Started modeling a chair that will be made of weaved tubular sections, perhaps I'll make them into a bamboo-like material in the end, or just plastic, not sure.  Here's the workflow I have so far of the top of the chair, which is going to be a Cappellini S Sile chair, but with interwoven tubular sections.

I'll probably adjust the thickness of the weave after I complete the process, so that nothing is intersecting itself or nearby geometry.  It could be close as hell, but no intersections.

This is going to be poly heavy Lux! What's the story then, are you extruding spines?

Quote - This is going to be poly heavy Lux! What's the story then, are you extruding spines?

Oh yeah, this will be a little heavy on the poly side, but not as bad as you might think.  I'll be posting more screencaps later of the step by step, but the process here is just as you expected; splines.  I started by shaping  the chair from a surface using patch topology in 3DSmax (because patches are very easy to get perfectly smooth curved surfaces, kinda like Nurbs surfaces), then converted the result to quad polygons.  Then, in edge mode, I extracted splines from the edge loops of that poly surface, which gave me a series of splines in the exact shape of my S chair.  The hard part, so to speak, is now creating a weave by pulling the vertices of the "vertical" splines above and beneath the horizontal supports.  I tried writing a Maxscript to do it automatically for me, but my scripting skills aren't yet at the level where I could make it happen.

I don't need to actually extrude the splines, since that part is automated in 3dsMax,  simply by selecting "renderable" splines, as you see here.  The resulting geometry of renderable splines is quad based, and visible in the viewport, while still allowing me to manipulate the underlying spline itself at the vertex or segment level, with instant feedback of the geometry changes.  Once I'm done shaping the weave here, I'll simply convert them to mesh geometry, and they will be like any other polygon manifold.  At this early stage, I still have parametric control over the density of the spline meshes (to make the tubes smoother or blockier if need be), or I could create rectangular "strips".  There's a lot of options while they are still linked to the underlying splines, and I'm not dead set on exactly what I'll do in the end.  At this geometric resolution, the object will end up somewhere in the 150k quad range I'm guessing, but I may add even MORE detail if I decide to make them bamboo-like (such as insetting "ribs" at each ring edge along the length of the tubes to simulate how bamboo reeds look).  Plus I'll remove the caps at the ends of the tubes, which at this time are simply Ngons, and make them Quad caps before exporting.  That part is easy and automated, but adds to the poly count of course.

I modeled a weaved Easter Basket a couple months ago (render in my gallery), which used a totally different process, but the result was excellent.  This particular approach was advised to me by a classmate,  as he said the control over the splines is far better than the technique I used for the basket, so I'm giving this way a try.  So far, so good I think.  More screencaps to come.

As I feared, the weaving of the tubes are becoming more tricky to manage as I make my way down the shape of the chair, since the chair tapers to variable degrees, but the tubes, of course, are all the same diameter...

As you can see, they are now starting to intersect pretty badly. I have to think about how I'll handle this problem. Unlike the weave I used for the basket, this is going to present some problems, because of the odd shape of the chair. I'm either going to have to reduce the size of the tube diameter until they no longer intersect at the tightest part of the surface, or rethink the weave process entirely, and start over.

What's the issue if they do overlap?

Did you try an FFD. I know it will scale them but is it noticable

Overlapping geometry is something I typically try to avoid; it could cause artifacts when smoothing or rendering, and there's some other issues when it comes to compatibility with other software, especially if a geometry is going to be fabricated.  This one isn't, but it's still a rule I like to adhere to whenever possible.

I could taper the geometry along the spline using a modifier like FFD, but for aesthetics as well as realism, I'd rather keep the tubes a uniform diameter.  I'm thinking intersection may happen here regardless.  I may reconsider the weave technique for this particular chair, and go for a more simple surface geometry.  Perhaps I'll do some detailed pouf cushions for it or something like that instead.  Undecided.

Fair enough, It does look slightly fat atm. What happens on the ref image?

Quote - Fair enough, It does look slightly fat atm. What happens on the ref image?

