basics of cutting holes in curved shapes without creating noticeable smoothing artifacts. The same basic concepts can be applied to these rectangular slots on the side of the cylinder.
A few pages back there's also a couple of other posts that cover different approaches to similar topology issues.
There's a few different ways to approach this problem so it makes sense to start with the easiest solution first.
The current edge bevel weight layout stops at the corners of the intersecting shapes and this means there's no path for the new edges to run out across the rest of the mesh. This forces the bevel modifier to merge the new edges in the corner of the shape, which deforms the side of the cylinder and the corners of the rectangular cut out.
Adding bevel weights to the loops directly behind the rectangular cut out provides a path for the new edges to run out across the rest of the mesh and connect to the adjacent shapes. This improved topology routing should solve most of the smoothing issues around the corners of the rectangular cut outs but may cause minor smoothing issues where the new edges run parallel to the edge segments that make up the walls of the cylinder.
Provided there's enough geometry density, the new edges that run parallel between the existing segments of the cylinder shouldn't cause any major smoothing issues. If the new edges do cause minor smoothing issues then it may be necessary to slide them along the surface of the cylinder to relive the smoothing tension caused by the subdivision.
It is worth noting that this kind of adjustment isn't entirely optimal since it can contribute to shape inaccuracy and potentially cause other types of smoothing artifacts. While there are some situations where it makes sense to manually adjust the supporting geometry, in most cases it's generally considered best practice to use an appropriate amount of geometry in the base mesh to accurately hold the shapes.
Below is a comparison of two different edge weight paths used to control the support loops generated by the bevel modifier.
Another strategy would be to use the existing geometry of the cylinder as part of the outer support loop around the shape intersection. In most cases this can be achieved by offsetting the intersecting geometry or using inset operations to create the inside profiles of the cut outs. When joining or subtracting shapes from curved surfaces it's important to match the segments of the intersecting shapes and provide some space between the existing geometry for the support loops to take up any difference between the geometry that defines each surface feature. This topic is covered in depth a few posts up in the thread.
Here's an example of what this process could look like when cutting a rectangular shape out of a larger cylinder. Start by adding the support loops on either side of the intersecting shape. Select the interior faces that define the area of the rectangular cut out and run an inset operation. Dissolve the interior geometry to create the rectangular cut out. Add the perpendicular support loops near the corner of the shape intersection then add the support loops on the inside of the rectangular cut out.
Either approach will work, when using the appropriate amount of geometry, on both solid and hollow shapes.
Recap:
Provide a clear route for support loops to run around the shape and terminate unnecessary support loops in an area that won't disrupt the underlying topology. When merging shapes with curved surfaces it's generally going to be a good idea to match the segments and provide some space between the existing geometry elements to act as a support loop and take up any difference between the shapes.
@FrankPolygon Hello, many thanks for your help. I wouldn't think about it myself... I did it according to the first option, but i got this "stretch marks". Can i fix this or is it acceptable for this element?
UPD: so, I just need to add more faces to the cylinder?
@rudenko_je Deciding whether or not certain smoothing artifacts are acceptable really depends on a couple of different factors:
Are the smoothing artifacts on a part of the model that's visible to the player under normal game play conditions?
Are the smoothing artifacts significant enough to still be noticeable when high frequency texture details are added to the textures?
If the answer to either of those questions is yes then it's probably worth trying to resolve the smoothing issues. Whether or not it makes sense to try adjusting the existing mesh or completely rebuilding the model really depends on the project's quality standards and time budget.
Assuming that this part of the model is further away
from but will still spend a significant amount of time directly in the
player's field of view: increasing the number of segments in the
cylinder to match the size of support loops required to hold the desired
level of corner sharpness is probably a good starting point.
However... Arbitrarily increasing the amount of geometry in the primitives does tend to improve the overall surface quality of a model but after a certain point there's diminishing returns and a steep fall off in editability. There's also some subtle differences in surface quality that makes choosing a specific topology layout important when there's a limited amount of geometry to work with.
Below is a comparison of 16 and 24 segment cylinders with both overlapping and offset shape intersections. There tends to be only minor differences between the different topology strategies at lower geometry densities but as the width of the curve's segments matches the width of the support loops on the intersecting shapes the difference in smoothing quality becomes more apparent.
