Balian.dev
St. Paul, MN · remote Book a call

Services

Science visualization that runs in the browser and gets the physics right.

Scrollytelling explainers and 3D with the real model underneath. For a museum or a lab or a publisher with something hard to explain. The demos are live and you can try them.

Who hires a science visualization studio

  • A museum or a planetariumAn exhibit that has to work on the floor, on the website and on a phone in the gallery, with the same model under all three. The visitor turns the knobs and the physics answers.
  • A lab, an observatory or a research instituteThe communications office has a result to explain to the public, a funder or the press, and the figure in the paper does not do it. A scroll through the finding, with the real data driving the picture.
  • A university online programA flagship course needs a lesson that is more than slides. The subject's model runs under the page, the learner moves through it at their own pace, and it lands in the LMS as SCORM, xAPI or over LTI.
  • A publisher, an edtech company or a magazineAn interactive for an article, a textbook chapter or a feature, built to your design and embedded in your CMS, with a still and a text version for every reader who cannot run it.
  • A biotech, aerospace, energy or medical device companyA mechanism, a device or a process that takes a whiteboard to explain to investors, surgeons or regulators. An explainer they can scroll and take apart, built from your CAD and your numbers.

Scrollytelling, when the subject moves

Scrollytelling is a page where scrolling drives the picture, so the reader sets the pace and the explanation unfolds in order. It suits any subject that happens over time or in steps: the universe from the first instant to today, a tide rising in a bay, a watch movement coming apart, an argument built one premise at a time. The demos on this site are built that way, and each one has a caption at a time, a clock or readout that keeps the comparison the stage is making, and a chip that says what kind of claim is on screen.

The difference from a video is that the model is live. A visitor who scrolls backwards watches the thing come apart. A number that changes on screen is computed from the parameters, so if the parameter changes, everything changes with it. The page is text first for a screen reader, the keyboard steps one caption at a time, and anyone who asks their device for less motion gets a still frame per stage.

3D explainers and medical device animation

A device or a mechanism from your CAD files, cleaned and compressed so it loads in seconds on a phone, then taken apart on scroll with every moving part moving at the real ratios and a label on each one. The watch demo is that pipeline applied to a hand-wound movement. For a device company the same build shows a mechanism of action, a delivery system or a procedure step by step, in the browser instead of a rendered video, so the sales engineer can stop on the frame the surgeon asked about.

I am in St. Paul, which puts me in the middle of one of the largest medical device clusters in the country, and most of what those companies explain to investors and clinicians is still a PDF and a rendered animation. An interactive explainer that is correct, that runs on a phone and that the regulatory reviewer can check is the version I build.

How the science stays right

  • No physical value is typed into a scene. Parameters go in, a tested module computes the rest, and the readout on screen calls the same function as the geometry, so the number and the picture cannot disagree
  • Every stage declares what kind of claim it is making: measured, established, simulated, an argument, open, or a metaphor, and names its exaggerations in a note that is in the page's HTML
  • Sources per stage, in an appendix the reader can open, with the papers the numbers come from
  • A subject-matter reviewer on your side or on contract, a graduate student or a postdoc in the field, reads every caption and every number before it ships
  • Learning objectives and checks where the piece is a lesson, so an instructional designer can place it in a course
  • WCAG 2.2 AA from the start: the story in the DOM for screen readers, stage changes announced, keyboard stepping, reduced motion honoured, and stills with the same text for anyone without a GPU

Work I have done

  • Cross Rhythm: polyrhythm and polymeter with live sound, three against two as rings on one axle, meshing gears for a shared beat, a ratio as a torus knot, a rhythm sped up until it becomes a chord, the Rhythmicon and Euclidean rhythms, ending in an instrument you can play. Every sound is synthesized in the browser from the numbers that draw the picture.
  • Tide Machine: how the tides work and how a brass machine predicted them, with San Francisco Bay from real terrain and bathymetry, water driven by NOAA's harmonic constants, Kelvin's 1872 predictor and the 1912 Old Brass Brains, and a finale that predicts the tide on any coast on Earth
  • Block Universe: the block universe theory built as an argument, two observers slicing one block of spacetime, every tilt computed from tested relativity and every stage labelled by what kind of claim it is. See the post.
  • Cosmic History: the history of the universe from a fraction of a second after the beginning to today, each stage computed from the Planck 2018 cosmology on the GPU and labelled with how accurate it is. See the post.
  • Rose Engine: a guilloché lathe in the browser, where the rosettes you stack drive a 3D machine and the plate it cuts, in ink or engraved metal, with SVG, PNG and G-code export. See the case study.
  • How a mechanical watch works: a hand-wound Swiss movement drawn from CAD, taken apart step by step on scroll, with every gear turning at the maker's published ratios. See the case study.

What you get

  • The piece itself, on your site, in your article or inside your exhibit, in your design system
  • The subject in plain files your expert can edit: captions, how we know, sources and status per stage, with the engine untouched underneath
  • A fallback that works without WebGL and a text version with the same numbers, so every reader gets the explanation
  • Performance measured on a mid-range phone, which is where most of it will be looked at
  • SCORM, xAPI or LTI packaging when it is a lesson, and an embed code when it is an article
  • Source files and models you own, with the physics module and its tests
Questions
Do you work with our scientists?

Yes, and I would rather. Your expert writes or reviews the captions, the numbers and the sources, and I build the scene under them. Where you do not have one, I bring in a reviewer in the field on contract and the proposal says who.

Can it go inside our article or our CMS?

Yes. A piece ships as an embed that drops into WordPress, a custom CMS or a publishing platform, with its own styles contained so it does not fight the page. The text version and a still come with it for feeds and for readers who cannot run it.

Does it work in an exhibit?

Yes. The same build runs on a kiosk in the gallery, with touch in place of scroll, an idle mode that attracts, and a reset between visitors. The website gets the same piece, so a visitor can pick it up at home.

Is this a video?

No. A video plays at one speed and a visitor cannot ask it anything. Here the model runs live in the browser, so the reader sets the pace, scrolls back to the frame they missed, and the numbers on screen are computed, which means a scientist can check them.

What about accessibility?

The story is in the page as text first, in order, for screen readers. Stage changes are announced, the keyboard steps one caption at a time, anyone who asks their device for less motion gets a still frame per stage, and with no graphics card the stills and the text still read. It is built that way from the start because retrofitting it costs more.

How is this priced?

A fixed price for a written scope, like everything else I do. The proposal says what the subject, the number of stages, the review and the packaging add up to before anything starts. One engineer building with AI is what makes that price fit a communications or exhibit budget.

Tell me what you're building.