I haven't found any reference images of a Cappellini S Chair with an interwoven surface, and this may be the reason why.  It was a novel idea I was going to try, because I felt the result would have looked impressive.  There was one photo out there of the chair done in  some kind of burlap-like "wrappings" that appear interwoven at first glance, but really aren't.  There's plenty of true woven chairs out there, but most of them are constructed from simple planar surfaces, or surfaces with a mathematically simple curvature, which would make it a much easier process.

Unless there's something obvious I'm missing here, the only real solution I can think of is to start over, and this time use many more, and far thinner tubes or vertical splines.  My process was flawed from the start really; I'll need to create the diameter of the tubes by first measuring the narrowest part of the chair's surface, and start laying out the weave based on the diameter, and number of verticle tubes I'll need to comfortably cover that surface area.

Just doing a quick google and it seems to come up with the tubes runnig horizontal - which kind of makes sense from a strctural support point.

Quote - Just doing a quick google and it seems to come up with the tubes runnig horizontal - which kind of makes sense from a strctural support point.

Yep.  It hit me last night... I'm running the weave the opposite way! If I run the support tubes vertically,  and the weave horizontally, which is what makes the most sense structurally, then I shouldn't have problems.  Total brain lapse here! haha.  Thanks, airflamesred!

I think now I'll need a way to automate some of the process, or it will be a nightmare to do manually.

Here's an example of a mesh weave I did for another chair I modeled:

That one was fully automated. I simply modeled one woven strip, and a maxscript did the rest for me. However, it was easy, because as you can see, it was a planar surface area. This one will be more complex.

Here's the new spline layout for the Cappellini chair, with the green support tubes now running vertically, and the "weave" tubes, in blue, running horizontally. Much, much better, and more sensible design.  Only now, as you can see, there will be far more weaved tubes to deal with:

This is looking a lot easier, the build, which I suspect is underlined by some metal construction, and maybe too many loops on the top one. There's a slightly stretch thing going on - two less loops might help.

I tried to create the twisted and woven meshsin the Blender before but very minor success.  I have the test 3dsmax logiciel to model now. I like it and the power of the tools, particularly in the tools for selection. Your works with this looks to be very well. I do not see many people making this kind of difficult modelisation of a chair..  Can you show the results please? I have use of your chairs before time, and they are again very well. you can see your other chair being employed in my gallery. je vous en remercie!

It's a good start,you can make it better,just need more time I think.

Any updates Lux?

Yes, I finally came up with a technique that works better for weaving a mesh on this abnormally curved surface area for the chair, but my workstation crashed the other day, and I lost most of my WIP screengrabs and many of my previous source files.  Luckily, I was able to salvage some of my work from this project, and move it to a temporary computer.  So I've started working on it again (had to start almost from scratch), and I should have new screengrabs to show soon.

It's a hassle working on this temporary CPU however, because it lacks the GPU and CPU horsepower of my defunct workstation, and especially when it comes time to set up production renders, I will be waaay behind with most of my work.  Hopefully I'll be able to go out and purchase a new system this weekend or perhaps Monday at the latest, and get back up to speed.

Here's some new WIP screengrabs of the weaved mesh.  I still have some tweaking and work to do, but it's a good representation of how the finished work will look...

As you can see, some of the weaved strips are angled incorrectly. I could just use tubes for the weave, and I may end up doing just that. However, I was trying it with poly strips first. I like the look of the weave, but can't figure out how the angles got thrown off slightly. Shouldn't be a problem to fix, but I think I might stick with tubes, and make them just slightly thicker.

Ok, i've decided I'm going to go with the tubular style weaving after all, and give it a rectangular border support.  Here's a WIP screengrab of the chair with tubular style weaving:

Just a few more tweaks to get some problem areas sorted out, I think..

Prefering the squarer edge, a nice contrast to the weave

Quote - Prefering the squarer edge, a nice contrast to the weave

Thanks.  You mean the squared border with the tubular weave, as in the last WIP?  That's the one I like best myself, and will be using for the final model.

Click the attached thumbnail to enlarge.

I added extra reinforcing struts to the inner support tubes, and I am currently modeling the metal base of the chair.  The tubular weaving is complete, but I'm thinking of doing something a little unique to the outer support structure.  I may twist it like braided bamboo, or molded aluminum, or possibly give it some reinforcing "leather" straps along the length, which will also be modeled into it, of course.  I know such things can be done with displacement or normal maps, but this is intentionally a high poly, detailed model intended for arch/viz scenes, or close-up rendering, so I really don't care about poly count.  The ending poly count won't be crazy anyway, as I've been conservative on the detail of the weaving itself.

