All posts by jjweiler

3D Print-Based Silicone Mold Culinary Study

This is a continuation of  Melting Materials for Mold Making, where we describe some of our experiments to create molds of wax, chocolate, and jello using 3D printed models and silicone molds, and 3D-Designed Molds for Baking and Freezing, where we experiment with baking and freezing food using silicone molds.

Our focus is using 3D prints to fabricate molds for culinary exploration. To determine what types of 3D designs and recipes work well to create customized, detailed dishes, we held a workshop with culinary enthusiasts.

Participants were invited to attend a workshop, which introduced them to our software system and workflow for generating 3D food molds. Over the course of the following week, they submitted drawings and photographs to be converted into 3D prints by our system. The participants then experimented with different recipes in their own homes, and kept in touch with the group by sharing their designs and recipes through a private group on a social network. During this time, they also had the option to create additional designs, and those were 3D printed and made into silicone molds for them to experiment with.

Types of Molds and Designs

The most common participant requests were to make multiple silicone molds of each print, create interconnected designs, and fabricate additional silicone molds of household items.

Several of the participants requested the option to make multiple silicone models of each 3D design. While it takes 2-3 hours to create one of the 3D prints, each silicone mold can be made in about thirty minutes. As such, participants were able to make several silicone molds from a single 3D print. This is a clear benefit of using molds over directly 3D printing the food, since having multiple molds allowed the participants to have several copies of the food design made simultaneously, whereas a 3D printer can only create one copy at a time.

Participants also noted the usefulness of interconnected designs. Such designs are beneficial because they allow relatively simple designs to be multiplied into complex forms, and, by changing the number of molds used, allow meals to be scaled to the needs of the person cooking.

below: examples of interconnected designs

p1-1-cornbread1.jpg  fan-1-silicone.jpg  model3-PB&J popsickle.jpg

In addition to making molds from 3D prints, two participants also made silicone molds of household objects. The downside of deeper shapes is that they limit the types of food that can be molded. In order to remove the original reference object, the silicone mold had to be cut in half and then pressed together when the food mold is being set. While this works for thick batter or melted chocolate, participants found that materials like liquid gelatin or egg whites will leak out through any cuts in the silicone mold before they have time to harden.

below: example of molds made from household objects

p5-3-silicone.jpg  p5-4-silicone.JPG  p5-3-chocolate1.jpg

Recipes and Food Experiments

Most of the participants focused on single-ingredient foods that could easily transform from a liquid to a solid state, such as chocolate, egg yolk, gelatin, pancake, and flan. As our participants discovered from their experiments, other materials, such as wonton wraps, can also be shaped in the models, though they require the use of simpler molds composed of smooth surfaces.

below: example of wonton wraps

p2-1-silicone1.jpg  p2-1-wraps5.jpg  p2-1-wraps6.jpg

For designs, participants suggested using food appearance and shape to encourage diners to make healthy choice. In addition, they were interested in using the shape of the food  to confuse or intrigue the diner as to what taste they may encounter.


Overall, our participants’ experiments revealed that molds with smooth surfaces worked well universally, whereas molds with fine details worked best with frozen and gelatin based foods. Hot foods were the most problematic, as they are often soft and difficult to remove form the molds. Depending on the ingredients, it may be more effective to freeze the meal into the mold, remove it, and then re-heat the food.

Future Opportunities

In the future, 3D models can be tailored more specifically to the foods they are applied to. For instance, our software might be altered to preview several different 3D models from one 2D image to show variable levels of detail and depth. Each 3D model could then be customized to maximize detail based on the specific attributes and limitations of the different foods being worked with. That way, culinary enthusiasts could visualize and compare what the finished dish would look like depending on their design and choice of ingredients.

below: examples of same molds being used to make different foods. In the future, models could be generated to best serve different food materials

p6-1-shortbread cookie4.jpg  p6-1-egg2

fball-poached eggs2.jpg  fballs-agar2.jpg

Since our participants enjoyed and appreciated the social-sharing aspects of this study, it could be beneficial to create a broader social sharing platform to aggregate 3D designs and recipes, thereby scaffolding a broad base of knowledge to advance and expand what food enthusiasts can create.

In addition, our approach offers insights for developing future high fidelity food-based 3D printing technologies. For example, our study shows that there is a definite interest in providing healthier options, as well as a desire to create several portions simultaneously in order to facilitate a shared dining experience. This indicates that future food 3D printers could focus on offering expanded food options outside of sweets and treats, and explore ways of generating food that encourages a communal, rather than isolated, dining experience.

