Modeling Challenge: Bottles

Bottles are a very common showcase for complex shapes. Plastic bottles, glass bottles, you name it. This challenge may be too ambitious, but I’m gonna see what you folks come up with. The quality of the work you have submitted in previous challenges is stunning. I’m not going to pin you down to a specific bottle, but I’ll let you choose what you want to attempt. The orange bottle on the left in the image above is an orange juice bottle from the grocery store, and is found in my surfacing book. The light blue one is something I did from one of Ed Eaton’s tutorials, and the glass bottle  is a Coke bottle. The green bottle is an old standard from theSolidWorks training manuals, and the red one on the left is something I bought at the grocery store and modeled by sight, and can be found in a surfacing class I did for SolidProfessor.

Well done bottles are enormously difficult, and will make you pull out all the tools and all the tricks. I suggest that you browse around the grocery store and take digital pictures of a bottle you’d like to model. These things are everywhere in your life, and the designs can be fantastic, some of them are fantastically difficult to model. Remember that this is a “modeling” challenge, not a “design” challenge, which is to say that this  about duplicating an existing shape as closely as you can, just from visual data. If you want to design your own bottle, that’s fine, but I’ll be commenting on the modeling, not critiquing design.

One thing I’d like to see is if you can cleanly execute a handle like the blue or red bottles. Handles like this are difficult because they flow into and out of the main shape of the rest of the bottle.

I’ll be looking for a several things:

• is it a solid?
• does it have a thicker neck than body?
• how is the bottom domed?
• is there a thread on the neck?
• is the neck round?
• surface continuity
• can you figure the volume of the bottle?
• extra points for multithickness shell feature
• extra points for shell to outside

Make sure to check back here every couple of days. I will add submissions to the end of this post, just like all the other challenges. Where possible, you will be able to download the models by clicking on the images. To submit a model, just email it to matt at dezignstuff dot com. Make sure you submit stuff that can be posted publicly and that you have the rights to, or at least say where you got something if it is borrowed. If you use a software package other than SolidWorks to make your bottle, just show or tell the basic commands that you went through to make the part.

Typically I post each part with an image, and make some comments on the modeling techniques used. I don’t grade the work or criticize, although I might make suggestions for improvement, mostly I point out the high points of models, and might pick out a model or two as exceptionally well done.

This is mainly an exercise to learn and have fun. So start modeling a bottle!

========================================

Be careful of what you ask for, because you just might get it. Entry #1 is from Garrett Brooks. This is a Klein Bottle, where the inside and outside surfaces are one continuous face. You can’t make them from a single surface feature in SW because of problems with self intersecting geometry, but you can construct it in multiple surfaces. Have a look at how Garrett accomplished this, and then made it solid to boot!

========================================

Entry #2 comes from Lin Shao Dun.  This is made with a single sweep. Lin turned this sweep 90 degrees and got some wild results. This is a great example of some of the control you can achieve with the sweep command. You need to download this one. just to see what he did and how he did it.

Sweeps don’t do domed bottoms very well when you sweep between top and bottom, but using this technique he could have easily.

Also, for anyone who’s curious, that finish on the bottle is the “rough draft” appearance. Lin is taunting me because I just put “rough draft appearance” low on my list of stuff I wanted to see in the software.

========================================

Entry #3 is from none other than Mark Biasotti. There are many things to point out here. First, click on the image to download the zipped file from the SolidWorks server. It is large enough (~35 MB), so by the time you’re done reading here, it may be downloaded. Next, notice in the FeatureManager, that it starts out with a long list of reference sketches, including sketches with +s next to them. These are sketch pictures, which are a great form of visual reference. This part, according to Feature Statistics has 175 features. Don’t be afraid to make a lot of features. You don’t get bonus points in a real model for using fewer features. I think some times people leave out stuff because they think a better model has fewer features. To me a better model has the correct geometry, regardless of the number of features.

If you roll back the part and roll it down feature by feature, you’ll also notice that Mark used a lot of reference surfaces. This is another great way to create design intent and get started on the model.

There are two problem areas on this part, as I see it, well three, really. I’m sure I’m not saying anything that Mark has not already agonized over a bit. I use the “Tangent Edges as Phantom” display setting. this allows me to see very quickly which edges are tangent, and which are not. SolidWorks in the past used a tighter definition of “tangent”, so these days almost everything you intend to be tangent actually turns out tangent. But on this model, there are two areas around the handle that are not tangent.

Here’s the first area. See that Mark applied the tangent constraint, but the Fill surface didn’t make it very tangent. Deviation Analysis shows it is at most 1.45 deg off. It’s close, very close, but no cigar. Is it close enough? Depends on your manufacturing method, probably. In this case, it is probably good enough. You never want to count on mold polishing to get rid of modeling imperfections, but in this case, you might not notice an edge in the molded part. And even if you did, is it really that important? I can’t answer that question. The answer would be different in different situations.

The next questionable area is on the bottle across from the handle. This time it was a boundary surface with a Curvature constraint applied. This can happen sometimes when the Direction 2 curves are not C2 to the face that the Direction 1 edge is on – so you are essentially asking the software to make the new geometry C2 when the existing references do not meet that condition. Without looking at this in great depth, I’d say that’s the situation that we have here.

None of this is meant to criticize Mark, but I just want to show that even in the hands of a recognized master, the software still has limitations. I don’t want to take away from the quality of the work, but rather to use it as a learning tool so you can see some of the situations in which you may have to go extra steps to really get perfect data. I also wanted to show this because this is probably one of the most complete examples we are going to see. I’m taking this apart in detail because I want to learn stuff as well. I hope that no one goes through my models like this, because they are full of stuff that you shouldn’t do. If you want to see me take apart my own models, I do it a few times in the surfacing book. Analysis of results is really a great way to learn.

