Integrating Feedforward into Data Physicalization authoring tools

I led the end-to-end UX research and design of DataCuts, an authoring tool for kirigami-based data physicalization (KiriPhys). Our goal was to implement feedforward mechanisms—predictive design cues— in an authoring tool and evaluate how it can support speculative, forward-thinking workflows.

Simplified example of Physicalization Design Process with DataCuts. A: data understanding and exploring, data organizing, and data mapping. B: transferring the design to the pattern using the tool. C: laser-cutting the physical artifact (realization).

Motivation

Unlike digital visualization, physicalization design introduces structural and material uncertainties that require immediate feedback and rapid iteration. This makes physicalization design speculative, labor intensive, and expertise dependent. Yet authoring tools remain scarce, with limited computational support to help anticipate outcomes.

Goal

In this work, we explore how feedforward mechanisms (predictive features) in an authoring tool might help physicalization designers anticipate post fabrication qualities like tactile feel and expandability before fabrication. We do this through the iterative development and study of DataCuts, an authoring tool for designing Kirigami based data physicalizations.

Methodology

We iteratively developed DataCuts and evaluated it through two studies:

An early formative study on low-fid prototype to surface the need for the authoring process,
and a follow-up evaluative study on the developed tool to observe how designers use feedforward while designing data artifacts.

formative study

to surface painpoints and authoring needs

To design the first figma based prototype, we drew from (1) an autoethnographic look at our own data artifact authoring process, (2) existing visualization tools, and (3) iterative co design sessions with colleagues experienced in visualization and interface design.

The Figma prototype used in The formative study: five-step workflow in the early prototype. A - Landing page; B - Selection of a pre-existing dataset or upload of a custom dataset; C - Data preview; D - Pattern design; E - Pattern finalization with options for download and customization.

formative study procedure

key results

The formative study revealed several needs in the physicalization authoring process, such as support for non linear workflows, tighter connection between data, mappings, and previews, and fast in tool iteration. Most importantly, participants consistently emphasized the need for anticipatory support to understand how their on screen designs would behave and feel after fabrication.
Through the in depth interviews and co-design sessions, this idea of predictive guidance gradually crystallized and we later named it feedforward, which directly informed our focus in the next version of DataCuts.

2. Evaluative study

to assess the developed authoring tool and evaluate the feedforward feature

The web-based authoring tool we developed features three panels, including a dedicated feedforward panel:

Controls Panel
Includes the interactive features needed to define a design. Users can select and switch datasets, set up data mappings, and see the pattern update as mappings change. Drag and drop mapping supports quick authoring, including applying a mapping across multiple rows. Re scaling can be applied through column headers. The panel supports free iteration on mappings, shapes, and property definitions, with no fixed sequence.

Pattern Pannel
Shows the designed pattern in a ruled area that corresponds to its real world size, with a legend. The preview is kept in 2D to support speculation about structure, material properties, and interaction qualities through design decisions rather than relying on a predefined 3D view.

Feedforward Pannel
A dedicated space for anticipatory support, helping users anticipate post fabrication qualities and better understand physical and material aspects before fabrication. It combines action based hints with physical indicators that summarize key qualities such as size, countability, expandability, and tactile qualities.

The workflow of creating a representation using the authoring tool

Methodology

Study 2 examined how designers used DataCuts and its feedforward features during a concrete design task.

After a brief introduction to physicalization and the tool, participants engaged in a think aloud design and making session where they created a data physicalization in DataCuts while considering its future fabrication. The designed artifact was then fabricated and given back to them, so they could reflect on how the anticipated outcome based on feedforward matched or differed from the physical result. This was followed by a semi structured interview and reflection on their experience with the tool and the role of feedforward in their decisions.

key results

on

feedforward

Our results are framed through participants’ interactions around feedforward. Overall, we observed:

Participants heavily relied on feedforward to map data to physical properties, using the indicators to understand tangible properties before fabrication and refine their expectations and design decisions.
Feedforward encouraged tactile representation over conventional mapping, prompting several participants to move from standard visual variables toward more interactive, sensory-rich designs, and it affected their mental model of mapping, shifting attention from low-level cut parameters to higher-level tactile variables.
Trust in feedforward increased confidence, enabling more adventurous yet “on track” exploration, but mismatches between indicators and fabricated outcomes also exposed limits in how far participants felt they could rely on the current indicators.

Page updated

Google Sites

Report abuse