What if you could design a real, physical object — not by drawing it, not by clicking, but simply by looking?
That question has been living in me for some time... 
This article is the story of how two very different projects, both called (almost) the same name, collided in my imagination and grew into a concept I now call SYNTHESIS: interactive evolutionary design in virtual reality, with human attention — measured through eye-tracking and heart rate — acting as the fitness function, and a 3D printer waiting at the end of the loop.
Two sparks walk into a mind
The first spark came from an unexpected place:
The ELViRA project at Trier University of Applied Sciences — E-Learning with Virtual Reality Applications. Their engineers built VR environments that make complex industrial processes tangible for students: instead of reading about a production line, you walk through it. 
What struck me was not the technology itself but the quiet philosophical shift underneath it — knowledge stopped being something you look at and became something you stand inside.
The second spark was older.
In 2007, software engineer David Oranchak presented an evolutionary art system at GECCO, the Genetic and Evolutionary Computation Conference (full paper, PDF · his project page). His idea was elegant: computer-generated images carry a “genome” of parameters; the ones judged most beautiful “breed,” mutate, and evolve. The human is not the artist and not the audience — the human is the fitness function, the force of selection in a small digital ecosystem. 
And Oranchak went one fascinating step further: he explored whether that human judgment could be approximated — in his case, derived from the collective taste of a social network like Flickr.

Two sparks.
One gives virtual worlds a pedagogical body. 
The other gives human taste an evolutionary engine. 
For days I couldn’t stop asking: what happens if you put them in the same room?

The concept: SYNTHESIS

Here is the scene I keep seeing.
A student steps into a VR gallery. Around them float thirty or forty parametric 3D forms — brackets, handles, bionic structures — each defined by a small genome: curvature, thickness, weight, texture. The student doesn’t click or rate anything. They simply walk and look.
The system quietly listens. Eye-tracking notices where the gaze lingers. A wearable sensor registers the small quickening of a pulse when a form resonates. The forms that hold attention become parents. Their genomes cross — half from each — with a whisper of random mutation, and a new generation appears in the air. The forms nobody looked at fade away.
Ten to fifteen generations. Perhaps half an hour. One form remains — the one that best fits this particular person’s sense of what is right.
And then my favorite part: the file exports as STEP/STL and goes straight to a 3D printer. The student removes the headset and holds, in their hands, a physical component that grew — quite literally — out of their own attention.
Where attention rests, form is born.
Underneath the poetry: solid ground
I care deeply about the poetic dimension of this idea — but I care just as much about its scientific honesty. 
So let me name the hard questions myself, because they are not flaws in the concept; they are the research:
Does looking mean liking? No — not automatically. People also stare at what confuses or unsettles them. Whether implicit biometric signals (gaze dwell time, heart-rate variability) can serve as a valid aesthetic fitness function is an open research question, and SYNTHESIS is designed to test it against explicit human ratings as ground truth.
Can we beat evaluator fatigue? Interactive Evolutionary Computation (IEC) is an established research field, and its classic weakness is well known: humans get tired of judging generation after generation. One promising direction is a surrogate model that learns from the biometric stream and pre-filters candidates — echoing the question Oranchak explored two decades ago with social-network data: can the human fitness function be approximated? SYNTHESIS proposes biometrics as exactly that approximation.
How does a 2D art genome grow a 3D body? Encoding parametric 3D geometry — with real structural constraints — into an evolvable genome is a beautiful, unsolved engineering puzzle.
And the ethics? Biometric data is intimate data. Any real implementation of this concept must treat GDPR-compliant, consent-based, minimal data handling as a foundation, not a footnote.
Why this matters beyond the lab
For me, SYNTHESIS is more than a technical pipeline...
It is a small, working demonstration of a principle I build everything at Delvira Art n Tech around: 
what you focus on is what you get — made visible, testable, printable.
We spend our lives being told that our inner world and the outer world are separate. 
Here, for half an hour, they are not: taste becomes selection, selection becomes form, form becomes matter. 
The algorithm is not a replacement for human creativity — it is a mirror and an amplifier of it. 
A student walks out of the lab not just with CAD skills, but with a physical proof of their own creative attention.
That is the kind of technology I want to exist in the world: tools that don’t just serve people, but show people to themselves.
What’s next?
SYNTHESIS is currently a concept. I am actively exploring where it can grow first, and I’ll share the journey here as it unfolds.
If you work in VR, generative design, evolutionary computation, biometric interfaces — or if this idea simply made your pulse quicken a little (see what I did there?) — I’d love to hear from you.
Where attention goes, creation follows.

 Araniya Lumira, the founder of Delvira Art n Tech, a studio exploring art, attention, and technology.