Creating intersecting shapes directly on the existing geometry of curved surfaces is fairly intuitive because it's quick and easy but does tend to distort the mesh by either directly displacing the underlying geometry or by disrupting the segment spacing along the curved surface. Below is comparison that shows how shape intersection and support loop placement can effect shape accuracy and subdivision smoothing. Increasing the geometry density of the basic shapes does counteract some of this effect but doesn't completely resolve it.
Offsetting shape intersections, using existing geometry as support and matching the segments of intersecting shapes all contribute to producing well optimized topology that subdivides cleanly. Below is a comparison that shows how even a slight increase in mesh density improves the overall surface quality of offset shape intersections.
There are certain situations where double looping an edge makes sense but it isn't always necessary and can sometimes introduce it's own type of shape distortion and smoothing artifacts. Using less shape offset or narrower edge widths on the initial geometry tends to provide better results with less complexity.
Here's an example that shows how double looping the edge of a lower density mesh tends to sharpen the inside of the shape intersection while also introducing some pinching artifacts neat the outside corners. As the mesh density increases the artifacts in the corners become less of an issue but are still partially visible.
When working with a limited amount of starting geometry it's generally possible to make some adjustments to the mesh that will partially reduce but not completely resolve the visibility of certain types of smoothing artifacts. As mentioned in the previous post, this isn't the most ideal way to correct smoothing issues but it can be a usable workaround in less then ideal situations where the geometry density is constrained by the adjacent shapes.
In this example: While it's certainly possible to move some of the geometry around and make some minor improvements, the support loop topology is causing enough problems for the subdivision smoothing that it just makes more sense to start over with a different approach.
In this example: The pinching artifacts in the corners have been resolved by providing a path for the perpendicular support loops but the placement of the shape intersection causes the parallel support loops to disrupt the underlying shape and the segment spacing on the cylinder. This generates it's own type of smoothing artifact that runs out and into the adjacent segment.
Applying the bevel modifier and spreading out the perpendicular edge loops does help soften the harshness of the mesh deformation but doesn't do much to resolve the underlying artifact. Manually adjusting the position of the highlighted edges does relive most of the smoothing stress that's generating the smoothing artifacts but doesn't resolve them completely.
In this example: Using offset geometry produces a smaller smoothing artifact that can be softened by spreading out the trailing support loops. Moving the highlighted edges down and into the subtracted shape does resolve most of the smoothing issues on the outside of the cylinder but ends up creating some minor shape deformation and softness on the inside of the shape transition. With a bit more experimentation it should be possible to find a workable balance between the shapes.
In this example: Increasing the geometry density (From 16 segments to 24 segments.) resolves most of the visible smoothing artifacts from all but the steepest glancing angles. Support loops are all generated automatically using a bevel modifier that's controlled by edge weights.
Applying the bevel modifier and spreading out the perpendicular loops provides a bit more room for the subdivision smoothing to soften the artifacts near the corner of the shapes. Manually adjusting the height and position of the highlighted edges pretty much resolves the smoothing artifacts without disrupting too much of the underlying shapes.
For most objects with low gloss materials this is probably an
acceptable compromise between modeling efficiency, shape accuracy and
surface quality.
In this example: Increasing the geometry density of the cylinder (From 16 segments to 24 segments.) and using offset shape intersections produces the cleanest result so far. Spreading out the end of the perpendicular support loops helps soften the edges of the smoothing artifacts and moving the highlighted edges down / inwards a little pretty much resolves all of the visible smoothing artifacts without producing noticeable distortion.
Any minor smoothing artifacts in the adjusted meshes would likely be passable for smaller shapes that won't be seen up close or for objects that have lots of texture based normal details that are added after baking.
Increasing the geometry density further would produce slightly better results (Examples in the previous posts were 32 segment cylinders.) but beyond that the improvement in quality will be relatively minimal for smaller objects. So resolving smoothing artifacts and increasing surface quality without over-complicating the base mesh is about more than just blindly adding geometry until the shapes look good.
Recap:
The basic premise of subdivision smoothing is fairly straightforward. It's the interactions with all the other geometry elements that adds the complexity. Increasing the geometry density can help in some situations but isn't the be all end all.
There's different schools of thought on geometry density, modeling directly off existing geometry on curves, manually adjusting the underlying geometry to counteracting smoothing stresses, acceptability of different types of smoothing artifacts, etc. All of these different approaches generally work.
