Custom Glasses Modeled in CAD

To begin, I brought my original roughly-modelled design into an assembly. I treated this like sketch templates, but it was 3D and I was able to snap sketches to it. I also brought the reference photos from the Smooth Operator sunglasses into the assembly. I used the initial model as a reference for the front view and the side profile of the arms. I used the photos for the top and side curvature of the frames and the top curvature of the arms.

I created a boundary surface, trimmed it with my new sketch and off-set it to create the basic frame structure. I did this several times to try to get the curves smooth.

My original design was trying to emulate the unabomber, with some added character. I like that these glasses are so big, that you feel like you can hide away from the world while wearing them. Because they give off an antisocial vibe, I thought I’d play into that by creating an angled brow. It’s cool how much the shape of glasses can suggest mood or create a character by emphasising particular facial features and expressions.

After I’d finished the basic shape of the frame, I went back to make sure it maintained my design intent. I’d rounded off the sharp edges to increase comfort and tone it down a bit. It’s difficult to tell when they’re not on a face, but I think the angles and slopes are definitely in keeping with my inspiration sources.

Next, I made the lens. I sketched construction lines between the two surfaces and made splines go through their centres. I creates a boundary surface and then trimmed it slightly larger than the hole in the frames. I like that the curvature analysis tool works on surfaces as well as sketches. This way I was able to judge the curvature against the existing object.

To create the built-in nose-pad as part of the frames, I extended the inner and outer lofts, trimmed away my desired shape and lofted between the two sections. I then had to trim away part of the back of the frame so that I could merge the nose pad to the rest of it. This was quite difficult, but turned out cleanly in the end. I only estimated the size and position based on real-world examples, so I’m not sure yet how comfortable they’ll be.

I made a split line so that I could see how some graphical differentiation might come across. I think it’s a bit too much and gives a feminine vibe, more like a lynx than a grumpy house-cat.


I based the arms on my original design, but changed them a lot. The initial arms were too short and rounded too sharply. While I loved their ability to cut out the sun, I gave up this aspect so that I could create a new hinge. My initial goal was to prevent hair from getting caught in the hinge (which is a huge issue with so many glasses!). I don’t think I achieved this goal as I got side-tracked by the aesthetics, mechanics and my own limitations in terms of modelling.

The hinge is built into the frame and the arm, with no extra parts. The ends of the arm are pinched together to be inserted into the female part of the hinge, built into the frames.

After sketching my new arm, I had to create the hinge section of the frame. This was by far the hardest part of this project. I tried a bunch of different methods, but my final body was a rather simple loft made up of some complicated sketches. Joining it to the frame seamlessly was exhausting. Because it touched the frame in so many places as it sloped into it, it was difficult to determine where to trim each body in order to join them together. I couldn’t join them as a filled surface because they came together tightly into a vertex, which created zero-thickness geometry. In the end, I made both bodies solids and combined them. This was messy and I wasn’t happy with the end result, but it created a part that I could work with and looks fine if you squint your eyes.

Once the part was solid, I could create a cavity for the lens to sit in. I had to manually input the XYZ values so that it would allow enough width for the lens to be installed, without cutting so deep that it would go through the frame completely.

Finally, I constructed the arm. For some reason the projected curve was being sent way off into the distance, despite previewing perfectly. After trying a variety of options to fix it, I wrote a forum post to ask for help. Before posting it, I tried one more thing – it worked! I can’t remember what it was and now wish I’d posted the question and answer after all!

Constructing the arms was a delight. I only take photos when I’m having problems, but the only one I have here is the final result. Time for a deep, relaxing breath…


And then of course, disaster struck.

When attempting to mirror the frame, a mysterious problem would occur that stretched one face into a near-zero-geometry shard. I tried every method of mirroring to no avail. One friend said the only way to fix it was to delete it and start again. That was not an option for me. But I could do the next best thing. Save it as a parasolid to delete all memory of its own idiosyncrasies. This did not work.

For some reason, it was displaying perfectly in Keyshot. As most of the output for this assignment consisted of Keyshot renders, I started to come to terms with the idea of having a shoddy model with perfect renders. Finally, my lecturer gave me the hot tip to adjust the fillet on the offending face. I adjusted it by 0.05mm and the problem disappeared.

Solidworks can be whacky.

As part of our assignment, I had to revisit the mannequin head I’d created for the last submission. The nose definitely needed fixing. So did the eyes. And ears. Instead I chose to give it a mouth. After attempting to loft the lips on their own and then fill to the rest of the face, I realised I could make some much smoother curves by lofting the lips into the face. This was simpler than I’d originally thought. I had two side profiles, the centre and the cheek. I then had three guide curves; the top and bottom outlines of the face and a spline going through the middle. The guide curves didn’t perfectly join the loft to the face, so I cut the outside away and filled with curvature-continuous edging.