Since food 3D printing technology could potentially become ubiquitous in future years, it would be prudent to make sure the technology does not inhibit, and hopefully tacitly encourages healthy eating and social engagement.



3D-Designed Molds for Baking and Freezing

This is a continuation of Melting Materials for Mold Making, where we describe some of our experiments to create molds of wax, chocolate, and jello using 3D printed models and silicone molds.

Here we are presenting new additions to the model making software and further experiments with different types of food.

3D Model Generator Additions

The program we have been using to generate our models works by taking a black and white 2D image and transforming it into a depth map, where the lighter parts of the image are raised up and the dark parts are lowered. In order to save on time and material costs for the 3D printing, we have also made the models hollow in the back.

Below-left: photograph of Antonio Canova’s Bust of Venus Italica. Center: bas-relief 3D model generated from that picture. Right: back of the 3D model

antonio-canova-bust-of-venus-italica.jpg  Untitled.png  Untitled 2.PNG

In addition, we created a version of the program that uses a color signifier (in this case, red) to subtract part of the image from the finished model. This way, the resulting model will not be limited to the rectangular dimensions of the original 2D image.

Below: model generated using the version of the program that subtracts red space. Left: Antonio Canova’s Bust of Venus Italica with red background. Middle: generated model. Right: back of model.

antonio-canova-bust-of-venus-italica copy.jpg  Untitled 3.PNG  Untitled 4.PNG

As in Melting Materials for Mold Making, silicone putty is used to create a negative of the 3D model. All food will then be cast using the silicone putty mold and will have no direct contact with the 3D print. This is because 1) the flexibility of silicone makes it significantly easier to remove molds after they have hardened, and 2) while we are using food safe 3D printed materials, the temperature limits of 3D printed material food safety is not entirely known. For the following food tests, we specifically chose to use Silicone Plastique putty, since it is food safe and can withstand temperatures up to 450 degrees Fahrenheit.

Baking Tests

Before each baking test, the silicone mold was throughly washed and sprayed with cooking spray.

Sugar cookie – We found that Pillsbury sugar cookies (oven, 350 F, 12 minutes) did not closely stick to the mold, largely because of air pockets that formed in the cookie. Below-left: silicone mold. Right: sugar cookie.

model2-sugar cookie.jpg

Pancake – While we could not get a complete result with the Aunt Jemima pancake mix, we were able to get some promising details in the pancakes, and further experiments with cooking time / temperature / pancake mix could likely result in a functional pancake mold.

Below-left: (oven, 375 F, 12 minutes) Pancake was still gooey

Below-center: (oven, 375 F, 17 minutes) Pancake was fluffy, though still slightly undercooked. Part with detail (hair) stuck to silicone mold

Below-right: (oven, 375 F, 12 minutes) Significantly less batter was poured into the mold with the hope that it would cook faster. This was successful, and the resulting pancake was fully cooked. Part of the pancake was stuck to the mold, but some nice detailing (bun and part of hair) was successfully preserved.

model2-pancake1.jpg  model2-pancake2.jpg  model2-pancake3.jpg

Eggs – The eggs cooked fairly evenly in the oven and were overall easy to remove from the mold without causing any damage. They were also successful in capturing details from the silicone mold.

Sunny side up (oven, 350 F, 12 minutes)

Below-left: sunny side up egg still in mold. Center: egg removed from mold with yoke still intact. Right: yoke broken open

model2-egg1-1.jpg  model2-egg1-2.jpg  model2-egg1-3

Whisked egg (oven, 350 F, 12 minutes)

Below-left: whisked egg still in mold. Right: egg taken out of mold. Part of the egg was still slightly gooey, which caused a chunk of the hair to become stuck to the silicone mold.

model2-egg2-1.jpg    model2-egg2-2.jpg

Freezing Tests

Liquid was poured into the silicone mold and then placed in the freezer overnight. Overall, the frozen models were the most successful in capturing fine details from the silicone mold.

Below-left: ice (frozen tap water). Right: popsicle made from Bolthouse Farms breakfast smoothie.

model2-ice.jpg    model2-popsickle-1

Designing for Extreme Heat

In the wake of global climate change, our world is projected to experience more extreme heat waves over the next few decades.

Phoenix, Arizona, where this research was conducted, is one of the hottest locations on the planet and presents a testbed for understanding and addressing heat-related challenges. This research focuses on adaptation as a design strategy that compliments existing approaches to mitigate human impact on the environment.

We held a summer-long diary study that helped us to understand how extreme heat impacts human lives and how participants cope with extreme heat.


