nexa
ExploreServicesAboutTeamFAQBlogContact
GitHubStart a project
Blog · Design

Designing a camera screen you can use with one hand

Ingredient Scanner is used in supermarket aisles, usually while holding a basket. That changed where every button lives.

Tina

Team Lead · Mobile & Web

July 14, 20262 min read

On this page

  1. The study
  2. What we saw
  3. Everything moves down
  4. Results without leaving the camera
  5. Fewer words on the result
  6. Design principles we kept

Our first version of Ingredient Scanner looked like most camera apps: close top left, flash top right, shutter at the bottom. It tested fine at a desk. In a supermarket, with a basket in the other hand, people struggled to reach anything above the middle of the screen. This is a field report on what we saw and what we changed.

The study

Where
Ordinary supermarket aisles, at normal shopping times. No lab, no desk.
Who
Us first, then friends who had never seen the app, given no instructions.
Task
Walk an aisle, pick up a product, scan the label, decide whether to keep it.
What we watched
Grip, thumb movement, and every moment someone paused or shifted the phone.

What we saw

01

Phones held low

Close to the body, one hand on a basket or trolley the whole time.

02

Top corners unreachable

People stretched for close and flash, gave up, and changed grip.

03

Nobody complained

They just slowed down. Friction showed up as hesitation, not feedback.

04

Results screen broke the rhythm

A new screen meant reaching up for back before every next product.

Test where the app will be used, not where it was built.

Everything moves down

We moved every control into the bottom third of the screen, inside the area a thumb reaches comfortably. Close became a swipe down that works anywhere on screen. Flash moved next to the shutter.

Close Flash Shutter
Swipe ↓ to close Flash Manual Shutter
Same screen, controls moved into the shaded area a thumb reaches while the other hand holds a basket.

Results without leaving the camera

A successful scan used to open a new results screen with a back button in the corner. Now the result slides up over the camera view. Swipe it down and the next scan is ready.

  1. Scan labelShutter at the bottom
  2. Result slides upOver the camera, no new screen
  3. Swipe downDismiss from anywhere
  4. Next productCamera is already open
The scan loop. The thumb stays in the bottom third the whole time.

Fewer words on the result

The first result card listed every ingredient on the label. It was complete, and almost nobody read it in the shop. Now the card leads with a short verdict, shows the two or three ingredients that matter for the user's own preferences, and keeps the full list one tap away for later.

Design principles we kept

  1. Reach before looks

    If a control is out of thumb reach, it does not matter how it looks.

  2. Stay in the loop

    Every action should leave the user ready for the next scan.

  3. Admit bad photos

    When a label is blurry, ask for another shot instead of returning a confident answer built on a bad image.

Mentioned in this post

Ingredient Scanner

Case study

Work with us

Need an app like Ingredient Scanner built?

Our services

Keep reading

All posts
  • AI can design an app now. Are designers redundant?
    Read post

    Design

    AI can design an app now. Are designers redundant?

    AI makes the first draft of a screen cheap. That shifts design work to what matters most: understanding users and choosing what is right for them.

    Tina· Oct 10, 20266 min read
  • AI in software development: turn change into an advantage
    Read post

    Engineering

    AI in software development: turn change into an advantage

    AI does not replace technical thinking. With the right process, it frees teams to focus on the decisions that make products better.

    Ethan· Oct 9, 20264 min read
  • Adding AI to an app without a surprise bill
    Read post

    Engineering

    Adding AI to an app without a surprise bill

    Most of the cost of an AI feature is decided before the first request. Model routing, caching, context size and hard limits, in order of impact.

    Ethan· Oct 7, 20267 min read
nexa
ExploreServicesAboutTeamFAQBlogContact

© 2026 Nexa Tech. All rights reserved.

Web · Mobile · Open source