Home Technical Talk

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

Replies

  • sixbysix
    Offline / Send Message
    sixbysix polycounter lvl 6
    Very interesting, thanks guys! I'm not working for any game, just practicing modeling for the heck of it ;)
  • sacboi
    Offline / Send Message
    sacboi ngon master
    IronLover64  why do you insist on continuing to edit a sub divided mesh? I would've thought by now our combined replies too your requests thus far were helpful in providing direction toward a more straightforward modeling approach? 


  • IronLover64
    sacboi said:
    IronLover64  why do you insist on continuing to edit a sub divided mesh? I would've thought by now our combined replies too your requests thus far were helpful in providing direction toward a more straightforward modeling approach?

      
    You're right
  • wirrexx
    Offline / Send Message
    wirrexx interpolator
    sacboi said:
    IronLover64  why do you insist on continuing to edit a sub divided mesh? I would've thought by now our combined replies too your requests thus far were helpful in providing direction toward a more straightforward modeling approach?

      
    You're right

    take your picture, put it in affinity or Photoshop and draw over it how your edgeflow should go, this can visually help you to build it in 3D =)
  • IronLover64
    wirrexx said:
    sacboi said:
    IronLover64  why do you insist on continuing to edit a sub divided mesh? I would've thought by now our combined replies too your requests thus far were helpful in providing direction toward a more straightforward modeling approach?

      
    You're right

    take your picture, put it in affinity or Photoshop and draw over it how your edgeflow should go, this can visually help you to build it in 3D =)
    noted
  • Filip5
    Offline / Send Message
    Filip5 polycounter lvl 5
    Hey guys, I wonder what might be causing these shadow stripes on model. May it be geometry ? Not sure if posting into right thread but not really sure if this is modeling issue or lightmap. Lightmap is cranked up to 2k. Those two parts are modeled as separate parts


  • tcollier91
    Hello, new here. Mostly I do motion graphics stuff which doesn't require much modelling but I'd like to get better during this time. I'm probably approaching this totally wrong, and could use some help.

    Here is my reference, plus my full model and one of my reattempts. I cannot for the life of me to get the thumb area to shade correctly. Once I try to create the curvature the shading falls apart. Thanks in advance.



  • sixbysix
    Offline / Send Message
    sixbysix polycounter lvl 6
    @tcollier91

    Take my comment with a grain of salt since I'm practicing myself but I think the problem here might be that the mesh doesn't conform to the shape? Here's my quick take, it shades without issues:


  • tcollier91
    sixbysix said:
    @tcollier91

    Take my comment with a grain of salt since I'm practicing myself but I think the problem here might be that the mesh doesn't conform to the shape? Here's my quick take, it shades without issues:


    Thank you for your response. I had flattened my mesh to troubleshoot. Mainly I am having shading issues around poles. Your mesh is neglecting the area I am having the most trouble with (circled below), here is my latest progress: 
    I found a few models of the exact mouse on turbosquid, and I tried to look at their topology but they also seemed to have a few poles. Perhaps I am over complicating the mesh in pursuit of perfection which is just beyond my capabilities right now.
  • navneethdodla94
    Hi! i'm new to polycount. A friend told me this would a good place to ask help for topology. I've been trying to model Ribs across cylinders but there's a lot of pinching. Tried to reroute some loops but i made a mess.
  • FrankPolygon
    Offline / Send Message
    FrankPolygon ngon master
    @navneethdodla94 Start by blocking out the shapes and matching the cylinder segments near the intersections. Try to minimize the amount of superfluous geometry by using existing cylinder segments as support loops for additional operations. Avoid adding support loops and secondary details too early in the process as this will lock in the existing geometry and make the mesh a nightmare to work with.

    In this shape the segment count on the smallest horizontal cylinder is determined by the size and position of the triangular gusset. This same process of evaluating the shapes and matching the segment count is repeated for the slightly larger perpendicular cylinder and again for the largest vertical cylinder.

    Work through all of the intersection areas until the segment counts are reasonably uniform. This shape will resolve cleanly, in all quads and with minimal geometry, but there may be some edge cases where it's more efficient to use triangles or n-gons in the transition areas to avoid creating an overly dense mesh that's difficult to edit. A good block out should be the jumping off point for adding details and support loops.



    There's more than one way to approach this and different packages have different capabilities but here's two approaches that provide clean geometry, decent shape accuracy and minimal re-work. Some of this process could be streamlined with further use of modifiers but the purpose of this is to illustrate what's going on inside the mesh. Subdivision modeling is an approximate process so it's all about balancing the trade-offs between editing efficiency and shape accuracy. Forethought and neatness are helpful but perfect can be the enemy of good enough.