It's one thing to understand what works to resolve a particular problem but it's something entirely different to understand how something resolves a particular problem. Which is why it's often beneficial to try and solve these topology problems at the lowest possible level, while still creating accurate shapes.
What it really comes down to is reading through the existing documentation and experimenting with different approaches to figure out how everything works. A significant part of self directed learning is researching different strategies, testing sample models and evaluating the results.
Catmull–Clark subdivision averages between existing points and this creates a smoothing effect. Uncontrolled subdivision smoothing tends to deform shapes and shape transitions that lack supporting geometry.
Creating shapes that have crisp edges and minimal smoothing deformation often requires adding support loops on both sides of the edges that define shape profiles and shape transitions. Placing support loops on both sides of the defining edges will constrain the area that's being averaged out by the subdivision which tends to also reduce the associated smoothing effect.
Supporting geometry isn't limited to just support loops either. The distance between the existing geometry in the merged shapes also has an effect on subdivision smoothing. Increasing the distance between the segments (reducing geometry density) tends to amplify the smoothing effect while reducing the distance between the segments (increasing the geometry density) tends to reduce the the smoothing effect.
Increasing the amount of geometry in the starting primitives can make it easier to accurately merge the shapes and does tend to increase the quality of the subdivided mesh but arbitrarily adding more geometry to existing mesh doesn't guarantee an increase in overall shape accuracy.
This is why it's generally considered best practice to match the segments of the intersecting shapes whenever possible. Matching the segments of intersecting shapes tends to produce a mesh with a more consistent geometry distribution and makes it easier to maintain the accuracy of the basic shapes. More consistent geometry density will also tend to make it easier to use the minimum amount of geometry required to hold the shapes. Which helps reduce unnecessary complexity and makes the base mesh easier to work with.
Below is an example of what this process could look like for this particular shape. Start by evaluating the overall size of the object and the relative size of the smallest surface feature then select an appropriate number of segments for the smallest intersecting shape.
Use primitives to preview the base of the shape intersection. Since the larger cylinder isn't constrained by adjacent geometry and the cylinder wall segments will be disrupted by the smaller shape it makes sense to adjust the number of segments in the larger cylinder to match the existing geometry in the smaller cylinder. Keeping these cylinder wall edges parallel to each other and concentric with the rest of the large cylinder's geometry will help preserve shape accuracy and reduce unwanted smoothing deformation artifacts.
Perpendicular loops can be added to the cylinder walls to match the rest of the intersecting geometry. The segment matching doesn't have to be perfect. It just needs to be close enough that any difference between the two shapes can be taken up between the support loops around the shape intersection.
There's a number of different ways to approach merging the shapes and adding the support loops but if the goal is to maintain shape accuracy then try to avoid disrupting the position of the cylinder wall segments. It's also important to maintain relatively consistent segment spacing on the intersecting shape but if merging stray geometry requires making a compromise it's generally better to preserve the shape of whichever feature requires the greater degree of shape accuracy.
While there are edge cases where it makes sense to only add a support loop on one side of a shape intersection it's generally better to have support loops on both the outside and inside of most shape intersections. Placing support loops on both sides of the edges that define shape transitions will better define the area that's used to average out the smoothing and this helps constrain any deformation artifacts to smaller area.
These same basic concepts apply to shape intersections with tapered sides and complex compound curves.
There may also be situations where the size of the object, geometry from adjacent shapes or some other constraint makes adjusting the number of the segments in the shapes impractical. When using less than optimal amounts of starting geometry the next best option is to move or scale the geometry components along the surface of the the larger cylinder to counter act the edge stress caused by the subdivision smoothing.
The results won't be as accurate as segment matching but when done correctly there shouldn't be any major smoothing artifacts and the result is often better than omitting minor details or having to completely rebuild major sections of a model late in the modeling process.
Here is a comparison of the previous examples. Matching the geometry of the intersecting shape by increasing the number of segments in the larger cylinder does improve the quality of the surface but resolving the shape accuracy issues at a lower level of complexity and adding support loops to both sides of the shape intersection also help control the subdivision smoothing behavior.
Adjusting the position of over stressed geometry to counter the deformation caused by subdivision smoothing also produces usable results but does open up the possibility of inadvertently creating other types of artifacts and is really only a good option when segment matching isn't viable.