When I mirrored the features, I had a seam running through the middle, despite putting tangent lines in my guide curves. To rectify this, I ran the loft across both sides of the mouth. The outside was too complex for a tangent fill to work, so I did this in two sections.

I also had to model a case for this assignment. I don’t have many photos of the process which means that it went swimmingly! I took photos of my own glasses case to use as reference. I used the same methods as the pump bottle to get my intended curves, then mirrored them 4-ways. I offset the surfaces as I had with the glasses frames to create the inner lining. Because my custom glasses are larger, with bolder edges than the glasses this case was designed for, I had to use the scale tool to make it large enough to contain the custom design. This caused issue with my mates and external references, so I did a dodgy; I hid the referenced parts and simply added in new parts where they needed to fit. Once I was happy with how everything looked, I broke the references and deleted the hidden parts (I think).

To assess the cost of 3D Printing my parts, I uploaded them to Shape Ways’ website. They instantly assess the cost of manufacture in various materials and provide information on the constraints of those materials and processes. I also tried Fictiv, which was much faster to use, but more expensive. I liked Shape Ways’ tool to repair broken models. Fictiv has a ‘print anyway’ option, but it’s difficult to tell what the outcome would be and they’re unable to quote accurately for broken parts.

When it came time to render, the poor joinery between my hinge and frame became really obvious. After failing to repair it, I tried to make a feature out of it, but that just showed up more problems.

I shrugged off the issue and chose a colour scheme I was happy with. I ran into a bunch of issues bringing the correct colour differentiation into Keyshot. I tried offsetting the surfaces to create new separate bodies, but that failed. In the end, parasolids fixed my problem. Praise be.

To import the case assembly, for some reason, I had to save the assembly as a STEP file. I can’t explain why, but it was great. Also, for some reason in Keyshot, the sliders max out at a small increment, but if you’re smart enough to just type the numbers in (I wasn’t), you can make the values as big as you want.

After rendering out a bunch of images, I had to create a presentation board. I haven’t fallen in love with informative posters. Design posters are like science reports, while advertising posters are like essays. I like essays. I like feelings. I like stories. I feel that a technical drawing is way more suited to presenting information than beautiful renderings. This is why I styled my poster like an ad.

I looked up sunglasses advertising and realised that they’re often focused on human models more than products. The ones that do focus on products either show lots of different pairs of glasses or one simple image of the hero glasses. I like the simple image, but don’t think that’s exactly what my lecturer has in mind. I took inspiration from a fellow student. His composition of one pair of glasses in multiple angles is really dynamic. It shows various aspects of the design, but feels like a single image. I incorporated this idea, alongside the human model aspect and some explicit branding to create my poster.

With everything complete, I’m still freaking out. I’m really excited about my animation. I learned how to control elements within the program and make the limited controls do what I wanted them to do. I made a video with moving and rotating parts, models, and cameras and played with the opacity to make particular elements fade in and out. I’m very proud of it. I gave myself plenty of time to render it, so that I could spend the last day of the assignment writing this blog and uploading all the deliverables.

I started the render last night and worked out how many hours til I’d want it finished in the morning. I set the maximum time to 10 hours and went to bed. To ensure that my computer wouldn’t stop the process, I plugged it into power and disabled the sleep mode.

For some incredibly stupid reason, Keyshot stops the render process when the screensaver is on. I woke up at 6 to see how my render was going and it had stopped after an hour. Luckily I’d woken early. With only 9 left on the render, I disabled the screensaver and continued rendering, confident it would be complete by 3pm, with plenty of time to submit.

It’s now been rendering for 8 hours and is almost halfway done. I’m not sure if it will keep going til 9pm (with my submission looming at 10pm) or just cut itself short after 10 hours. Either option is not ideal. I’ll post the final video here when it’s complete. Wish me luck.


Because I’d saved the keyshot setup for video as a package, I was able to copy it onto a bunch of the uni computers and output a desired section of frames from each one. To prevent the screens from locking, I had to walk around moving the mouse of each computer every nine minutes.

Running 8 computers at once, the render was complete within 45 minutes. I stitched them all together in premiere pro, submitted by 5pm and went off to the pub to celebrate with a beer.


I was annoyed that the cross-fade from the case to the head had a sharp cut. Other than that, I’m fairly happy. I plan on experimenting with Modo animation software next. It should give me much more freedom of control and hopefully a better quality animation too.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s