    In the process shown below the alignment difference between the cylinder is taken up inside of the support loops. This requires some minor cleanup work which can be done quickly by using tools like merge by distance, snap and merge, loop select and edge dissolve, cut edge loop, Tris to Quads, etc.

    How much cleanup work there is to do depends on how well the shape intersections matched, how the topology was routed, how the software handles chamfers, what the order of operations was and which tools were used. The overall goal is to be efficient by using modifiers and automated tools wherever possible, minimizing manual cleanup and keeping modeling operations non-destructive for as long as possible.

    Steps shown below: Block out the basic shapes, matching the segments in the intersection areas. Add in additional support geometry. Use Boolean unions to join the basic shapes. Add in secondary shapes and details. Add support loops with a chamfer / bevel operation. Split, mirror and merge down any overlapping areas or stray geometry.



    Here's an alternate workflow that uses a different order of operations with wider tolerances. This strategy requires less manual cleanup but is also less accurate since some of the edges along the cylinder walls are skewed. Since this difference is minor and spread out over a wide area it subdivides without creating any major artifacts. Whether or not this is acceptable for the project is up to the artist. It's also possible to quickly straighten the skewed edges using a constrained move snap to vertex operation.

    Steps shown below: Block out the basic shapes. Add in support geometry. Merge the shapes with Boolean union operations and use merge by distance to remove all stray geometry. Add secondary shapes and details. Create support loops with a chamfer operation. Use the snap and merge function to resolve any triangles or n-gons.



    This is not pointed at anyone in particular but more of a general note: As a courtesy to other artists and as an investment in your professional growth, please take the time to research, test and evaluate existing topology layouts and modeling strategies.

    Objects with multiple shape intersections are a great opportunity to learn by synthesizing the basic fundamentals used to construct each individual shape and apply them to creating something more complex. If we were to use this shape as an example, pages 175, 173, 172, 171, 170, 169 all contain detailed write ups on similar shape intersections.

    The key to learning is taking the time to research, test and evaluate the results of different strategies. Part of the evaluation process is to ask for feedback but it's also important to be self aware and to provide context by describing what problems you have identified, what strategies you have tried and how you would like to improve the results.

    Again this isn't a critique of any one person but rather a personal reflection on what I regularly see. Self directed motivation and education are two top recurring questions and I think it's something we can work to improve. It's not about discouraging people from asking for help but rather encouraging them to self evaluate and provide context that can facilitate learning.
  • navneethdodla94
    Whoa! Can't believe someone actually went through the effort of making rebuilding it and making such a detailed explanation. Thanks a lot @FrankPolygon. I guess i messed up because the rib was an after thought. If i had planned it early on by blocking i could have tired to figure it out but i remained stubborn and wanted to fix it with what i had. 

    One thing i could not figure out was how to add cuts and reroute loops on curved surfaces because it always seems to deform the flow of highlights when i use the knife tool to add loops. And since i'm transitioning from maya to blender i didnt know how to add edge flow to curved surfaces. 

    I'll go through 175,173 and 172, rebuild it in blender
  • FrankPolygon
    Offline / Send Message
    FrankPolygon ngon master
    @navneethdodla94 Not a problem, glad it was helpful. Thank you for being a good sport about tying your question to my thoughts on learning at the end of my previous post. That topic was on my mind for the last couple of weeks and just thought it was worth sharing.

    A shape that consistently fights multiple modeling strategies and tools is a good indicator that there may be a problem with the underlying geometry. Much of the topology strategy is decided early on so it's often worth revisiting the block out stage and resolving these issues at that level. Deciding when to toss existing work and start over with a clean block out isn't always an easy decision but it sometimes ends up being faster and less work in the long run.

    When adjusting topology layouts it's best to rely on tools that constrain the operations to the underlying geometry and preserve the existing shapes. Manual operations without constrains tend to bring things out of alignment and this can generate undulations and other artifacts when subdivision is applied. The overall goal is to preserve the basic shapes and maintain order when subdivision is applied.

    The way the intersecting geometry directly connects to the edge segments of the cylinders, without leaving a sufficient offset to act as support geometry, is likely the cause of most of the smoothing artifacts in the original model.

    If the knife tool caused additional issues it would be interesting to see a few examples but I suspect that some of the issues there may have been related to stray geometry which that tool can generate. The knife tool has it's place but adding all of the perpendicular loops with it would be a lot of manual work and moving the loops without edge constraints could introduce a lot of issues.

