Client Gallery With Face Recognition Included

Face search is off by default. The photographer switches it on for a specific gallery, one gallery at a time. Nothing happens on galleries where it stays off.

When it is on, the gallery shows a button. A guest taps it and the panel that opens carries a short explanation of what is about to happen, before any selfie is given. The guest then allows the camera and takes the selfie. The gallery filters down to the photos that matched, and the guest views or downloads from there like any other visitor.

One structural detail worth knowing: face search runs on the parent gallery link. A gallery can hold sections, and a section link is view-only, with no ZIP, no full-resolution handoff, no picks, no comments and no face search. If you want guests to search by face, send the gallery link, not a section link.

The selfie is processed on the guest's own device. It is never uploaded, never stored, and never seen by anyone. The comparison also runs in the guest's browser, so the matching itself is not a server round trip with a photo of a face in it.

What does get stored is a set of numeric face descriptors attached to the gallery's photos, not to the guest. They are computed in the photographer's browser when the photos go up. A descriptor is a list of numbers that describes a face region in an image. It belongs to the photo, and it dies with the photo: delete the photo or the gallery and the descriptors go with it.

On the server those descriptors sit behind the gallery password when a password is set. The gallery password is per-gallery and opt-in, and server-side it gates the gallery page, the whole-gallery ZIP, the full-resolution ZIP, comments, and face search. If you are running face search on anything sensitive, set the password. The Lightroom Classic plugin can set it from the Export dialog, so the gallery can go live already password-set.

Know what the password does not cover, so you do not promise more than it gives. Individual proof image URLs are public and do not expire. The password stops the gallery page from disclosing those URLs, but it does not protect an image URL that somebody already holds.

A longer treatment of the storage and deletion questions lives on the face recognition privacy answer page.

This is the commercial argument, so here are the figures as the vendors' own pricing pages state them today. Every number in this table is sourced at the bottom of the page.

Pic-Time is the honest benchmark here, because they ship the feature and document it. Their lower plans do not carry it. Selfie search is an Advanced item, which is the top of their list.

Pixieset is deliberately absent from that table. Every Pixieset URL returned an error on the date this page was written, so no Pixieset price and no Pixieset feature claim could be verified. Rather than repeat a figure from memory, we left them out.

The point of a single plan is that there is no tier to climb. Face search is not a lever to move you up a price list, because there is nowhere above it to move.

On the same plan: unlimited galleries, photos and clients under fair use, your own subdomain at yourstudio.clientgallery.io/project, client-facing pages carrying only your logo, colours and fonts with no platform badge and no link back to us, a per-gallery password, sub-galleries, full-resolution ZIP delivery, and the two-way Lightroom Classic plugin.

The fair-use ceilings, stated plainly because vague limits are worse than named ones: 1,000 photos per gallery, with the refusal telling you to split into another gallery and galleries being unlimited; 250 GB of proofs per paid account; 5 GB and one gallery on a free account. Full-resolution delivery is a paid feature, with 100 GB of standing ZIPs per account, and each handoff ZIP stays live for 3 days and can be re-uploaded at any time. Hosted galleries themselves never expire.

What keeps that price flat is scope. clientgallery.io hosts proof JPEGs. It does not warehouse RAW files or master video. If your requirement is a master archive, this is not the tool.

Face matching is a similarity score, not an identity check. It fails in ordinary conditions, and a guest who expects perfection will be annoyed rather than delighted.

• Shade and hard backlight. A face lit from behind at sunset often will not match the same face in flat indoor light.
• Angle. Strong profiles and faces turned away from the lens match poorly against a straight-on selfie.
• Motion blur. Dance floors, sports, children moving. Blurred faces produce weak descriptors.
• Sunglasses, hats, masks. Anything that covers the eye region cuts the match rate hard.
• False matches happen. A guest can be shown a photo of someone else who resembles them. Treat face search as a filter that saves scrolling, not as a guarantee of who is in a frame.

The practical consequence: keep the full gallery browsable alongside the face filter, and tell guests in your delivery note that the search finds most of their photos, not all of them.

This is the section a careful reader should weigh, and we would rather write it than have you find it out later.

Pic-Time's privacy policy documents a fuller consent flow than we run. It states that the feature is "by default disabled, if enabled it is at Photographers explicit opt-in consent", that they obtain "a written or electronic release", and that on withdrawal "Face Grouping will be disabled and associated biometric data will be deleted".

What we operate is narrower. Face search is off by default and the photographer opts a gallery in. The guest reads an explanation on screen before giving a selfie. Descriptors are attached to photos rather than to people, they are gated by the gallery password on the server, and they are deleted when the photo or the gallery is deleted.

What we do not have: there is no separate guest-facing control to delete face data, and there is no written release process. Switching face search off for a gallery stops the feature, but that switch is not itself a purge of descriptors that were already computed.

If you work under a regime that requires a documented biometric consent and deletion process, or your client contract demands one, that gap is real and it should decide your choice. Pic-Time documents the process. We do not, yet.

Face search alone is a bad reason to move a studio. Move for the whole workflow or do not move.

If your revenue runs through a print store, if you sell wall art and albums out of the gallery, or if album design lives in the same tool as delivery, stay put. clientgallery.io has no print store, no commissions and no album designer, and bolting a second tool onto a working sales flow to save a few dollars a month is a bad trade.

The reader this page is written for is different: a photographer who delivers photos, wants guests to find themselves without help, and does not want that capability priced as a top-tier upsell.

1. Create the gallery and upload. Descriptors are computed in your browser as the photos go up.
2. Set the gallery password if the shoot warrants it. It gates face search server-side along with the gallery page, the ZIPs and comments.
3. Switch face search on for that gallery. It stays off everywhere else.
4. Send the gallery link, not a section link. Sections are view-only and carry no face search.
5. Tell guests what to expect: a selfie processed on their own phone, a filter that finds most of their photos, and the full gallery still there to browse.

Two vertical walkthroughs go deeper than this page: guests finding their own photos with a selfie and event photos where attendees find themselves.

• Pic-Time, Pricing. https://www.pic-time.com/pricing (fetched 8 September 2026). Plan prices and the Advanced-only Smart Search listing.
• Pic-Time, Privacy Policy. https://www.pic-time.com/privacypolicy (fetched 8 September 2026). Opt-in consent, written or electronic release, and deletion on withdrawal.
• CloudSpot, Pricing. https://www.cloudspot.io/pricing (fetched 8 September 2026). Plan prices and storage figures; no face search advertised on the page.
• Pixieset pages could not be retrieved on 8 September 2026, so no Pixieset figure or feature claim appears on this page.