Shared gallery links have a dirty secret: most guests never find their photos in them. Facial-recognition delivery flips the flow — instead of people searching for photos, the photos find the people. Here's exactly how it works, how to run it at your event, what it costs, and how to handle consent properly.
What is facial-recognition photo delivery?
At a normal event, photos end up in one big gallery and a link gets posted somewhere. Every guest then has the same job: scroll past hundreds or thousands of strangers' photos hoping to spot themselves. Most don't bother — and the photographer fields “can you send me that one of us?” messages for weeks.
Facial-recognition delivery replaces that with a simple loop: each guest registers once with a selfie. Software converts that selfie into a face vector — a set of numbers describing facial geometry, not a stored copy of the face. As the photographer's images come in, every face in every photo is compared against registered guests, and matches above a high confidence threshold are delivered to the right person automatically.
The payoff shows up on both sides: guests get their own photos without searching, and with a WhatsApp-based flow like SnapStream's, registration takes seconds at the venue — which is why opt-in rates on the night run high, and photos can land in chats while the event is still going.
How to run it at your event, step by step
The steps below use SnapStream as the worked example because it's free to try end-to-end; the overall shape applies to any face-delivery platform.
Step 1 — Create the event and pick a capacity
Sign up free, create an event, and choose a tier sized to your guest list. A 20-guest Taster event is free, so you can run your first event without spending anything. The event gets a unique join code and QR code.
Step 2 — Put the QR code where guests will see it
Print the QR on welcome signage, table cards, badges, or the bar menu; for festivals, screens and entry gates work best. The QR opens a WhatsApp chat — guests don't install anything, which is the single biggest driver of registration rates. Add one line of plain-language signage: “Scan to get every photo you're in, sent to your WhatsApp.”
Step 3 — Guests register with one selfie
In the chat, each guest agrees to the terms (explicit opt-in — more on consent below), then sends a selfie. Good selfies make good matches: face the camera, decent light, no sunglasses. Registration is done in under a minute, and the system confirms it in chat.
Step 4 — Shoot and upload as you go
Photographers upload through the web uploader during or after the event — no special camera hardware needed. Each upload is scanned automatically: every face detected, every registered guest checked, matches recorded at a high confidence threshold so lookalikes don't get someone else's photos.
Step 5 — Delivery happens by itself
Matched guests get a WhatsApp message — “You've been spotted! 📸” — with their photos and a link to a personal, event-branded gallery containing only the images they appear in. They can download, share, and revisit until the gallery expires. Nobody on your team sends anything manually.
Step 6 — After the event
Galleries stay live for 30 days, giving guests plenty of time to download. Guest data doesn't linger: registrations and face vectors are deleted automatically after the event, and any guest can type delete in the chat to erase their data instantly at any point.
Consent and privacy: do this properly
Face data is biometric data, and regulators treat it that way — POPIA in South Africa, GDPR in Europe. The good news: a consent-first flow is also the highest-converting flow, because guests are explicitly told what they're getting. The checklist:
- Opt-in, never opt-out. No face matching before the guest explicitly agrees. SnapStream won't accept a selfie until terms are accepted in the chat.
- Plain-language signage at the venue saying photos are being taken and face matching is available for guests who register.
- Easy deletion. Guests must be able to remove themselves without emailing anyone — a single delete message should wipe selfie, vectors, and matches.
- Time-boxed retention. Data should be deleted automatically after the event, not warehoused.
- Vectors, not faces. Prefer platforms that store mathematical representations rather than building a recognizable face database.
What it costs
Pricing models in this niche vary wildly, so compare like-for-like against your event calendar (figures verified June 2026):
| Platform | Model | Entry cost | Best fit |
|---|---|---|---|
| SnapStream | Per event, no subscription | Free (20 guests) · $49 typical wedding · up to $299, then Enterprise for large events | Weddings, festivals, corporate, sports |
| SpotMyPhotos | Monthly subscription | $195/mo (8,000 photos/mo) | Agencies & corporate programs |
| Waldo Photos | Commission + credits | 10% of order value (+5% premium matching) | US camps, schools, youth sports |
| PhotoDay | Commission on sales | 10% + card fees | US school/sports picture day |
| Memzo | Per photo uploaded | ~$0.03/photo | Indian weddings & events |
| Premagic | Quote-based | Reported from ~$99/mo | Corporate conferences |
Two structural things to notice. Subscriptions make sense only above a steady monthly event volume — a $195/month plan is $2,340 a year whether you shoot or not. And commission models tie your cost to your sales, which suits print-package businesses but penalizes free-to-guest delivery. Per-event pricing keeps the cost attached to the event, where it can be passed to the client — many photographers sell “every guest gets their photos” as a premium add-on and make delivery a profit line instead of an expense.
And the gallery half costs nothing: SnapStream includes unlimited client galleries with 20 GB of storage free (more at cost — $7 per 100 GB/month), so the face-matching fee is the only real bill. If you're weighing this against a traditional gallery subscription, see our SnapStream vs Pixieset comparison and the full alternatives roundup.
Pre-event checklist
- Event created, tier matches expected guest count, QR generated.
- QR printed on signage at entry, bars, tables — anywhere guests pause.
- MC or host briefed to mention it once early in the program.
- Photographers know the upload flow and cadence (during-event uploads = during-event wow).
- Consent signage up; team knows guests can type delete anytime.
- Test run done: register yourself, upload a few photos, confirm delivery lands.
Common pitfalls
- Hiding the QR. Registration rate is a function of visibility. The events that hit very high opt-in put the QR at the entrance and had the MC mention it.
- Bad selfie conditions. A dim corner selfie station produces weak vectors. Put it somewhere lit.
- Uploading everything at 2am. It works, but you lose the live-event magic. Upload in batches through the night — guests sharing their photos mid-event is free marketing for the organizer.
- Treating consent as fine print. Make the opt-in proud and clear. Guests trust a system that asks properly, and regulators require it.