    Here's some alternate tools that would be helpful for modeling this shape. In Blender, most tools have useful functions hidden in the menu or accessible with secondary operators so it's helpful to look at the documentation and review at all of the features for each tool. Since this isn't a Blender specific thread I've put the links in the spoiler below:

    To add or adjust edge loops that flow across the walls of a cylinder try using tools like edge slide, loop cut and slide, vertex connect, etc.
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/edges.html#edge-slide
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/subdividing/loop.html#loop-cut-and-slide
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/subdividing/vertex_connect.html

    To add support loops around existing geometry try using tools like bevel (profile value of 1), offset edge slide and inset.
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/subdividing/bevel.html#options
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/subdividing/loop.html#offset-edge-slide
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/duplicating/inset.html

    To create, merge down and cleanup geometry try using tools like loop select + variants, bridge edge loops, dissolve edges, transform snapping, auto merge, limited dissolve, triangulate faces, triangles to quads, etc.
    https://docs.blender.org/manual/en/latest/modeling/meshes/selecting.html#edge-loops
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/edges.html#bridge-edge-loops
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/basics/deleting.html#dissolve-edges
    https://docs.blender.org/manual/en/latest/scene_layout/object/editing/transform/control/snap.html#transform-snapping
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/mesh_options.html#auto-merge
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/basics/deleting.html?dissolve#limited-dissolve
    https://docs.blender.org/manual/en/latest/modeling/meshes/editing/faces.html?quad#triangulate-faces
  • sacboi
    Offline / Send Message
    sacboi ngon master
    Curved surfaces can present a number of challenges whilst designing an effective cage especially when cutting shapes that in some situations individual vertex placement can be crucial too preserve a quality smoothed surface. I've a limited background in automotive visualisation and various key principles are usually adhered to when a 'class A' reflective finalised output is an intended goal so I'd suggest if you're interested in further researching optimised topology guides, then this series of resources may prove useful in specifically defining not only the 'how' but more essentially the 'why' something is done.


    Also a quick comment on your posted query. N-poles are common and a requirement for proper topology which normally occurs via modeling 'organic' curvature based objects so at a glance the circled region should shade nicely with better placement.     

     
    edit:
    Polycount wiki - Topology and Sub Division Modeling
  • IronLover64
    I did not plan on saying this, but I will shamefully admit that I have no idea how I should model the wing and main body. This is what my second attempt looks like and there are many problems. The biggest mysteries is how the topology should be done and what the modeling process should be. I am desperately requesting for help.
  • natec
    Offline / Send Message
    natec polycounter lvl 13
    IronLover64 said: "I am desperately requesting for help."

    I think you should delete what you have. Re-read the previous posts with the help already given. Really take time and understand what each person has spent their time showing you. Its already all there. Before you model anything, plan out your model a bit before you start. Maybe re-read it all again.

    It hurts to throw your work out, but I've done this over and over myself and it does help. You've been learning and are better than when you first started this model and it will go quicker with better results. Just try a few more times, maybe go through this entire thread a few pages a day, and then try a few more times.

  • wirrexx
    Offline / Send Message
    wirrexx interpolator
    I did not plan on saying this, but I will shamefully admit that I have no idea how I should model the wing and main body. This is what my second attempt looks like and there are many problems. The biggest mysteries is how the topology should be done and what the modeling process should be. I am desperately requesting for help.

    heybro, sorry for being late again, i've been sick still am, but i tried to do something to explain what we mean easier. Dont use a subd modell, use a mesh that is less dense, it's easier to control.

    Step one, i have three seperated pieces.
    step two, i align them to eachother
    step three i use boolean operaters and then merge vertices.
    step four add your support loops.

    I would lie if i am happy with results, but I just gave it ago 15 min ago and my head is killing me! =)


  • sacboi
    Offline / Send Message
    sacboi ngon master
    As mentioned you've made progress already just take a minute to step back a bit for a moment and try too absorb the various techniques we've shared. Believe it or not all of us, at one point or another have similarly been where you are, grinding away learning this stuff full on like a 'Trojan' until *that ball* happened to finally drop so stay frosty you'll get it and I definitly believe that.