Yes I think it needs something. How is the weave held in place at the framework?

Here's a wire of the base.  All of the cappellini S chairs seem to have this same base, some with, and some without the stoppers at the bottom.

Airflamesred, in the virtual model, the weave consists of a meshed spline object, seperate from the frame border.  However I imagine, in fabrication, this would be made of metal rods, welded to a molded metal frame, then coated in a thick plastic/rubber polymer (kind of like electical wiring would be).  It could be further coated with, say, a bamboo shell, leather wrappings, or some kind of material for aesthetics.

I'm not sure I understand why you think it's the math of the surface that makes the problem.  As far as I can tell, it's a pretty basic issue that you'd see in any shape with a non-uniform width or height, even a planar triangle. 

It looks like your problem is very simple: you're changing your density with your width.  In real life, the density of the weave would never change, but the amount of it would.  The ribs would lead down and stop when they hit the edge.  So the middle might be as you'd have it now, but the the top and bottom would  have at least one more rib on either side.  Given the current spacing of ribs, of course.  Or, in the first design, the woven parts would stop when they hit the edge.  Just like you've got them doing horizontally in the second design.  As they followed a path in the z and y planes, they shouldn't have shifted in the x direction.

But maybe I'm missing something.

Quote - I'm not sure I understand why you think it's the math of the surface that makes the problem.  As far as I can tell, it's a pretty basic issue that you'd see in any shape with a non-uniform width or height, even a planar triangle. 

It looks like your problem is very simple: you're changing your density with your width.  In real life, the density of the weave would never change, but the amount of it would.  The ribs would lead down and stop when they hit the edge.  So the middle might be as you'd have it now, but the the top and bottom would  have at least one more rib on either side.  Given the current spacing of ribs, of course.  Or, in the first design, the woven parts would stop when they hit the edge.  Just like you've got them doing horizontally in the second design.  As they followed a path in the z and y planes, they shouldn't have shifted in the x direction.

But maybe I'm missing something.

If you look at the final design, I've got weaving technique figured out, and corrected that shifting problem.  I could add one more on either side in the wider areas, and just cut them off at the border, as that would be more correct.

The issue initially was in the technique I chose to use, which wasn't working out for this kind of shape.  The technique of extracting splines from the original mesh edges was getting screwed up because of the variable density required for this kind of curvature.  I needed to find a different solution, as I wasn't about to manually place each and every weave along the surface area, even though that would be the most accurate way.

Thanks.

So how did you achieve the final look?  I think it looks good, but I hope you didn't manually weave each of those splines, because there are several automated ways to do that.  Being you are using 3dsmax for this, there's probably a dozen ways this could be done efficiently.

Personally, i would have just created one spline weave as a "template", then extract another spline from the profile of the original surface to use as an animation path constraint.  Then, simply animate the template weave model you made along that animation path for say 100 frames (perhaps more, depending on how many weave strands you think you might need), while also adjusting  the width and rotation of the template weave model accordingly, as keyframes.  It would only take say 10 minutes tops to do that.  Then, I'd make use of the SNAPSHOT TOOL in Max to automatically extract each keyframe as a static model.  Using the snapshot tool, only the parts that are animated are cloned into a new static mesh, so in this case, only the template weave you created would be cloned along the path constraint.  The result of this method is quick, easy, and you would automatically have a weaved surface that perfectly compliments the curves and variable width of your chair profile.

The final result is all that matters, but I hope you took an easier route.

I saw that you avoided the problem by reorienting the weave, and noticed that it did follow the edges horizontally.  It looks nice, and kind of makes me think of an outdoor lounge chair.  But I also noticed that you still didn't maintain vertical density, which is what I was talking about.

No, you shouldn't need to do it manually.  I don't see why you should need to manage the vertical shape manually when you already have it following the frame horizontally.  Not that I don't see how your original method caused problems, just that I'd think there would be a fairly similar way to make it work.  Especially since there doesn't seem to be a change in the curve along the x axis.

