Science Fair

The science fair was a blast. As we had a pretty simple set up, but rather efficient. Fortunately, we could use a few prints of Tim as well, including the red colored prints, which grabs the attention of people.

Setting up

Red balls

As soon as the lunch break started, people came by and looked around. There were people that looked around and went to the stands that grabbed their attention. And there were people who looked at every stand. The ambiance was great, everyone that passed by, was really looking into the projects.

The visitors were intrigued by the project, it shows them the possibility of 3D printing. As the subject is pretty technical and theoretical, we did manage to give them a clear explanation of what we have been doing and the results. The people who were more interested or who were involved with 3D printing themselves, asked rather clever questions. It was clear that they were really interested in the new techniques of 3D printing.

There were also people who did not know a lot about 3D printing but despite the lack of knowledge, they still wanted to know what we did. They tried to understand it and it was noticeable that they were trying the comprehend what we told them. In the end, they were intrigued by the project. Mission accomplished!



There were a few kids who went by as well, most of them were not that interested. A few did come by to have a look and we showed them that the prints were compressible and squishy. It peaked their attention immediately and they wondered what this project was about. It was noticeable that the children were quite clever and have a great understanding.

The event was really nice to have, luckily we had no complications and we could have a chat with several people, technical, non-technical and even children.

And this concludes our journey with the Fractal Space project!

Printing full size soles is not possible

Today we decided to try to print a full size sole. Since there is not space enough on the build plate to a fullsize print. We therefore split the sole in half, so it could be printed in two prints and then glued together. But we encountered a lot of problems that we didn’t think of in the beginning.

We tried multiple CAD programs and it was always the same. First we converted the STL file in the program. We converted the mesh to a BREP, so we could work with it. Then we tried to split or cut it in two. This is only possible with a BREP-file. The two parts we now have are not solid and it was not possible to make them solid, at least not in Fushion 360. So we tried Rhino – same problem.
We then drew an entirely new sole and cutted it in two (in Rhino) and then use the “Cap” function to make the parts solid. It would only cap planar curves, so we couldn’t make an arrow shape as we first wanted to do.

Then the real problem. We only have 21 hours before the science fair and the back sole would take 1 day and 17 hours, so it was truly not possible to do.

So the 3 hours we spend on modelling today, was basically a waste of time. We instead tried to print the same sole as before, but with a new color map. The new color map makes the sole hard where the foot needs support and soft where most of the pressure is (for example the heel). The sole makes the pressure from the body more evenly defused.


Variable Density Infill

This is the first time we printed models with variable infills. Giving a model variable infill is still experimental and will not always work as we like. The varying infill is done by Cura and it is done by using an image. The lighter the colour, the less infill. Currently, we work with black and white images as these are the easiest to change and determine the right infill.

Black and white image of the foot

So first of, we tried to print a variable infilled shoe sole. It was done with black and white image above, which resulted in the sole.

Sole with variable infill

Sole with variable infill side

The sole looks pretty good and the infill looks alright, but the infill density still has to be reversed. The high infill spots should have lower infill instead and vice versa, so we will do a new print soon.

As this image is taken from the 3D scan, we did not determine the right infill. In the next test, we made a gradient image from white to black and printed that in a rectangle, to see how the program exactly determines the different levels of black.


Variable infill rectangular

Variable infill rectangular side

The rectangle has been printed alright with different infill density, but there are some issues. There is this issue, if you make the infill to white, it will have the lowest infill, below 5% infill. If you go lower than 5% infill, the fractal will not be connected and will hang loose. A lower limit would be helpful in the program. The program does not accurately determine the right infill per colour. We also tried to make a print with a different image, which has pre-determined ”infills”. It’s basically an image with different levels of black, ranging from 0% to 100% with increments of 10%. Cura did not recognize the levels of black after 60% as you can see in the image below. So there might be some software tweaking.

Gradient with 10 percent increments

Cura Rectangular

Repeated Compression Test

The soles of your shoes gets compressed a lot on a daily basis. It is recommended to walk 10,000 steps a day, so about 5,000 steps per foot. That is a lot of repeated compressions. To test whether the new pattern suffers from this repeated compression, we had to test it.