  • IronLover64
    Thank you everyone, I took your advice and have started redoing everything. However, the wing poses a huge problem: the shape is too sophisticated. It has a strange curve I currently know two ways of modelling it without subsurf and none of them work out. I can make the airfoil shape first from the side view, then make the rest of the wing with extrusions. This leads to a bad wing surface (the surface is full of random bumps rather than being a smooth surface). I can model the wing surface first from the top view, then position the loop cuts vertically, but this leads to a bad airfoil (the side shape of the wing is very awkward and each vertice has an inconsistent angle). I'm out of ideas on the modelling operation
  • IronLover64
    Another person I spoke to says that I will have to sculpt the whole thing. Would this be the best way to do it?
  • sacboi
    Offline / Send Message
    sacboi ngon master
    IronLover64 said: Thank you everyone, I took your advice and have started redoing everything. However, the wing poses a huge problem: the shape is too sophisticated. It has a strange curve I currently know two ways of modelling it without subsurf and none of them work out. I can make the airfoil shape first from the side view, then make the rest of the wing with extrusions. This leads to a bad wing surface (the surface is full of random bumps rather than being a smooth surface). I can model the wing surface first from the top view, then position the loop cuts vertically, but this leads to a bad airfoil (the side shape of the wing is very awkward and each vertice has an inconsistent angle). I'm out of ideas on the modelling operation
    If you're able, can you please post shaded with wireframe screencaps of what you currently have?
    Another person I spoke to says that I will have to sculpt the whole thing. Would this be the best way to do it?
    As one possible alternative to define displacement detail using a base mesh, sculpting complex surface variation would certainly capture an intended result however I think at this stage, will hinder more than be in anyway helpful in terms of firstly nailing down those key aspects of subd modeling, prior too implementing additional workflows.

    edit:
    I'll also point out  @wirrexx  recent example that basically outlines those underlying concepts, including:

    - Blocking out either a low poly shape or separated meshes
    - Then further refine using existing geo alongside support loops too control the object's overall silhouette

    Whilst keeping in mind throughout the modeling process, making use of constrain too axis or indeed the 'Edge Slide' transform function to help preempt potential shading errors from occurring, I'd initially touched on in the previous page.   
  • IronLover64
    Here's what Ive been doing: I modeled the wing from the top orthographic view so I can get the cuves right and add loop cuts. Then I select each half loop and move it to the top or bottom one by one. However, the airfoil is as rough as rock. The process isn't finished yet as I still need to move the bottom ones and some of the top ones
  • IronLover64
    another reason why I had to resort to subsurf a few months back was because not even my best blueprints were able to show what the inner wing's airfoil, and since subsurf was able to create a decent wing at the time, I went for it.
  • IronLover64
    sacboi said:
    IronLover64 said: Thank you everyone, I took your advice and have started redoing everything. However, the wing poses a huge problem: the shape is too sophisticated. It has a strange curve I currently know two ways of modelling it without subsurf and none of them work out. I can make the airfoil shape first from the side view, then make the rest of the wing with extrusions. This leads to a bad wing surface (the surface is full of random bumps rather than being a smooth surface). I can model the wing surface first from the top view, then position the loop cuts vertically, but this leads to a bad airfoil (the side shape of the wing is very awkward and each vertice has an inconsistent angle). I'm out of ideas on the modelling operation
    If you're able, can you please post shaded with wireframe screencaps of what you currently have?
    Another person I spoke to says that I will have to sculpt the whole thing. Would this be the best way to do it?
    As one possible alternative to define displacement detail using a base mesh, sculpting complex surface variation would certainly capture an intended result however I think at this stage, will hinder more than be in anyway helpful in terms of firstly nailing down those key aspects of subd modeling, prior too implementing additional workflows.

    edit:
    I'll also point out  @wirrexx  recent example that basically outlines those underlying concepts, including:

    - Blocking out either a low poly shape or separated meshes
    - Then further refine using existing geo alongside support loops too control the object's overall silhouette

    Whilst keeping in mind throughout the modeling process, making use of constrain too axis or indeed the 'Edge Slide' transform function to help preempt potential shading errors from occurring, I'd initially touched on in the previous page.   
    I did try to follow those instructions and started to make the thing in subsurf 1 (hardsurface modelling the wing does not result in a good airfoil as mentioned) until I realized that I had to somehow connect the back of the wing and body together as seen here. Subsurf doesn't allow that and neither does hardsurface for the obvious reason.
  • Dawid Gomola
    Offline / Send Message
    Dawid Gomola polycounter lvl 6
    @IronLover64

    Your last post shows your main problem. Working on subdivision levels all the time is the easiest way to messy model that you simply can't edit.

    I will give you a little bit different approach to this. And that is: Take a break. Don't get me wrong, i don't want to discurage you at all. What i mean is to take a break from this specific model. You are throwing everything at it, and you will burn sooner or later. Learning through frustration is not that good.

    To be honest, i think this model is super tricky to start with. It is fairly simple, it does look very simple. But modeling a model with this amount of nice curves all over is very difficult. And this is bad place to start learning fundamentals of modeling.

    That is why i recommend you to take a break. For a week, for a month and do something else. Something simplier, to learn different workflows and solutions. Then go back to this model, and you will see a massive difference. There is one more bonus to that and this is also something very important. Success. Almost any success boosts confidence, and solidifies knowledge. You have a finished solution, that you know is working, and you can apply it. A lot of ppl before proper work, does some kind of warmup. Modeling something small and fairly easy, just to relax and boost said confidence.

