On a screen, moving a cup’s rim a little further outward takes a change to one line. In ceramic, it means forming it again, waiting for it to dry, trimming and firing it. Only after picking it up and drinking from it do you know whether you still want to make another change.
When developing the predecessor to Coffee Stage Sharing Cup, we were looking for a more continuous, comfortable feeling when taking small sips of coffee. If we made every angle in ceramic first, we would spend our time and budget before trying many of the possible directions.
So we wondered: could we move some of that exploration onto a computer first? With digital modelling, physics calculations and AI, we could at least turn ideas in our heads into shapes we could see and change, then choose which directions were worth taking further.
That possibility was what drew us in. The feeling of a sip, usually so hard to describe, could also be explored through the angles of a cup, the liquid surface and movement.
The first questions arrived before we had made a cup
It is easy to say we want something “comfortable to drink from”. Drawing the cup means making choices.
Where should the body curve inward? How long should the flared section be? If the proportions of the opening and body change, how much room remains inside? A shape may look right from the side, but does it still look pleasing from the front, and is it easy to pick up? A broad feeling began to become a set of questions to compare one by one.
At the time, we used Claude, ChatGPT and digital modelling tools such as Blender to explore angles, proportions and simulation effects. We then used Claude to develop a fluid simulation environment. We tried several digital versions and made more than ten physical prototypes before gradually arriving at the later direction.
Left to right: S Size, M Size, L Size, Alpha Size, Beta Size.
This describes the development of Sharing Cup’s predecessor. The five current outlines in the diagram help show the series’ shapes; they are not an archive of every prototype from that time. Looking backwards from a finished piece, it is easy to feel the answer was there from the beginning. While making it, even which proportions to keep for the next step was a decision still to be made.
When liquid moves on screen, ask what question it answers
An animation of flowing liquid is compelling. The cup tilts and water moves along its wall. It already feels close to actual drinking.
But if the aim is simply to compare where the cup body curves inward, the other conditions should stay as similar as possible. Otherwise, if one version contains less and is lifted slowly while another contains more and turns quickly, where does the difference we see actually come from?
It helps to think of a model as a question with a limited scope. First say what will change, then decide which result is worth comparing:
| What you want to understand | Conditions to state for the comparison | What you still need to know beyond the image |
|---|---|---|
| When coffee approaches the rim | Internal cup shape, amount poured and tilting movement | Whether you are only looking at its appearance or recording when liquid flows out |
| How liquid flows out after reaching the rim | Liquid settings, edge shape and how surface contact is handled | Whether there is a flow rate to compare, rather than just an animation |
| Whether small sips feel comfortable | The rim, how the cup is held and the person’s movement | You still need to touch and drink from the physical cup |
This is one way to read a model, rather than a checklist of validations we completed at the time. We do not currently have enough information to set out the specific contact coefficients, liquid properties and solver settings used then, so we will not fill in a set of professional-looking numbers.
More important than a tool’s name is which question it helped narrow down. Being able to change a shape on a computer does not mean knowing how it will feel against the lips. A physical object still stands between the two.
A cup study narrowed the question to “how much flows out”
A 2021 study of a controlled-flow drinking cup put cup shape, liquid volume and tilting movement into a model to observe outflow and changes in flow rate. Its simulations used water, and the subject was a transitional drinking cup for children. Comparison with results from physical cups remained future work. Read the original research
This example makes “simulating a cup” more concrete. It did not begin by answering whether the cup was good to use overall; it first defined which flow results to observe. Readers can therefore see how far the model went and what kind of comparison was still needed next.
NASA’s guidance on simulation validation also stresses assessing a model against experimental data within its intended use. For a cup, reproducing a particular flow and explaining whether a sip of coffee feels comfortable are questions of different scope. View the validation guidance
We like digital exploration because it can bring choices that were once vague into view earlier. If a shape is worth trying further, we can put resources into it. When the question becomes contact with the lips, feel in the hand and actual drinking, the physical object takes over.
From the top left: ① at rest; ② liquid surface near the waist; ③ front edge reaching the flared rim; ④ entry position.
Digital proportions must work inside and out in a physical cup
The real difficulty later was turning digital proportions into a usable cup, then making the next one.
A cup is more than a single line in profile. Between the waist you see outside and the contour touching the liquid inside lies the cup wall. Wanting a thin rim does not mean we can simply make the whole model thinner. During forming, removal from the mould and trimming, the physical piece must be in a condition that can withstand the work.
In conventional slip casting, a wall gradually forms inside a water-absorbing mould, then the excess slip is poured out. See how the forming process works So the unfired cup in front of us is more than an outer shape. Its internal space and wall thickness also need to be handled during production.
The more than ten physical prototypes should therefore be understood as more than “proof that the computer calculated correctly”. They were also answering a more practical question: how can the chosen shape be formed, removed from the mould and still retain the contour we want?
One reason we chose moulds and slip casting was to narrow the differences between cups. Finding one prototype we like and continuing to make a batch in that direction are two different challenges.
We are still asking about that feeling when drinking coffee
Before this series, we found it hard to connect the feeling of drinking coffee with physics and digital modelling. Once we started making it, the question in our minds became clearer: why does coffee offer so much variation? Of the sip we enjoy every day, how much comes from the drink, and how much from the way we use the cup?
We are still learning, and there is plenty we cannot yet answer. But rather than label a simulation image “scientific”, we would like to share the exploration so that the person picking up a cup also knows the thinking behind its curve.
So when friends ask why we do not simply throw the cups on a wheel, the answer continues from here. We care both about making one cup look right and about how to carry the curve we want into the next cup and the next batch.
