Key Faces: Faster Filtering for Weddings and Events

A wedding gallery might run past a thousand photos by the time the night is over, and somewhere in there are the shots someone will actually ask you for by name: the two moms together, the whole bridal party, that one uncle who wanted a photo with every cousin. None of those requests map neatly onto a star rating or a single search term. They are about specific people showing up across a huge, unsorted pile of frames, and right now finding them still takes a human eye scanning thumbnail after thumbnail.

This is the exact gap a feature called key faces is being built to close, and it is worth being upfront about where that stands today: it is on the roadmap, not yet available, and there is no date attached to it. What follows is what already helps with people-heavy galleries right now, what the gap still looks like, and what key faces is aimed at once it ships.

None of this is a reason to hold off on using the tools already in front of you. A wedding or event gallery is easier to work through today than it was a season ago, even before a face-specific filter exists, and the habits that make it easier now are the same ones that will make any future filter more useful the moment it arrives.

Why weddings and events are the hardest galleries to search

Most shoots have a fairly small cast of subjects. A portrait session has one family. A product shoot has no faces at all. A wedding or a large event flips that completely: dozens or hundreds of people move through the frame, many of them appearing in only a handful of photos each, and the people who matter most to the client, the couple, the parents, a handful of close family, are scattered across the entire day rather than grouped together in one obvious block of time. Asking “find every photo of the groom’s parents” is a completely reasonable request from a client. It is also, today, one that still needs a person to answer by scanning the gallery directly.

Scale makes this worse, not better. A small elopement with forty guests is manageable to scan by eye. A four hundred guest wedding, or a multi-day conference with a rotating cast of speakers, is not, and that is precisely the kind of gallery where a faster way to find specific people would save the most time.

What already helps today

Several tools already cut down the manual scanning, even without a feature built specifically around recognizing faces. Rating photos as you cull means you can ask the assistant for every five-star frame and get a strong shortlist of the day’s best moments in seconds. Plain-language search lets you describe a moment rather than a person, every photo where the whole group is dancing, or the shots from the ceremony itself, and get an answer without a person-by-person search. Smart tags catch out-of-focus frames and closed eyes automatically, so a first pass through a massive gallery clears out the obvious technical misses before anyone starts the more subjective work of picking favorites.

Stacked together, rating, plain-language search, and smart tags remove most of the tedious scanning from a large event gallery. What they do not yet solve is the specific, recurring request to find one particular person or small group across hundreds of unrelated frames.

The gap those tools do not close yet

The honest limitation is this: the assistant currently knows what is happening in a photo, laughing, dancing, a group shot, but it does not yet recognize a specific individual across different frames the way a person would recognize a familiar face in photo after photo. Asking for “every photo with the couple” works reasonably well because there are exactly two people involved and the request describes a role rather than an identity. Asking for “every photo with a specific relative” is a different, harder problem, because it requires recognizing one particular face across lighting changes, angles, and a crowd, and that is not something today’s tools are built to do.

It is a subtle but important difference. A role, the couple, the officiant, the flower girl, is really a description of what someone is doing or wearing in that moment, which the assistant can work with. An identity is about recognizing the same face again and again regardless of what that person is doing, which is a fundamentally different kind of task and the one key faces is specifically aimed at solving.

Key faces: what is coming, and what is not confirmed yet

Key faces is the feature aimed directly at that gap. It is on the product roadmap as a future AI filter, described for now as a direction rather than a committed date, and it is worth treating it exactly that way until it actually ships. The idea behind it is straightforward: let you mark or identify the people who matter most for a given gallery, then filter the entire shoot down to just the frames those specific people appear in, cutting a thousand-photo wedding down to the handful of albums a family actually asks for by name.

Searching a gallery by what is in the photos in Imagen

Until it arrives, nothing about today’s toolset changes, and nothing here should be read as a promise about exactly what the finished feature will look like. What can be said with confidence is the problem it is meant to solve, because that problem is the one described above, and it is a real one for anyone shooting weddings and large events regularly.

Why this matters most for people-heavy shoots specifically

A feature like this earns its keep in direct proportion to how many people are in a gallery and how few of them appear in every frame. A wedding is close to the ideal case: a large, shifting cast, a handful of people who matter disproportionately to the final delivery, and requests from clients that are almost always phrased around a person rather than a moment. A conference or a large corporate event shares the same shape, a stage full of speakers and a room full of attendees, with specific requests usually landing on one speaker or one executive rather than the room as a whole. Contrast that with a single-subject portrait session, where there is no search problem to solve in the first place because there is only ever one or two people to find.

Even within a wedding, the need is uneven. The ceremony and the formal portraits already tend to be organized and easy to review. It is the reception, hours of loosely structured candids with the widest possible mix of guests, where a specific-person filter would save the most scanning time once it exists.

What to do today while you wait

Until key faces ships, the practical move is to lean harder on the tools already available and to build habits around them now, so the transition is seamless whenever the new filter does arrive. Rate generously during culling rather than sparingly, since a broader five-star set gives the assistant more to work with when a client asks for favorites. Use plain-language search for moments and roles, the ceremony, the toast, the couple, rather than trying to force it to recognize a specific individual it is not built to identify yet. And let smart tags clear out the technical misses early, so the gallery you are manually scanning for specific people is already smaller than the one you originally uploaded.

None of this is wasted effort once key faces does arrive. A gallery that is already rated, already searched by moment, and already cleared of technical misses is a smaller, cleaner set for any future person-specific filter to work against, rather than the full, unsorted upload.

For more on building a search-friendly workflow for large galleries, our valuable tips library covers rating habits and plain-language search in more depth.

What’s new in Imagen: People-heavy shoots are a good fit for the new Imagen, an intelligent, agentic Studio for professional photographers where you cull, edit, review and export in any order. Rate as you cull, using AI culling or manual culling for free, then ask the assistant in plain language for the moments and roles you need, with you steering. The Studio is rolling out gradually as a beta.

Frequently asked questions

Is key faces available now?

No. It is on the roadmap as a future AI filter, with no confirmed date. Today’s smart tags cover out-of-focus and closed-eye detection only.

Can the assistant already find a specific person by name today?

Not reliably. It can describe and find what is happening in a photo, but recognizing one specific individual consistently across a large, varied gallery is the gap key faces is being built to address.

What can I search for today on a large wedding or event gallery?

Ratings, moments, and roles work well now: five-star frames, a specific part of the day, or a description like “the group photos.” These cut a large gallery down significantly even without face-specific filtering.

Will key faces replace rating and plain-language search once it launches?

No. It is expected to work alongside them as another way to filter a gallery, not a replacement for rating your favorites or describing a moment in plain language.

Should I change how I shoot or cull now to prepare for key faces?

No changes are needed. Rating consistently and using plain-language search for moments and roles are useful on their own today, and they will still be useful once a face-specific filter is available.

Want to see how much easier your next gallery is to search? Start a project in Imagen’s new Studio, where you cull, edit and export together, and put today’s rating and search tools to work at imagen-ai.com.