    Now going back to first thing i said. You are misunderstanding a lot of feedback here. Even without reading anything, looking on pictures ppl posted, you can get what is common ground in every of them. And that is simple geometry. Subdividing is just a cherry on top. But the base model is very simple. Managable. Easy to edit. You build whole form like that, and then work on subdividing it. You are just starting with wrong step.

    About that sculting thing. Don't do that to yourself. Sculpting will not solve any of your problems and will add magnitude of it's own. Sculpting is different and hard skill. Sculpting nice hardsurface is just a completly different beast. Work on base skills first.

    There is a perfect topic on this forum i think to learn all of this:
    https://polycount.com/discussion/129938/week-8-the-weekly-hard-surface-challenge

    Those are different, easy and hard hardsurface problems. But there you can find solutions to all of them, wireframes, shaded models and different approaches to problems. Do some of these models, do all of them if you feel like it. There is even more:
    https://polycount.com/discussion/132146/week-40-the-weekly-hard-surface-challenge

    Another 30 weeks of challanges. Trust me and i bet most of ppl here will agree, this will skill you up much much faster. Work on it for some time and then do your space ship.

    And if you want to persist on this model, there is another thing you can try. Instead of subdividing model, do the oposite. Limit yourself. Try to make whole shape and use max 1000 tri or poly. Make it 2k if it is too hard. Build a full base mesh. Edit it nicely. And then you can work on each part to make it look good smoothed. Just don't go straight to 10-20k, that amount could be a nice mid poly model ready for subd including support loops.

    Best of luck to you. You will get it, sooner or later.
  • Prime8
    Offline / Send Message
    Prime8 greentooth
    @IronLover64
    I recommend you take a few steps back and use a less complex object for practice.
    Try to get comfortable with subd modelling first.

    Generally if a surface is continuous, like the wing and body, you should model it as a continuous surface, otherwise don't.
    Wirexx already showed you how to do this.
  • pr3stl1
    Hi,
    How about such a problem? The second day I can’t decide, it's stupid :s 
  • ned_poreyra
    Offline / Send Message
    ned_poreyra triangle
    So, this. I'm doing an exercise on subd modeling and I want to make a sphere with triangular-ish incision with sharp edges. I know I have to quadify the area around the triangle, but I can't figure out the edge flow. Here is the obj if anyone wants to help me solve the case https://www.sendspace.com/file/bh992d




  • .Wiki
    Offline / Send Message
    .Wiki polycounter lvl 7
    Is there any reason why you did the intrusion in such a weird angle? The quad-sphere you were using gives you the perfect base topolgy for this intrusion. If you want it to be rotated then simply rotate the whole sphere.
  • rage288
  • ned_poreyra
    Offline / Send Message
    ned_poreyra triangle
    .Wiki said:
    Is there any reason why you did the intrusion in such a weird angle?
    Yes. I'm doing it as an exercise. It's intentionally made this way, this is not a real thing I need to model. I want to learn how to make any shape with quad topology. I've been recently watching Arrimus on youtube and his workflow is amazing. He combines smooth shapes with sharp shapes, doesn't care about angles or amount of detail needed, he just brute forces through everything. I want to learn to be like that, so I started simple with the sharp pyramid inside a smooth sphere.




  • wirrexx
    Offline / Send Message
    wirrexx interpolator
    edit:Admin delete message please.
  • IronLover64
    I decided to halt Sky Whale for another time and build a TAC Falcon. What would be the best topology for the main body shape?
  • perna
    Offline / Send Message
    perna quad damage
    .Wiki said:
    Is there any reason why you did the intrusion in such a weird angle?
    Yes. I'm doing it as an exercise. It's intentionally made this way, this is not a real thing I need to model. I want to learn how to make any shape with quad topology. I've been recently watching Arrimus on youtube and his workflow is amazing. He combines smooth shapes with sharp shapes, doesn't care about angles or amount of detail needed, he just brute forces through everything. I want to learn to be like that, so I started simple with the sharp pyramid inside a smooth sphere. 
    post of the year

  • FrankPolygon
    Offline / Send Message
    FrankPolygon ngon master

    Nothing wrong with testing different workflows but it's important to focus on the results and evaluate whether or not a given strategy justifies the time spent. Work through each modeling strategy and pick the one that produces the best results for the least amount of time and effort spent.

    Here's a few more examples of the previously mentioned strategy of blocking out the shapes, matching the segment counts on intersecting shapes and using the primary shape geometry as support loops. Like Eric mentioned, simply rotating the sphere to match the cut out solves the problem.