You might find a generic and holistic solution if you look into chain mail techniques.  Different type of weave, but the same basic issue.  Or, of course, how to make a peacock chair.

Quote - I saw that you avoided the problem by reorienting the weave, and noticed that it did follow the edges horizontally.  It looks nice, and kind of makes me think of an outdoor lounge chair.  But I also noticed that you still didn't maintain vertical density, which is what I was talking about...

Yes, you're right about that, and I'm in the process of correcting that.  Good catch there. After a while of looking at certain things, you often miss the obvious, which is why it helps me, at least, to post these WIP's and get some "real time" feedback.

This turned out to be a fun, but sometimes tricky build.  I obviously didn't have many references to work with, but I think it's going to turn out ok.

Oh, I think it will turn out great.  It looks good as is.  I'm just interested in this as a general mesh building problem.

I looked at the Blender solutions for this type of problem, just because I was curious.  I thought I'd pass along what I found in case you were interested in the methodologies for other projects. 

The most interesting by far was using particles for chain mail.  You make the particle mesh the 2 or more rings you need to for your chosen interlocking pattern to work (there's lots of different chain mail patterns), and make sure the emitter emits one particle per vertex and has as regular a topology as you can make.  It's not absolutely perfect, but it works most of the time.  The benefit is being able to clothify the emitter and have the chain mail automatically follow the cloth.  I've seen people sim that sort of mesh with rigid bodies, and it was generally more involved.

The second was pretty simple, but I think it works the opposite of what you did in the first one (though I'm not sure).  You make the flat mesh shape first, then you bend it.  That was naturally the way I would have done it, so I thought maybe you had avoided it because it breaks.  But seeing it in action, it seems to work OK.

My weave is actually more complex than the real one, but only because I thought it needed to be, when in reality, I had already nailed it on my second attempt.  Sometimes things aren't as complex as we imagine they should be.  I don't care really, I'd rather have a more complex chair, but wow.  I didn't think it would be that simple.

Hi, all.  Just thought I'd share some more updates.  This is the final progress of the weave pattern itself, considering that since I began this endeavor, I've discovered several ways to achieve weaved meshes far more efficiently and correctly than the way I went about doing this one.  There's some imperfections still in the weave itself, but I'm just going to leave them, as I think it will make it look more "hand made" in the end, and  also I'm about ready to just put the finishing touches on this and move on to unwrapping and texturing the model.  I've been working on it off and on since I started this thread, and learned a lot about technique, and how to approach intricate pattern modeling, and also discovered some really useful plugins that would have made this project a piece of cake had I started out with them.

I didn't, in the end, use any plugins for this, nor did I employ any scripting (although it would have been smart of me to do so in retrospect).  It's all basic spline modeling and polygon modeling.  Here's the final raw screengrabs of the "almost finished" model:

The next, and final step is to just add "wraps" to the outter border around the chair, to make it appear as if the weaved tubes are somehow fastened to the outer edge, or perhaps just for aestetics.

Obviously, this model doesn't follow the real weaved Cappellini S Chair exactly, because I wasn't aware one existed until I was well into the modeling process. So it's just going to by my own personal RE-vision of one for now. The base, however, is very close to the original, and the shape, obviously, resembles the actual design.

The next update will include the fully textured model, and possibly a final render. Hope this thread inspired some of you to try some pattern modeling of your own, and thanks to everyone for all the help and input along the way!

Très bien! Objet très compliqué.  Will you share it?

I've added a border that I hope resembles leather-like, or rubber polymer wrappings to the border (possibly even fabric of some kind; I'll decide on the texture once I start adding materials and UVmapping it).  This was done by lofting a star-like shape around the border spline, and twisting the resulting mesh until it looks like what you see here.  I think it looks cool.  Now I think it's time to UVunwrap and texture it.  Turned out to be quite a detailed chair.

Here we go again with tweaking, tweaking, tweaking. haha.  Figured out a MUCH cleaner and more realistic method to model the wrapped border.  See the attached image for the results.  Still used Lofting, but changed the shape, and the resulting mesh is much more like a wrapped frame now, at least in my opinion.  Cleaner, more defined geometry too.