Above: Data from our diary study of extreme heat: thermal camera image captured by a participant and participants’ journals

These findings motivated our critical making work themed around adaption, focusing on artifacts for visualizing, coping with, and utilizing extreme heat. In constructing these artifacts, we were able to critically reflect on both the benefits and drawbacks of designing for adaptation.


Above: Solar Cooker made from re-purposed materials


Above: A sensor-enabled hot composter deployed outside


Above: Solar-powered chiller


Above: “Phoenix, a survivor’s guide” is designed to provide local knowledge and resources to the uninitiated in surviving the extremes of the desert climate. The survival guide is intended as a low-cost, DIY style, self-printed zine to be distributed amongst vulnerable populations.


Above: Visualizing extreme heat: screenprinting with thermochromic ink and a paint-based heat visualization

To see the full paper, click here.

The paper will be presented at the International Symposium for Electronic Arts (ISEA 2017).

Melting Materials for Mold Making

While 3D printers have become more accessible, it is still relatively expensive and time consuming to generate 3D prints. In addition, the materials commonly available for 3D printing are limited to certain types of plastics. We wanted to explore the possibilities of broadening the affordability and material variety for making multiple models by using traditional mold-making with 3D printed sources. This way, by 3D printing a single model, users would be able to create multiple finished molds using a variety of materials.

Crayon – Solar Heating

As an early experiment in melting materials, we used solar heating to melt Crayola crayons into different shapes. Crayons have a relatively low melting temperature, becoming completely molten at between 120-150 degrees fahrenheit, which is an easy temperature to reach while sitting outside in Phoenix, Arizona during early Autumn.

left: fused crayons after heating; middle: cut up pieces of crayon before being heated; right: pieces of crayon fusing as they melt

Wax – 3D printing tests

Since we wanted to make these models useful for molding several different types of materials, including food, we wanted to insure that the 3D printed models would be food safe. The main ways to insure that a 3D print are food safe are to 1) use a food safe printing material (some types of PLA are food safe, but it is important to check with the manufacturer), 2) use a 3D printer that has a stainless steel extruder, 3) wash the 3D model with antibacterial soap, and 4) spray the model with a polyurethane spray to prevent the risk of bacteria growing in small cracks in the 3D print.

For testing the viability of the 3D models as molds, we used wax as a test molding material, since wax is solid at room temperature but can melt at between 110-150 degrees fahrenheit (depending on type of wax), and so is easy to melt using normal household items.

The first 3D print prototype used a raised image inside of a hollow cube, with the hopes that the wax could be poured in when hot and would be removed easily once hardened (a no-stick spray was applied to the 3D print before the wax was poured in). However, it appeared impossible to remove the wax from the model intact. By putting sheets of saran wrap between the wax and the 3D model, it was possible to remove the wax once dried. However, the resulting wax molds were unable to properly adhere to the shape of the 3D model because of the presence of the saran wrap.

left: first 3D printed model; right: the only way to remove the wax molds in one piece was to place a layer of saran wrap in between the plastic model and the wax as it melts. However, that resulted in the wax not adhering to the shape of the 3D model well.

After the first model proved ineffective for casting wax, we created a second model which had a hollow bottom, in the hope that the wax mold could be pushed out of the model once it had hardened. This did succeed, but it was still fairly difficult to remove the wax mold and there was some surface abrasions to the wax.

left: liquid wax cooling inside the second model. The model was placed on top of wax paper to prevent the liquid wax from leaking out from the bottom of the model. right: resulting wax model. While a marked improvement from the previous attempts, it was still difficult to remove from the 3D print and had sustained minor damage.

Wax – Silicone mold

Because the wax was proving difficult to remove from the inflexible plastic 3D prints, we chose to look for possible in-between methods of transferring a 3D print into a molded material. We decided to use silicone putty, a material that starts out with a texture resembling play-doh but will solidify into a permanent shape while still maintaining flexibility. To create the silicone mold, silicone putty was spread around a 3D print and left to dry. Afterwards, hot wax was poured into the silicone mold, whose flexible properties made it much easier to remove the wax after it hardened.

left: original 3D print; right: silicone mold that was created from dried silicone putty that was molded around the 3D print; bottom: resulting wax piece that was cast in and removed from the silicone model. Despite being much thinner (and therefore more fragile) than the previous wax molds, the wax piece was removed from the silicone with significantly less damage than the wax tests that were cast directly in the 3D prints.

wax-silicone mold.jpg

Given the success of molding with wax, we did some preliminary experiments with food materials using the same silicone mold, which was created from a type of silicone putty that has been designated and labeled as food-safe.


