Process
The process was strongly dominated by our intuitive approach, at the beginning mostly carried by visual impulses.
Two worth-mentioning ideas at the beginning of the conceptual phase were a landscape, a visual representation of files—or more like a summary—that would be organized by time and contain relationships. This idea was partially based on the explorations from Task 1 in this course, where a 3D space with all the files extracted from the computer was visualized in a matrix of size, depth of directory, and access, which then transformed into birth time (as a result of the data extraction). This laid the ground for designing with time as the main compositional and organizational factor.
The second idea was the concept of gems that carried the notion of the deconstructed concept of a file— trying to grasp the idea that files are semantic entities and that it is possible to create semantic clusters around one main idea, and in doing so let go of the static nature of the file as we know it today.
As philosophically appealing as this idea was, it didn’t really work conceptually, but it still carried a strong visual impulse that led to the decision of following it regardless of the meaning behind it. We started reimagining what both images—a landscape and a gem—could mean from scratch and came up with the idea of translating the information behind a file, or within a file, into 3D geometry.
Form is information
The translation of conveyed information, is carried on two levels: on the intra level and on the inter level.
On the intra level, geometry is generated from data through rules. These rules can be universal, describing general properties of data, and at the same time specific to file types. For example, brightness in an image could define height along the z-axis, or in a design file, elements positioned higher on the canvas could appear higher in the geometry.
On the inter level, material and colour communicate this relational information. A change in colour can indicate changes between versions, and a glassy material marks the most recent version.
With this concept we opened up a new position of visual representation of a file on the scale of minimization or abstraction that stands between the icon as a strongly abstract entity and the preview as a direct, non-translated representation. When bringing in the new axis of informational value into the scale, we might again find ourselves in the same position, logically following the premise that with abstraction information is transformed into more global, broad ideas. The position of the geometrical representation withholds a very similar informational value as the preview, but differentiates in the actual type of information—it tells us about relationships to other files and version history, which previews do not, at least not without a reference. And so the preview and the geometry form a complementary composition that we tried to integrate into our prototype.
This composition begins to be even more interesting when bringing a third dimension into the matrix, which is the user distance, an expression summarizing informational value and abstraction. While the preview seems to be very near the user, the user basically places themself within the file, the icon stands very far away from the user. The geometry positions itself again somewhere in between, possibly nearer the preview, holding up the position for even further deconstruction of its parts and so leaning towards the preview and the potential for recognition and file-specific memorability.
An aspect that, following Philipp Steinacher’s feedback at the final presentation, could be interesting to test and explore further in the context of possible user integration, although our project should be seen more as an illustration of the requirements of modern file management and a visual exploration of a 3D space (leaning towards designing for XR devices) that provides and requires different visual solutions than the desktop.
First Prototypes
In order to explore the idea of 3D forming, the concept was with code in Blender that actually translated an image into geometry, using rules such as bright pixels being higher in space than darker pixels. This approach gave our idea more substance, but it did not manage to actually answer the core questions: how is this geometry really created, how does the translation work really work, and what specific rules does it follow?
Two options for answering these questions presented themselves. The first one was a translation based on entities which would result in a more illustrative visuality: one cube in 3D space would represent one other entity, such as a pixel, letter, or sentence. The other option would be more abstract and would work with actual byte sizes, so one cube would be one data chunk.
In the end, the question was not really answered successfully, and Alex Obernauer’s comment—that this is probably something engineers would additionally need to be asked to answer—gave space to see this idea more as a visual goal, or a vision.
The first prototypes (paper and digital) were focused on exploring different perspectives and ways of sorting the 3D space. The main struggles were the definition of time in this space and the classification of a timestamp according to it. Questions arose such as: does a file exist only once in the space? Do files exist on parallel levels? Do timestamps define moments when files were changed, opened, or created?
At the same time, we struggled with communication within the team regarding the at-that-moment existing relationship window and the role of the filters, as the process of letting go of the idea of the directory and traditional sorting methods was not easy to pursue.
Organisation is replaced by customised search and automated characterisation
The system is search-based, and so finding something must not be linked to the name (because this relates strongly to directories), but rather to time references, relationships, type.
And so two stacks of filters that allow users to customise the search in much detail, and to use relationships as a chain to move between files in time were developed. In addition, files can be grouped according to collaborative spaces and tagged more complexly.
The notion of nodes as an automatic feature allows the system to give suggestions and enables the user to manually create automated tagging and collaboration classifications. For example, files of a given type, created in a certain time span, or within a certain group, can automatically be tagged in a specific way or sorted into a particular group space.
Retrospect
The project emerges from navigating between two opposing impulses, turning the prototype into a form of mediation between them—though it ultimately leans more strongly toward one side. On the one hand, there is the need for user-friendly, customizable, flexible, and adaptive systems that can shape themselves around individual personalities. On the other, there is the drive for new visual languages and new solutions—an impulse that often requires a kind of destructive energy to break away from existing models and is characterized by very narrow usefulness.
This seemingly arising tension is not easily resolved. In many cases, and particularly in formfinder, the process results in a design that is powerful in a conceptual and illustrative sense, yet more distant from the opposite pole of being user-friendly or even practically developable. Although attempts are made to mediate this gap through ideas such as nodes, references to Apple’s UI, and filtering systems, the geometric logic of the 3D environment does not yet feel fully anchored.
At the same time, the project evokes a compelling vision—a future possibility. This raises a central question: does the development of modern systems require visionary prototyping in the first place, opening conceptual space ?
And yet, even when working speculatively, usability and logic cannot be abandoned. They are what bridge the initial visionary gesture with the possibility for testing and development, transforming abstraction into something navigable. Without them, the path remains symbolic rather than actionable.
With formfinder, this mediation became tangible. It clearly stages how intertwined these aspects are, appearing only as opposing tensions while in fact also being dependent on one another.