    The last example is not a direct copy but an abstract illustration. Producing sharp corners on these sort of shapes is relatively more difficult than producing rounded corners so if this approach works here it will also work on the previous example in the original question.



    Not going to critique a three year old video but there's a few things of note: It's a concept type sample model, there's a lot of manual work and both n-gons and tris are used in various places on the model.

    With most concept models the focus is the artistic aspect rather than the technical. I have seen a few other long time professionals who do 3D concepts and they do tend to do a lot of manual work. Tweaking each vertex one at a time, cutting in additional loops one segment at a time, etc. This might be fine for 3D concept work but the time cost of manually adjusting every facet of the model could be problematic in a reasonably busy production environment.

    Just because a mesh is all quads doesn't mean the topology is any good. Likewise triangles and n-gons don't mean a mesh is inherently bad. It all comes down to how and where they are used. When used correctly n-gons are a huge time saver and the results are visually indistinguishable from an all quad mesh that takes longer to make and is more difficult to edit.

    There are numerous comparisons, discussions and examples of topology strategies in the pages of this thread. To anyone interested in different modeling and topology strategies, just roll back over the last 20-30 pages and take the time to test each of them. There's very little magic here. It's all repeatable, testable and verifiable. Everyone builds their own modeling process and has their favorite strategies.

    Using all quad geometry solely for the purpose of using all quad geometry, without a specific and legitimate technical reasons to do so, does not make the work better nor is it an optimal strategy. Manually brute forcing all quad or even grid topology is, in my opinion, an egregious waste of time. It would be more efficient to leave these tasks to automated re-meshing tools.

    This is just my general philosophy and not pointed at any one person in particular, so that said: There's a difference between well organized neatness and the compulsive pursuit of perfection. There comes a distinct point where perfect is the enemy of good enough because nothing is ever going to be perfect. There's no benefit to suffering for technical art. Time wasted on the unnecessary fetishization of technical details is time that can't be spent doing more productive and enjoyable things.
  • ned_poreyra
    Offline / Send Message
    ned_poreyra triangle

    Nothing wrong with testing different workflows but it's important to focus on the results and evaluate whether or not a given strategy justifies the time spent.
    He's like 10x faster than me. And I've been doing 3D for 7 years. Does it mean I should basically quit? Because time is the primary reason why I want to learn his technique. I bet he can model a full car with insides and everything in like a day.
  • FrankPolygon
    Offline / Send Message
    FrankPolygon ngon master
    @ned_poreyra At the risk of going off topic, that particular question is something that you and only you can answer. My schedule is somewhat limited but if you want an honest evaluation of your modeling process (and some feedback with links to resources that may help) then record a video of your high poly modeling process, upload it to YouTube and PM me the link. No need for a voice-over or anything special.  It can be a simple 5-10x time-lapse.

    The Polycount Hard Surface challenge thread has a good selection of manageable projects so pick one of those if you'd like to take me up on the offer. Here's a link to that thread: https://polycount.com/discussion/132146/week-40-the-weekly-hard-surface-challenge/p

    No obligation. Just wanted to put the offer out there.
  • Blaizer
    Offline / Send Message
    Blaizer greentooth
    We say here: "the harder you try, the faster you get".  But man, don't be exaggerated/hyperbolic @ned_poreyra.

    BTW, I think all time wasted with logic in modelling is justified (playing videogames would be different, and that's what too many people do instead of improving).

    If ned_poreyra says he's practicing to get better/faster, well that's good, and there's nothing wrong with that. Doing things in a more difficult way makes us to analyze things better, but one thing is this, and other that "sphere". Like it's said here, your doing it in the wrong way, and you know what, you are going to nowhere.

    With time and effort we can achieve great things, that's for sure, but effort applied without logic gives us nothing. I'll be honest and i'd say you are wasting your time, and as result, others are wasting time trying to help you with an exercise i find stupid and noobish.

    I've lurking this thread since years, and i'm astonished, perplexed to see that too many people still don't use triangles or pentagons in their meshes. But if they are 2 of the 3 pillars of subdivison modeling, for god's sake!!. It may be the result of not trying harder? not knowing anything about 3d modelling?. Do you really have 7 years doing 3D? really? quality years? 5-12+ hours per day of modeling!? i have very serious doubts.

    Frankpolygon nailed it with its reply and help. kudos for him.

    You could save time, effort, and have a clean and better topology using triangles where they are needed. And Pentagons aswell. A Quad polygon would be like a sheep, and triangles and pentagons shepherd dogs.

    If you can't imagine it, a 100% quad mesh has "poles", and there's always poles with 3 and 5 edges. So you can use triangles and pentagons without fear in the lvl 0 of a subdivision model and take advantage of them. And later, if you need more geo, apply a subdivision iteration and that's all. That's the methodology of subdivision modelling, you are not making a "raw" mesh directly.