The chocolate tests were mostly successful, with the only notable problem being that the surface detail of the chocolate molds appear to have lumps or pockets in their surfaces. This may be caused by air pockets being stuck under the liquid chocolate as it is poured into the silicone mold, or the type of chocolate that was used (standard chocolate chips) is not designed to remain smooth after being melted and re-solidified.


The success of the gelatin molds varied based on the type of gelatin that was used.

below: When using Jello brand gelatin mix, the resulting gelatin did not maintain its shape when being removed from the silicone mold.


below: Alternatively, coffee agar mix (which is intended to be cut up into cubes or other shapes after hardening, and thus was designed with greater internal resilience than Jello) was easily removed from the silicone mold without any damage.

Crafting colorful objects: a DIY method for adding surface detail to 3D prints

Even as 3D printing technology continues to advance, there are still limitations as to the availability of detailed color 3D prints. The best commercially available 3D printers use binder-jetting (where colored ink is used in the construction process of the object), which results in a fuzzy, diluted image on the object’s surface.

As an alternative, it may be possible to create 3D prints with detailed surface color by pasting 2D printed images onto the surface of mono-color objects. In order to turn flat, 2D images into shapes that can fit onto uneven 3D forms, I created a program that would alter the proportions of the image so that it can be printed, cut, and pasted in such a way that the image will be evenly displayed over the 3D object.

right: original image;  left: image morphed to fit orb


below: test pasting image onto orb

print test_paper107

below: 3D printing tests, using symmetrical shapes

bendy cylinder bare

below: 3D prints – symmetrical shapes and Russian nesting doll prototypes after paper images are glued onto them



In order to get outside perspective on whether this approach would be feasible for an average user, we held a small workshop in order to gather feedback. For this case, the workshop attendees would be evaluating the process of cutting out images and pasting them onto the prints. The actual 3D printed models were printed ahead of time (since the 3D printing process can take too long for the span of a workshop). All workshop participants were presented with a set of 3 Russian nesting dolls, and were requested to bring in their own image files to be pasted onto the objects.

below: example set of 3D printed Russian nesting dolls and printed paper images before images are cut and applied to dolls


below: workshop pictures


On Cutting Out Paper Shapes – workshop attendees could choose between using scissors or exacto-blade. While all the attendees were able to cut out the shapes, some found the process tedious. Based on this feedback, future endeavors will focus on using laser cutters or similar tools to cut out paper and avoid unnecessary human labor.

On Pasting Paper onto 3D Forms – participants varied how they applied the paper onto the 3D prints. Some brushed the paper with gel medium before applying it to the 3D form, while others lathered the 3D form with gel medium before applying the paper. Both approaches were successful, though it seems applying the gel medium directly to the 3D form was faster.

below: if an extra layer of gel medium is not put on top of the paper, it does not have a glossy finish. While some attendees preferred the matte finish, it does leave the exposed paper at greater risk of abrasion or damage. By using an alternative gel with a matte finish, it may be possible to provide protection to the paper while not giving the object a glossy finish.


The results of applying the paper to the 3D prints were mostly successful. However, in some cases the paper ended up crinkled or not correctly matched up. For participants who completed multiple Russian Nesting Dolls, the results appeared smoother after their first attempt.



Certain types of printed images hid the cuts in the paper better than others. Generally, large areas of a single color (especially white) clearly displayed where the paper had been cut. In contrast, designs with bold and complex patterns mask the location of the paper cuts.

below: large white area clearly shows where cuts in the paper occurred


below: bold and complex patterns make the paper cuts much less visible


The most consistent problem was applying the paper in such as way that the “head” was completely covered. For many, this resulted in the black 3D print being visible between the strips of paper.

As a potential alternative, we used a different algorithm that would allow the paper to be cut into rectangular strips, as opposed to the curving triangles in the previous models. While this did allow the paper to completely cover the form, the extra overlapping pieces of paper made gluing more difficult. In addition, a fault in the algorithm caused additional warping of the image. Overall, the curving-triangle method appears more successful, although an extra buffer of paper should be added to the design so that the paper will completely cover the form.

left: Paper cut into strips. Overlapping pieces of paper bunch up slightly. No underlying part of the underlying object is exposed.

right: Paper cut into curved triangle shapes. Near the head, part of the underlying object is exposed.


Participants in the study suggested many possible uses for this technique, including fine art pieces (such as vases), figurines & game pieces, and biological medical models.

Going forward, we are looking to expand the capabilities of the program, so that it will be capable of morphing images to more complicated 3D models. Ideally, this would allow a user to design a complex form in the program and then map a 2D image onto it.