Segment matching is an important part of subdivision modeling and one of the keys to getting consistent results with complex shape intersections on curved surfaces. It's also important to preserve the attributes of the intersecting shapes and try to take up any differences between the intersecting geometry in the support loop around the outside of the shape intersection. This helps preserve the accuracy of the underlying shapes and will generally constrain any smoothing issues to a relative small area within the support loop.
Below is an example of how these concepts can be applied when matching the segments of the smaller intersecting shapes to the existing geometry of the larger shape.
Resolving smoothing artifacts caused by adding
surface features to a curved mesh is an often discussed topic so there's
a lot of prior art in this thread an elsewhere on the forums. The
previous page has a couple of good discussions about other methods of
adding holes to curved cylinders and there's been a recent discussion
about holes in curved surfaces and mesh complexity.
https://polycount.com/discussion/comment/2755033/#Comment_2755033
It's also worth mentioning that subdivision modeling is an inherently approximate process. The smoothing algorithm itself is averaging things out and behaves in a fairly consistent and predictable manner. It's art not purely mathematical CAD modeling. When it comes to using overly dense or complex base
meshes there's a definite point where the marginal returns start to have
steep time costs.
So there's always going to be tradeoffs between shape accuracy and modeling efficiency. This is why it's generally considered best practice to evaluate the overall size of the object, relative to the general view distance, and choose an appropriate level of mesh density based on that. Sometimes getting the desired results is about creating better geometry rather than using more geometry. https://polycount.com/discussion/comment/2746328/#Comment_2746328
Recap:
Match the edge segments around shape intersections. Use existing geometry as support for intersecting shapes whenever possible. Reduce unnecessary complexity by using the minimum amount of geometry required to hold the shapes.
Thanks. I assume the outer edge loop was added manually with the knife cut as extruding around such a circle is impossible in Blender?
Yes, you are right, it was done with a knife tool. I didn't care much about topology, for sure it can be done better. It's just to illustrate how adding control loops can tighten up geometry when used with a subdivision surface modifier.
I have been getting some annoying pinching This is the shape I set out to achieve this is what I'm getting I would also like it to be flush with the curved surface if someone has a way to fix this or has any solution to re-model this that would be great because I've tried everything I know and all the results have been crappy. I've tried increasing the poly count and trying it like that but it made it hard to model other parts of the model. So please if you got ANY solution please tell me. Thanks
Hi, not sure that I understand what exactly you want to achieve. Could you please provide a blend file with two initial shapes (before you try to boolean one out of another)?
I'm having trouble with modeling a pillow from a tutorial on youtube. This one to be exact.
I take a cube and scale it down the z to 0.03. In edit mode, I give it 30 cuts on each side and one going around. I add cloth physics with the silk preset. After tweaking the settings like the tutorial and adding a force object for the simulation, the cube just collapses on itself. When the cube is scaled up it blows away almost. I don't know why this is happening. Here are some photos.
Replies
Yes! have a lots of segments or resolution.
Workflow
Orthographic Views
Creating a spiral is not too difficult, but the problem is with the start/end parts of it.
The reference I have for it doesn't seem to have a gap between the start of the spiral and the cube it needs to connect to., and I'm not quite sure how to achieve that.
show us what you've done and where you are stuck mate?
I have been checking out both ChamferZone and Eugene, but still have some problems with above.
Some pieces will have to be made only by following the refs, so there'll be a bunch of guesswork, but some pieces, like the slide of a Glock 17 is easy to find the outer dimensions of. If you start with that which you know the dimensions of, you can line the rest up to fit.
Block out everything and check widths and lengths to see that you're happy with it before starting to add compound curves and shit.
If you're REALLY struggling, I would suggest that maybe you've started at a too high difficulty level. If you're completely new to hard-surface stuff, there's a buuuuuuuuuuuuuuuuuuuuunch of stuff that you won't learn until you make a mistake and get to correct it. If you start with a very detailed piece, it's gonna give you all of the challenges at once and that will feel overwhelming.
I suggest you start with something extremely simple, like a throwing knife. Model, bake and texture it and ask for help once you've tried and failed at something. When you're done, ask for crits in your own thread. Fix the stuff that's broken, and then step up to a fork.
Then a shovel. Then a serrated knife. Then a pistol.
As for learning, the best tool at your disposal is this or any other large community. If you want thought out responses like this one, you're gonna have to put in the work first. This one is a freebie. If you spend 5 or 0 minutes modeling before asking for help, people will spend the same amount of time helping you. They have better things to do than try to help someone who hasn't shown they've put in the time yet.
