Skip to content

Selects: Home

shriansh edited this page Mar 1, 2026 · 1 revision

Selects — Overview

Because no one needs 15 shots of the same latte.


What Is Selects?

Your camera roll is a crime scene. The evidence: seventeen nearly-identical burst shots taken at your friend's birthday, forty-three blurry attempts at a sunset, and a growing collection of accidental pocket selfies. Selects is here to clean up the mess.

Named after the professional photography term for "the photos that actually made the cut," Selects is an iOS app that helps you rapidly triage your photo library — group by group, swipe by swipe — until your camera roll looks like it belongs to someone who has their life together.

No subscriptions. No cloud uploads. No server ever laying eyes on your blurry brunch photos. Everything runs entirely on-device.


How It Works

Selects groups your photos into clusters — shots taken within 60 seconds of each other (i.e., "the burst where you were trying to catch the dog mid-jump"). Instead of reviewing 500 individual photos, you review moments. The deck shows you one cluster at a time, stacked like a contact sheet, and you make a call:

Gesture What Happens
Swipe Up 🗑️ Cluster goes to the Trash Holding Pen
Swipe Down ✅ Cluster goes to the Keep Bin
Tap 🔍 Open the cluster to pick individual shots
Pinch 📅 Jump to the Archive Heatmap

It's like Tinder, but for the seventeen identical photos you took of your food.


The Secret Weapon: On-Device AI

Before you even see a cluster, Selects has already done its homework. A silent AI analysis pipeline (powered by Apple's Vision framework, running entirely on your device) evaluates every photo for:

  • Blur — because the moment was fleeting but the motion blur is forever
  • Blink detection — someone's eyes were closed. Not you. The photo.
  • Aesthetic score — an attention-based saliency model proxies for "this one looks better than the others"
  • Document & receipt tagging — for the running archive of parking tickets you've been meaning to delete

The result: by the time you swipe, the best photo in each cluster is already on top. The hero photo earns a yellow border. It worked hard for it.


Key Features

Nothing Is Deleted Until You Say So

Swiping to trash doesn't delete anything. Trashed photos land in a Holding Pen — a 2-column grid where you can rescue individual photos before pulling the trigger on permanent deletion. The safety net is intentional. You're allowed to change your mind.

The Keep Bin

Photos you swipe to keep land in a staging area, not your library. Review them, change your mind, send them back to the deck. The workflow is non-destructive in both directions.

Archive Heatmap

A calendar-style photo density heatmap visualising every year of your library as a grid of days, shaded by how many photos you took. That one suspiciously dark month in 2019? Tap it. The entire deck jumps to that date. Turns out time travel is mostly useful for confronting old photo bloat.

Real-Time Space Tracker

A live HUD at the top of the screen tells you exactly how much space is waiting to be freed. Watching the megabytes climb as you swipe is, frankly, more satisfying than it has any right to be.

Cluster Detail View

Not ready to commit? Tap any cluster to open every shot in a scrollable grid. All photos start selected. Deselect the ones you want gone. The action button adapts in real time:

  • Keep All — you liked every shot (overachiever)
  • Keep N & Bin N — a measured, responsible choice
  • Delete All — a bold move, and the carpet does look good

Design Philosophy

Selects is unapologetically dark mode. The typography is American Typewriter Bold — a deliberate nod to the contact-sheet aesthetic of analogue photography. The colour language is simple: indigo for navigation, red for trash, green for keeps, yellow for heroes.

The app is built for photographers — amateur, hobbyist, or anyone who shoots in bursts and has been quietly afraid of their camera roll for the past three years.


Technical Bits (For the Curious)

  • SwiftUI throughout, targeting iOS 17+
  • Photos framework for library access and edits
  • Vision framework for all AI analysis (face landmarks, image classification, saliency)
  • Zero network requests — your photos never leave your device
  • SelectsEngine — a central state machine managing clusters, holding pens, analysis results, and heatmap cache

Get In Touch

Found a bug? Have a feature request? Want to share that Selects just helped you delete 2GB of near-identical sunset photos and you feel like a new person?

📬 selects-support@shriansh.com


Selects — made with care (and mild horror at the state of the average camera roll) by Shriansh Shrivastava.