How we 3D printed mini mannequins to explain clothes size labels

The idea for the Sept 12 interactive graphic Inside The Confusing World Of Women’s Clothing Sizes stemmed from project lead Stephanie Adeline’s personal frustrations with shopping for clothes. After moving home to South-east Asia from the US, she found shopping to be really stressful.

The clothes readily available in South-east Asian shops are much smaller than those in US ones. Also, the size of outfits labelled with the same alpha sizing (like S, M, L) may vary from store to store. For example, a size M top might fit perfectly in one shop, while an XL one is needed in another.

There was only one way to solve this mystery: by diving into the data.

We knew that it would be a largely data-driven story, where size guides would need to be collected – but it would also be a human-centred one, involving real women and their lived experiences.

We kick-started this project with a painstaking process of collecting size charts from brands’ websites. We decided on eight brands – COS, Cotton On, H&M, Love & Bravery, Love, Bonito, Lululemon, Sandro and Uniqlo – based on their popularity in Singapore and availability of alpha size charts. Because the data gathering was done manually, we took extra care to ensure accuracy. When a brand did not have a general size guide page, we carefully cross-checked individual product pages to verify the body measurement charts, confirming that the numbers were consistent across multiple items.

After gathering the data from just a few brands, we realised our hunch was right. A size M at H&M is an L at Love & Bravery, and an XL at Uniqlo.

From there, we set out to answer two key questions: Why is this happening, and how is it affecting women?

To help answer them, we began by learning from the lived experiences of real women. We created an anonymous survey to gather insights into their shopping experiences – covering the range of sizes they wear, the fit challenges they encounter, and the frustrations they face with inconsistent sizing. The responses revealed a common pain point: Many women find themselves fitting into clothes with different alpha size labels, depending on the store. We explained this frustration through the story of Shannon – a hypothetical example inspired by our survey respondents.

To dig even deeper into these frustrations and better understand the pain points of finding the right fit, we also designed a series of shopping experiments. For this, we roped in volunteers from the newsroom: Hannah, Irene, Shabana, Friday and Matilda.

This story began with a personal frustration, and both our lived experiences and reporting make one thing clear: every woman’s body is unique.

Our visual story therefore needs a strong “me layer”. It should serve two purposes – first, to remind women that there is nothing wrong with their bodies, and that clothes are meant to fit them, not the other way around; and second, to provide a useful tool that helps them shop for clothes with greater confidence.

While brainstorming how to visualise the data, we came across this body visualiser project, where the 3D model would morph into different body types based on user input. That sparked our idea to show how the size charts would look like, visualised as an actual human.

Clothing sizes are typically based on three main measurements: bust, waist, and hip. These do not capture full body proportions, and we state that clearly in our interactive graphic. To simplify things, we focused on the three main measurements as the variables and kept the other aspects to be somewhat average looking. Our model’s height was set to 160cm, which is the average height for women in Singapore.

We decided on the smallest and largest measurements that we can reasonably cater to, created a base sculpt of the smallest possible woman with our measurements, and additively sculpted three separate shape keys to represent each of the three measurements. We used Blender, a free open source 3D programme, for all modelling tasks.

We then used shape keys to adjust the measurements of the woman. A 3D mesh is made up of multiple points in space or vertices. Shape keys store alternate positional values of these vertices that we can then interpolate between. If we move points away from one another, we increase the volume of a mesh. This forms the basis of how we can morph our human model from smaller sizes to larger ones, and vice versa.

While we kept the effect of the bust and hip keys to be fairly localised to the breasts and upper thighs respectively, we decided to use the waist shape key as a way to globally adjust for fat levels, giving us the ability to thicken other regions that are not accounted for within the three main measurements, such as the arms and the face region.

The model was then hand-painted in Photoshop using a watercolour-inspired art style to create a friendly and welcoming impression, with the intention to encourage our readers to interact with it.

Earlier in the ideation stage of this project, once we had a base 3D model that we could play around with, we considered the idea of 3D printing mini mannequins. At first, we thought, “No way, that’s too ambitious”, especially since we were busy with the 2025 General Election at that time. But then, developer intern Tang Hao Liang casually blurted out: “Oh, I have a 3D printer at home.” The next day, he showed up with our first prototype, Shannon 1.0.

Hao Liang went on to test different printing materials by printing out dozens of animals that represented each team member’s spirit animal. Finally, we settled on a resin that looked more like that of a mannequin, without the typical 3D-printed look.

These 3D-printed mannequins had another purpose beyond transforming data visualisation into a tactile, tangible form. We also wanted to observe the reactions of real women – now that they could see brands’ sizing charts represented in real life, how would they perceive themselves?

We laid out the mannequins on a table, then asked our colleagues to guess which doll represented themselves best. The results were expected but shocking. Most of our five participants could identify mannequins that matched their body shapes, but when it came to size, many chose dolls that were bigger than their own measurements.

Here was another way we were making this abstract problem tactile: our life-sized print graphic.

We wanted the size charts to be placed in real scale on the page – for readers to be able to compare their body measurements against the print. This required multiple rounds of checking to ensure the accuracy of the scale. The final print graphic included a small measuring tape for readers to measure themselves, in case they did not have a physical one on hand.