It could've been a bit easier if there was only text, then I could find a similar font and put it on the height channel in Substance Painter. In fact I already did it, but I can't say that I'm happy with the result. Here it is:
Anyway, I also want to add a logo as on reference photo, so probably it's better to make text in some different way so I could reproduce the original font manually. So the question is how would you approach this problem? Maybe I should create an alpha (not sure if the terminology is right, but hope you understand what I mean) based on the photo and then use it as a brush in sculpting software or in Substance Painter?
If you know more than one way of doing that, for example "workflow to achieve best visual result" or "fastest workflow", I would love to hear about all of them 🙂
If you have somewhere in bookmarks a tutorial that will help me I would appreciate it.
Thanks in advance!
Now I have made several things, but my topology is absolutely horrible and can be done better by a toddler, but I do have some liberties. There will be virtually no close-ups and if it looks right, it's good enough.
Anyway, I am (trying) to model a Husqvarna 701/KTM 690 SMC-R engine to be used in renders, but it's just way too complicated for me to get close to a good result. The entire cilinder is almost one huge intricate casting and I have no idea how to tackle it. I just make shapes and boolean them together and try to clean up as much as I can, but it's becoming a mess and i'm just not getting anywhere near to a good result. Any idea on how to tackle it, or what modeling techniques I should utilise to make this? Especially the part in the center of the cilinder head where the tube sticks out.
I did find a 3D scan of the same engine (with a few small differences), but it's not making it any easier for me. I have made another engine previously for the KTM 390, but for some reason that one was way simpler for me.
Workflow
What exactly are you having issues with? The only thing I can see is that some unions are a bit sharp. This looks perfectly passable if you ask me.
Maybe you'd have a better time if you planned out ALL the booleans before booleaning. That'll make cleanup easier, as you would have planned how many sides you want each cylinder to make the geometry match as well as it can.
(^ Splines used in a loft to make base geo ^)
(^ 360 Topology ^)
This should be quite similar to the shape you presented above, but with a bit cleaner topology. I've made it in Blender with help of the Bevel modifier. Here is a .blend file if you want to mess around with it.
The topology is not "all quads", but probably it's good enough, and I guess trying to make it all quads can make topology unreasonably more complex.
heres some reference:
I can't seem to get good shading around this area:
It's been frustrating me so I scrapped that attempt (mostly cause the side view isn't aligned with the front) and started a new model just using the front and on my second attempt I couldn't figure it out either, any help would be super!
heres the second attempt (different approach) It's got a sub smooth modifier
Thanks
Most of the control loops are generated from a modifier equivalent to the bevel modifier in blender
Something like this, there are many tutorials and videos explaining this workflow. I suggest starting by looking up "Arrimus3D" on youtube.
EDIT;
For the subdivision workflow you're going for, your base cylinder has way too few subdivision to begine with. I suggest starting over with a more dense cylinder (not too dense, maybe just double it), because if you're going to model detail into it later, you're going to run into problems, especially shading and smoothing artefacts.
Use existing geometry as support for intersecting shapes whenever possible.
Reduce unnecessary complexity by using the minimum amount of geometry required to hold the shapes.
obj - https://disk.yandex.ua/d/TgHdsRvOLhulmg
I did it according to the first option, but i got this "stretch marks". Can i fix this or is it acceptable for this element?
UPD: so, I just need to add more faces to the cylinder?
If the answer to either of those questions is yes then it's probably worth trying to resolve the smoothing issues. Whether or not it makes sense to try adjusting the existing mesh or completely rebuilding the model really depends on the project's quality standards and time budget.
This is the shape I set out to achieve
this is what I'm getting
I would also like it to be flush with the curved surface
if someone has a way to fix this or has any solution to re-model this that would be great because I've tried everything I know and all the results have been crappy. I've tried increasing the poly count and trying it like that but it made it hard to model other parts of the model. So please if you got ANY solution please tell me. Thanks
I take a cube and scale it down the z to 0.03. In edit mode, I give it 30 cuts on each side and one going around. I add cloth physics with the silk preset. After tweaking the settings like the tutorial and adding a force object for the simulation, the cube just collapses on itself. When the cube is scaled up it blows away almost. I don't know why this is happening. Here are some photos.