The setup was pretty easy, we printed a cube with 50% infill with a size of 20 by 20mm. The cube has to be compressed 100 times under a load of 4,000N. As we had to do this test manually, we set the load quite high and the amount of repetition significantly lower than the actual amount steps.

Graph of the repeated test

In the graph above you can see the results. The first test is the blue line, the second test is the red line. The odd thing about this test is, that the cube gets compressed rather easily in the beginning and becomes more stiff at a certain point. Starting from the second test, the cube is rather stiff in the beginning and gets less rigid at a certain point. Going on from test 2, the cubes becomes less stiff in the beginning and more rigid in the end. This is rather intriguing, it looks like the cube gets deformed in the first test and becomes more stiff in the repeated testing. It is like a metal, it has the same progress as cold working a piece of metal.

If you compare the first and the last result, there isn’t much changed. The pattern should be suitable for daily use for walking and running. The material shouldn’t be a problem as Adidas also uses TPU in their premium running shoes, Ultraboost and NMD for example.

A day of modelling is a day of frustration

Today we want to create a more interesting design for the shoe sole. To do that we need some inspiration:

We simply imported a picture of a shoe in Rhino and used the sole as a way of created a sole. The scan of our feet were then subtracted from our sole and we had a sole with the imprint of our feet in.

The new sole was then imported in MeshMixer to clean up the surfaces and make them more clean. This was also done to the scan before it was imported in Rhino.

The result was this:

We now need to work with Tim to create the different infill patterns on the basis of our 3D scans of the feet (the scans we refer to is the ones that shows were the compression is with varies tones of black and white).
This has to be done in Tim’s Cura, since it is not possible to do in the Cura that are available for the public.

Status from the first shoe sole print

The result from the first true shoe sole print still needs to be improved. The current print has the following challanges:
– A uniform infill everywhere of 25%. We want different infills in the print, so that the sole can be compressed more some places that others.
– A ugly design were the bottom and top had planar surfaces. We want a more organic design with no flat surfaces.
– Scaled model of 70% of the actual size, since it could not be printed in full size on the Ultimakers print board. We want a full size, so we can actually use the sole.

In the design team we have been wondering how we are going to use the sole. We don’t want to create an insole for our shoe, since we don’t believe that this uses the full possibilities of the 3D-prints options. We instead want to use the sole as a base and maybe lasercut textile to use for holding the feet.

We see the following shoes as a source of inspiration when it comes to holding the sole in place:



The Birkenstock would be easy to make with a lasercutter. Here we tried with paper:

Modeling the sole hardness

With the 3D models of out feet and clay prints from the last post, we were one step closer to our goal. Now we need to determine the required hardness of the final sole. We decided the best way to do this is to compare the model of our feet without a load (in mid-air) and our feet when a load is applied (in clay).

3D model of one of our feet

To make a good comparison we only needed the underside of our feet. This meant we could clean up our models and remove a lot of unnecessary data points from the .STL files. After cutting and trimming we are left with the bottom 20mm or so of the shape of our feet.

After trying some different programs we finally found a CloudCompare, a good package for comparing two similar 3D models. The program also contained an option for aligning the models as closely as possible. This is done by choosing recognisable points on both models, which CloudCompare then used to find the orientation which yields the lowest total distance between the points. This is followed by a cloud registration which does the same, but with every single point on the meshes.

With the models of the loaded and unloaded foot matched as closely as possible, the cloud/mesh distance is calculated. This assigns a value to each mesh depending on the distance to the corresponding mesh of the reference mesh.

Cloud/mesh distance in colour scale

Cloud/mesh distance in 8-bit grayscale

By comparing the two models the change is a measure of how much the tissue in the foot compresses. For practical purposes, the values are converted to an 8-bit grayscale. This is done to properly define the various hardnesses we hope to achieve in the final product.

Foot scanning

In the past week we have been in contact with Bertus Naagen who is a staff member at TU Delft and has a lot of experience with 3D scanning. He introduced us to the 3D scanner Artec Eva which is a scanner that is used for small objects.

We made a mold out of Clay and scanned each of our feet:

Bertus Naagen scanned the Clay mold:

Foot scanning


We also did a 3D scanning of the food itself:

The scanner created 12 STL files for each of our feet.