    Try harder, but try harder to be smarter in the way you work. In a job we can't waste time, and we are asked a lot. You can't pretend to do a "hyperbole".

    The topology is defined by the shape and edge flow
    . It's not intelligent to do a sphere with a box cut in the way you pretended. If I were your teacher, and you present me that sphere, I would suspend you without doubts. And i'm not being harsh.

    If you want to get faster, don't mimic/copy what others do, analyze the what and the why. Try to understand if something could be done in a faster way instead of copying a way of doing things. Don't waste your years without trying to understand what others are doing. We work with geometry, and geometry is maths. don't memorize tutorials, understand them, and if you can, improve them.

    this is incredible.
  • perna
    Offline / Send Message
    perna quad damage
    @ned_poreyra
    as it appears you are being serious, I'll reply seriously as well.

    You propose that you may increase your efficiency by intentionally making a mess of your topology. Which system of logic would have to be applied to take this out of the realm of the absurd? If you already struggle with speed you are not going to see any improvements by misaligning your geometry on purpose.

    Don't seek "alternative" modeling approaches. Efficiency boils down to understanding the fundamentals of sub-d and employing a concrete workflow.
  • ned_poreyra
    Offline / Send Message
    ned_poreyra triangle
    You either completely misunderstood my goal or I'm still horrible at communication.

    I want to create meshes that:
    - can have any shape that could possibly exist
    - are lightweight (to share, to load, to have in viewport)
    - easy to modify (no rebaking anything, no dependencies on specific software or plugins)
    - clean (easy to understand by someone else)
    And I want it fairly fast.

    Let's say I want this thing:

    Doesn't matter what it is. I want it precisely this way, up to a milimeter, no compromises. You'll say: "just sculpt it"! Ok, but then I have a high poly mesh that I can't easily unwrap, load or share. "Then retopo and bake normals". Ok, but I can't easily modify it, I have to retopo and rebake every time, I have to share it with textures. "Well then don't bake normals, just do a full retopo for subd with details". So... why wouldn't I just model it that way in the first place? Because that's exactly what I'm trying to learn here.
    Blaizer said:
    If you can't imagine it, a 100% quad mesh has "poles", and there's always poles with 3 and 5 edges. So you can use triangles and pentagons without fear
    I don't know why you all sticked to that "100% quad mesh". I used "all quad" as a synonym of subdivision modeling - a mesh that doesn't have pinching or artifacts when smooth is applied. Of course I'll leave triangles or ngons if they're on a flat surface, not causing pinching etc. But that's not my case. My case is a sharp-edged incision on an organic shape. Sphere is there just as an example - like I already said - it could be any soft, curvy shape and the pyramid incision could be inserted at any angle.
  • perna
    Offline / Send Message
    perna quad damage
    Doesn't matter what it is. I want it precisely this way, up to a milimeter, no compromises. You'll say: "just sculpt it"! Ok, but then I have a high poly mesh that I can't easily unwrap, load or share. "Then retopo and bake normals". Ok, but I can't easily modify it, I have to retopo and rebake every time, I have to share it with textures. "Well then don't bake normals, just do a full retopo for subd with details". So... why wouldn't I just model it that way in the first place? Because that's exactly what I'm trying to learn here
    So you simply want to use sub-d modeling. Then why not align the sphere cutout properly the way it should be done with sub-d modeling? And why the talk about "doesn't care about angles or amount of detail needed, he just brute forces through everything"? Arrimus advocates clean topology, not making a mess of things in the hope that it will somehow increase efficiency.

    Model the sphere the way @.Wiki and @FrankPolygon pointed out. Master the basics before you spend time on weird alternative approaches.

  • zachagreg
    Offline / Send Message
    zachagreg ngon master
    ned the reason they are explaining it the way they are is because what you expect is a pipe dream and not realistic. You have just said you want is every possible thing that professional artists must make choices about throughout their day to day. You just want a perfect pipeline where programs know what you want before you do it. Reality doesn't work like this. Polygons don't work like this. Sub-D doesn't work like this. Arrimus is a talented artist but as mentioned it's concept and he has been doing that kind of concept for a while so he knows what geo and kind of shapes can go in various geo types because he has experienced similar forms before.

    He isn't doing things completely random even though it may seem he is. If you want that thing work within the bounds of the technology or get to work with modeling in clay and photoscanning things. You want clean geo without the work of creating and maintaining clean geo? No plugins? Too bad. It just isn't how things work, or you better be damn good at coming up with the technical methodology and programming as to how a machine could do it.