I would venture to guess that most folks aren’t going to put in 6 hours or more to make a model for this challenge. The more realistic your modeling task needs to be, the more detail you have to show, and the more correct it needs to be, the more real life errors you will run into. In looking at the feature properties, I see that the model was made about a year and a half ago, so some of the feature functionality may have changed in that time, which may be what is causing some of these issues. If the part were rebuilt today using SW2009, I would be willing to bet that the results would be somewhat different.

Some of the things Mark did here are very difficult. First the three corner blends. Around the top of the bottle there are 2 3 corner blends, shown here inside the red rectangles. They are both involving the label face. Spend some time looking at the model to see how Mark achieved these very difficult blends.

Bending the handle into the top area of the bottle is also a difficult task.

Here’s one that maybe Mark didn’t intend.

Many label faces need to take a paper label, and paper labels don’t stretch. With a paper label, you can only have curvature in one direction. This face has curvature in both directions, so it is probably intended for one of several types of non-paper labels that are available these days. I just point this out in case you are tasked with making a bottle that requires a paper label. Anyway, on a spline used for the label face, you can see a small section of the curvature comb goes below the spline. This means that the spline flips convexity, and that the resulting face has a dip in it. This may be too small to see on the real part, but the curvature comb doesn’t lie.

Mark skipped a few things like shelling the part, doming the base, the neck, the threads, etc, but it is a great example of the exterior features of the bottle, and bottles in general.

Thanks to Mark Biasotti for submitting a great learning example. And a rendering he sent a link to.

========================================

Entry #4 is from Adrian Siu, his first entry in these challenges. This is what Adrian had to say about his model:

I didn’t have an actual bottle to model, I just had a picture of the new Tropicana jug, so I made something that had 1L volume and than scale everything and obtained an internal volume of 2.63L (89oz.).

I didn’t focus so much on design (the feature tree and approach are obviously not optimized and reflect the “pain” of trial and error) but I rather focused on the modeling challenge. Got the measurable volume, you have to roll back before the shelling operation, got the multithickness shell feature, got the shell to outside, got the threads on the neck. What I found really difficult to achieve by doing solid modeling was to get the label placement area, which I could not do… I’ll try more, maybe a surface design.

Entry #5 is from Michael McCune. This one is done as an assembly. I was expecting to see an inserted part somwhere to start the inner part, but he approached it differently with a lot of assembly layout information. Click the image to the left to download an 18 mb zip file (missing one part – the OJ! someone on my server drank it…)
Here is what he had to say about the model:

Here is my challenge bottle; it is a 3 Liter Tropicana Orange Juice bottle from the near future. Sorry I couldn’t include the part file for the Liquid OJ as it’s 100+ mb so I gave you a few pictures…the 1 and 2 Liter marks on the back are accurate btw…fun stuff. Uh, there is really no grand Industrial Design Thesis to this thing, I was just having fun. The shroud is injected and the inner bottle is blow molded and glued in place (OJ is very expensive in the future, allowing me an unlimited budget).

========================================

Entry #6 is from Mike Dzieciuchowicz. Mike used some good tools in the course of making this bottle. The shape on the top half of the bottle was wrapped onto it. Nice touch. The ring around the middle was a separate body and then subtracted. He also got in the multithickness shell to make the neck thicker than the body and the shell to outside so you are modeling the fluid volume rather than the bottle.

Mike also pulled off a thread, cleaned up extra faces with the Delete Face tool. His thread uses a variable pitch helix to avoid the common problem of “how do I end this thread?” That I think is the first time I’ve seen the variable pitch helix used that way. Thanks for the submission!

========================================

Entry #7 is from Matt Lombard. This is a glass bottle I did for a project. The “thread” and neck were copied from a Snapple bottle. It was done with a wrap feature. The fruit is supposed to be a mango. Looks more egglantish to me. All of that was done with patterned thickened surfaces. The pattern on the bottom was just a sweep and a fillet. Chewed up cpu time, though. The text was the most difficult thing here. It was wrapped onto a similar surface, then extended and trimmed to fit the double curvature of the mango. This was an interesting project, although I’m not convinced I was the best person for this kind of work.

The bottle has 136 features, about half of which are shown in the list. The Dome feature was used to dome the bottom. It was not shelled, but I used a cut revolve to make the inside smooth. In reality, the glass is molded using a blow-mold like process, and the inside takes on the shape of the tooling minus the glass thickness, with some natural smoothing of small detail. The text was what I was most apprehensive about. I was not sure that level of detail was achievable in the glass process, which I haven’t worked with enough to know stuff like that. Plus, the text on the bottom dome of the bottle might complicate opening the mold. I told the customer that they would have to deal with the mold maker on these issues.

I went around with the customer also on the text originally because she wanted a label, but also wanted the double curvature. I told her I could design anything she wanted, but the labels weren’t likely to work correctly.

The display here is just RealView with the Glass. Gloss, Blue Glass mater… er, appearance. I think it looks pretty cool.

The project was meant to put a bunch of people to work on a tropical island where they had a lot of wasted fruit, and no knowledge of preservation techniques. I hope it succeeded, although I have not heard back from the customer. I get a wide range of interesting projects like this that come in from time to time.

I can’t provide a download for this because it is customer’s data.