← All work
Lux Optics
· 2026
Halide was losing a quarter of new users before their first photo
Halide’s first-run manual lost 25 to 30 percent of new users on the first page. I replaced the page flipping with a hands-on demo of the Looks, and that change led to a single paywall for every Lux app, built so the team can test what converts.
Role
Product designer. Onboarding and paywall design, prototyping, and code contributions to the shared paywall
Team
Lux Optics founders and engineering
Timeline
2026
Platform
iOS (Halide Mark III, Kino, shared paywall package)
Problem
Halide’s first-run manual lost 25 to 30 percent of new users before they took a photo. Each Lux app also kept its own paywall, with no easy way to test what converted.
Approach
Cut the page flipping in favour of a demo of the Looks, keep one page turn as a moment of delight, then fold every paywall into one repo built around swappable backdrops.
Outcome
An onboarding that shows people why they’d want Halide before asking them to pay, and a unified paywall Lux can iterate on and A/B test across Halide, Kino and future apps.
A manual most people never finished
Halide Mark II opened with a camera manual. You flipped through it page by page, like the booklet that comes in the box with a film camera. It was charming, and it was costing Lux: 25 to 30 percent of new users dropped off right at the start, before they’d taken a single photo.
For Mark III that loss mattered more than ever. Halide’s original audience, camera enthusiasts who wanted RAW and manual focus, had been saturated since 2020. Mark III needed to win over people who didn’t yet know why they’d want a third-party camera at all.
What people want from a camera app
Lux’s user research pointed away from feature lists. The people Mark III needed to reach, mostly Gen Z and then millennials, weren’t shopping for f-stops. They were looking for a look of their own, something they couldn’t get from the default Camera app. What they told us they wanted:
A unique look. Process Zero, Halide’s mode that skips Apple’s computational processing for a natural, film-like image, was the clearest reason anyone gave for switching.
A camera that’s more fun to use. Something more intentional than point and shoot, with manual controls there when they want them.
More than a camera. Being able to edit after capture was part of the appeal.
Proof that it’s good. Apple’s Editor’s Choice and the Apple Design Award carried more weight than we expected. Apple had told Lux’s founder it’s a huge selling point.
Cutting the page flipping
My recommendation was to drop the manual pages and spend those first seconds on the Looks instead. Rather than reading about Process Zero, you see it on a real photo, then tap through Looks and watch the image change. The research said people come for a look, and the Looks demo better than anything else on a phone screen.
It also lined up with the principle the founder kept coming back to: let them have something to control before asking for money.
Exploring the demo. Early screens split the same photo between Process Zero and iPhone processing, with Looks in a carousel underneath.
My first pass leaned too hard into realism. It had too many dials, which made it intimidating and risked misrepresenting the real camera. We landed on evoking the camera instead of replicating it: an abstracted UI with just the Looks, and pre-rendered photos bundled with the app so the demo works instantly on any device without depending on the live camera stack. The Look you pick carries into the app after the paywall, so your first real photo already looks the way you chose.
Pick Your Look. The shipped first run. Tap a Look, see it on the photo, and keep it in the camera.
Keeping one page turn
The manual itself wasn’t the problem. Making everyone read it was. People who love Halide love that kind of detail, and the page turn is part of what makes the app feel made by hand.
So I recommended keeping a single page flip in the first run as a moment of delight, not a chore, and keeping the full manual in Settings as a fun interaction for anyone who goes looking for it. New users get to the camera faster, and the people who want the manual still find it.
From one onboarding to one paywall
The onboarding change worked, and it pointed at the next step: the paywall. Halide and Kino each had their own, maintained separately. Every copy change or new feature meant touching each app, and there was no simple way to compare which message was converting better.
So we consolidated them into a single, unified paywall repo that every Lux app pulls in. I contributed to the GitHub project directly. The goal was to make the content easy to iterate on and A/B tests easy to run, so Lux can track what’s actually converting instead of guessing.
A paywall built for testing
I refactored the paywall’s Xcode project so each page’s backdrop can be a static image, a looping video or a carousel. The card on top stays the same: an icon, a headline, a couple of lines of copy and the trial button. Trying a new message means swapping a backdrop and some copy, not designing a new paywall.
Kino paywall drafts. One shared card over different backdrops: Looks, Cinematic Footage, Instant Grade and Easy to Use.
For Kino, the research pointed the same way as Halide. People want their own look, so the paywall leads with Looks, the colour grades built by award-winning colourists. Cinematic Footage and Instant Grade come next, with manual controls, external drives and the App of the Year accolades further along.
What I took away
The biggest improvements on this project came from taking things out: the page flipping, the extra dials, the duplicate paywalls. The research gave us permission to do it, because it was clear what people were actually looking for.
The other lesson was about what to build after a win. Once onboarding was working, the most useful thing wasn’t a prettier paywall. It was a paywall that’s easy to change, so the next improvement can be a test instead of a guess.