Photo & video

What photo and video apps are worth building?

Media tools earn their place by handling volume. One photo is easy to crop by hand. Four hundred from a wedding, or sixty clips from a school play, are not. The valuable ideas take a large pile of files and turn it into an ordered result: a sorted set, a cut film, a consistent export, or a shortlist of frames chosen by a criterion the user can read. If the tool would take a minute on one file and an afternoon on a folder, that gap is the reason to build it.

The recurring shapes are sorting tools, which group and remove duplicates; assembly tools, which cut clips into a sequence with titles and music; preparation tools, which resize, caption and export for a destination; and recovery tools, which rebuild an album or repair missing dates. Assembly and preparation are the easiest first versions to test, because the output can be checked at a glance. Sorting is harder, because a grouping that is almost right still needs a person to correct it.

A shortlist tool that picks the “best” frames is tempting and difficult. It is simple to score sharpness or exposure. It is far harder to score composition, expression or meaning, and users notice when the score and their judgement disagree. Start with the measurable criteria and label them as such.

Who uses photo and video apps, and how do you reach them?

Event photographers and videographers are the clearest professional group. They deliver many files on a deadline and are paid for the result, so a tool that saves real hours on a batch they already process is worth testing with them. Families and friend groups who share an event form a larger and less organised group. They want one product from many contributors, and they rarely have a shared process. Small shops and creators need consistent images for listings and posts, and they often work alone with a tight budget of time.

Reach the professionals where they already share work: the forums and groups for their trade, and the places where they post finished galleries. Show a before-and-after on a real event, with the couple’s or client’s permission, rather than a generic demo. For families, the shared moment is itself the reach. The question to test is whether the people who receive a finished film are also the people who would make the next one, and that is a hypothesis to measure, not a fact to assume.

What makes a photo or video idea good?

• It starts from a folder the user already has. A tool that needs a new capture workflow before it does anything loses most of its potential users.
• The result is judged in a minute. A thumbnail grid, a 60-second preview or a sorted list the user can scan is enough to decide whether the tool worked.
• It leaves the originals alone. Write outputs to a new location and keep a mapping from each output back to its source. A tool that overwrites or moves source files can destroy the only copy of a memory.
• It handles real volume. Test with four hundred files or the largest event you have, not with six photos. Export time and memory use are the failures that appear only at scale.
• It does not identify people. Anything that recognises faces or auto-tags individuals in a crowd raises questions a first version should not take on.

State the input folder, the output and the person who receives it in one sentence. If the output has no receiver, the idea is a tool looking for a use.

What does a first version look like? A worked example

Hypothetical wedding tool: an assembler for guest clips. Guests upload short clips to a shared folder before the event ends. The organiser runs the tool, which sorts the clips by the time they were recorded, trims each to a maximum length, adds a title card and produces one film in the order the clips were made. It does no face recognition and no automatic choice of the best moments, so every decision about what goes in stays with the organiser.

Build a command-line tool that runs on the organiser’s computer. Input: a folder of video files, each with a recording date. Steps: sort by recording time; trim each clip to at most 12 seconds from its start, skipping clips shorter than 2 seconds; add a title card from a plain text file; concatenate the clips in order into one file at 1080p. Never modify, move or delete anything in the input folder. Write a CSV listing each input file, its trim range and whether it was included, with a reason for every exclusion. Print the total output duration at the end. Do not detect faces, rank clips or change audio levels.

Test it on the folder from a real event. The first version is working when the CSV matches what the organiser expects and the film plays without manual repair. Then try the hardest cases: a clip recorded sideways, a clip with no date, and a folder of three hundred files. Each failure should appear as a line in the CSV with a reason, not as a crash that loses the batch.

What goes wrong with photo and video apps?

• Overwriting originals. A tool that edits or moves source files can destroy the only copy of an event. Write to a new folder and never change the inputs.
• Face and identity features. Recognising people in event footage can be a question of biometric data in many places, and the guests did not agree to it. Leave it out of a first version.
• Music in a finished film. Background tracks usually need a licence for public showing, and the rules differ by country. Offer a choice of licensed or royalty-free tracks, and make the terms visible to the person who publishes the film.
• Quality and compression. Re-encoding many clips can lose quality and take hours. Measure the export time on the largest event you have before promising a turnaround.
• Clients who expect a finished edit. Photographers and videographers have a craft of their own. Position a tool as a first draft that they finish, or it will be judged against their own work and found wanting.

Where to start: live ideas and the next post

One idea on the board matches the worked example closely: Every guest’s wedding video, cut into one film. The main open question is how the clips arrive, since guests will not upload them in the order things happened. A first version can sort by recording time and report the clips it could not place, which is more useful than guessing.

A photographer’s culling helper or a family album merger is a different tool: post it as an input folder, an output and the person who receives it. For choosing between several such ideas, how to scope an MVP shows how to cut the first version to one step that can be tested in a day.