    You are spending more time bashing your head at this hypothetical problem which is one you've fabricated presumably because you are having an art block at the moment or are finding frustration in your work. Stop trying to find ways around things that thousands of professionals everyday are dealing with and haven't found a better method. Just bunker down and do the work man, you have skill, refine it, grow with it, and learn. If it is speed you want making things harder for yourself doesn't help. Try also to stop consuming tutorials and what not, some are scripted and planned before hand. Some people are just really fast. Some people take forever and do weird things. Focus on yourself and not what others are doing and stop comparing yourself.

    If big groundbreaking geometry modeling tools suddenly come out of the woodwork that allow for this freeform auto clean modeling everyone will know and a lot of people will be happy. This is not what's happening currently though nor in the foreseeable future.

    Take some time, maybe step away from tutorials and other peoples work or even your own. Realize that every one of us on Polycount and outside have failed much more than we have succeeded and that everyone grows and learns and performs differently. Give yourself a break man.
  • ned_poreyra
    Offline / Send Message
    ned_poreyra triangle

    perna said:

    So you simply want to use sub-d modeling. Then why not align the sphere cutout properly the way it should be done with sub-d modeling?
    Because maybe I would want a shape like this.
    How do you use your edge aligning trick here? You can't, you just have to figure out how to insert those triangles at any possible angle, into any possible geometry. How many times do I have to explain that the sphere is just a hypothetical example?


  • perna
    Offline / Send Message
    perna quad damage

    perna said:

    So you simply want to use sub-d modeling. Then why not align the sphere cutout properly the way it should be done with sub-d modeling?
    Because maybe I would want a shape like this.
    How do you use your edge aligning trick here? You can't, you just have to figure out how to insert those triangles at any possible angle, into any possible geometry. How many times do I have to explain that the sphere is just a hypothetical example?


    Good luck.

  • ned_poreyra
    Offline / Send Message
    ned_poreyra triangle
    I decided to halt Sky Whale for another time and build a TAC Falcon. What would be the best topology for the main body shape?
    IronLover64's request got buried under my discussion, so I feel obliged to help him.

    Here is how I'd do it. Maybe someone can figure out a cleaner edge flow. The part that slopes down and creases at the same time is tricky. https://pasteall.org/blend/23fdae71ff074a3aae2b36925e1caa96

  • Blaizer
    Offline / Send Message
    Blaizer greentooth
    @ned_poreyra you are making things more complicated than needed. Those examples can be modelled with subdivision modelling, but it's a total waste of time because those shapes are not very realistic for hard surface models. You'll need too much time for a small piece, and you'll need to work with a good amount of polygons, a very dense mesh. The mesh wouldn't be the typical low poly subdiv model.

    For those kind of weird meshes is faster to use Meshmachine for Blender, or Zremesher used wisely in Zbrush.

    Anyways, if you want to do that in subdivision modelling, you would need to do booleans or mesh projections/inserts and mesh cleanups like crazy. We also have kitbash tools. Don't forget to always use subdiv primitives/meshes as boolean operators.

    this is an example of mesh projection:



    and this one aswell, with cuts:


  • IronLover64
    I decided to halt Sky Whale for another time and build a TAC Falcon. What would be the best topology for the main body shape?
    IronLover64's request got buried under my discussion, so I feel obliged to help him.

    Here is how I'd do it. Maybe someone can figure out a cleaner edge flow. The part that slopes down and creases at the same time is tricky. https://pasteall.org/blend/23fdae71ff074a3aae2b36925e1caa96

    thanks a lot
  • IronLover64
    I decided to put wirrexx's idea to the test. This is what I got for the front wing. Any fix for the "lower poly" part?
  • ArcticTauntaun
    Hey guys!

    I've been trying to model a helmet lately (in Maya, no ZBrush) as practice for hard surface subd modeling as opposed to sculpt + retopo.

    However, I'm stuck trying to figure out how to best connect these different shapes with optimal topology while preserving the sharp insets in the visor as seen in the reference image. 

    I modeled the base of the helmet using a cylinder scaled then moved into place via soft selection, is there a better way to get a more accurate curve for this kind of shape using pure subd? I'm using holding edges to keep parts of the visor sharp but that seems to cause some shading issues with super visible seams, would changing edge hardness/softness fix this?

    In ZBrush I'd just take a sphere and gizmo extrude it, dynamesh and go to town, any general workflow tips for approaching this type of model (hardsurface but organic) in pure subd with interesting shapes and insets would be greatly appreciated.



    Should I be modeling each piece separately like my current attempt or should I model out a half sphere + extrude out the cylindrical base and then add on the visor and other parts later? Then boolean the different intersections?

    Thanks!
Sign In or Register to comment.