Looking good. Personally I would have left the metal frame - I like the effect of the green - just a preference.

Quote - Looking good. Personally I would have left the metal frame - I like the effect of the green - just a preference.

Thanks!  Well, I think you might find the wrapped border more interesting once it's combined with a proper texture.  I imagine it kind of like a "faux" wicker-style wrapping.  In fact, if I keep the tubes with minimal segments, they look like wicker also.  I was thinking about a wicker texture for the entire thing, and it just might work.  You'll probably shudder at the final poly load on this, but I don't think it's unreasonable for this amount of modeled-in detail.  Unfortunately, the green is just a viewport object color, and won't be part of the final textured look.  

Quote - Click the image to enlarge.

Here we go again with tweaking, tweaking, tweaking. haha.  Figured out a MUCH cleaner and more realistic method to model the wrapped border.  See the attached image for the results.  Still used Lofting, but changed the shape, and the resulting mesh is much more like a wrapped frame now, at least in my opinion.  Cleaner, more defined geometry too.

Wow.   I have to admit, that is looking damn good.

So the model is almost fully textured now.  I've shifted gears in the end, and chose to go with a wood/wicker/Rattan look for the weaved pieces and overall appearance of the chair, and I think it was a good last-minute choice.  Here's the screengrabs so far with the new textures:

I've decided I may offer this chair as a freebie when it's all done. There aren't many fully-modeled wicker furniture pieces out there for free; certainly not of a Cappellini S Chair.  I think the polygon count will be on the high end, but quite manageable; especially in this day and age, where most rendering apps can handle high polygon objects/scenes. The textures are from CGtextures.com, and high res enough for very close-up rendering. I'll be working on setting up a final production render tomorrow.

Let me know your thoughts. Hope you guys like it; I'm pretty happy with it. Thanks.

Fine work with the textures. Would you have done this any differently, in retrospect?

Quote - Fine work with the textures. Would you have done this any differently, in retrospect?

Thank you so much for all you input in this thread, Mark.  I'm really happy with how the chair looks with these textures.  I'm working on a render today.

In retrospect, I suppose there are a few things I would have done differently, not that I'm unhappy with how this turned out, but simply because I've discovered some new ways to attack this kind of modeling procedure.  If I were to model another object that was made of interwoven mesh, like this one, I probably would take a much different approach; namely, scripting.  There's nothing inherantly wrong about the technique I used for this chair, to achieve the weaved geometry, but some of the procedure could have been dramatically easier if I had known how to script them out in Maxscript when I started.  I think also the one thing I would do differently is in my planning.  I really walked into this build with very little reference material (I only stumbled on one photo of a real wicker cappellini chair half way through the build), and I didn't have a clear vision of what the final piece was going to look like until late in the procedure.  All I had planned for with this chair was that it was going to be in the shape of a Cappellini S, and that it would be made from interwoven mesh.  Looking back, I should have thought it out far better, and had a plan for how it would ultimately look.

Here are the production renders of this model.  I may tweak the textures a bit, especially to add a little more detail to the metal base texture maps, before I release this as a freebie item.  Look for it under my account on Turbosquid, and ShareCG.com in the next couple days.  I'll also announce it on my DA page, of course.

Anyway, here's the renders of the chair, along with an interior room I quickly modeled in 3dsmax as well.

Hope you guys like it. I'm excited to see what some of you do with the model.  I'm proud, and fairly confident, that it's the only one out there of it's kind, and will be one of only a few fully-modeled wicker items available for free.

Wow.  That's damn impressive.  I think that metal base could use some better bump maps, but aside from that, it's a wicked model.  That's the most realistic model of a wicker chair I've seen yet, and those renders look like photos man.  Can't believe you're gonna give this out for free man.

Wow! c'est incroyable! Render is too so real!

Hey, all.  I finally got around to uploading this model for free download.  

You can download the model from either of the following locations, and it will also be available in freestuff here on Renderosity shortly, in both .max and .obj versions:

Turbosquid.com

ShareCG.com

There are full preview renders and wireframe previews to view, and a readme file with further details is located in the zip file download.

Hope you find it useful, and if you have any problems or questions, feel free to reply to this thread, or hit me on email or PM.  Thanks!