Photo

Photo challenges collect visual proof: "take a photo of your whole team in front of the fountain." Instead of a typed answer, the team submits an image — and you decide how it gets judged: automatically, by AI, by a facilitator, or AI-assisted facilitator review.

What players experience

The challenge screen shows the Clue, difficulty badge, {points} pts chip, and a "Photo Evidence" section with two buttons: "Take Photo" and "Choose Photo". The app requests camera/photo-library permission (with an "Open Settings" fallback if denied), uploads the image before submitting ("Uploading…"), then submits.

What happens next depends on your event settings:

• Accepted instantly — the confetti "Challenge Complete!" modal with the animated point award.
• Held for review (when Review Before Points is on) — the modal instead says "Submitted for Review!" / "You'll earn" with the footnote "⏳ Points pending facilitator review". The challenge shows an amber box: "Your submission is with a facilitator. {points} points are on hold until it's approved." The Home screen's points tile shows +{n} pending review in amber. The form locks to prevent double submissions.
• Rejected by AI — the "Not Quite!" modal; the team can retake and resubmit. If a facilitator rejects a held submission, the team sees the rejection reason with a prompt to resubmit.

Hints, access codes, and activation gates work exactly as on other challenge types.

Scoring & answers

Photo challenges use the Photo answer type and are judged by media validation rather than string matching:

1. AI validation runs only when both are true: the challenge has AI Validation Criteria set, and the event's AI Photo Verdicts toggle is on ("AI pre-checks each photo/video (uses tokens)"). The photo is sent to OpenAI's gpt-4o vision model with your challenge name, description, clue, and criteria; the model is instructed to be "reasonably lenient" and returns a verdict with a confidence score and reasoning — all stored on the submission and shown to facilitators in the review drawer's AI verdict panel.
2. Review Before Points ("Photos/videos hold points until approved") holds every media submission for a facilitator, even when AI has already judged it — the AI verdict becomes advice for the reviewer, not the decision. Approving awards the challenge's points (or the reviewer's Points to award override); rejecting awards 0 and prompts the team to resubmit.
3. If the AI call fails, the submission is not lost — it is accepted into the queue as pending manual review with the note "AI validation failed, pending manual review".
4. If neither AI nor review is enabled, every photo submission is accepted automatically and scored as correct.

Standard scoring applies on acceptance: Hint Cost deduction (never below 0), Speed Bonus tiers for the first non-staff finishers, Max Completing Teams, and a live Double Down effect doubling the award. On facilitator approval, points are the reviewer's Points to award value (defaulting to the challenge's full points).

Configuring in the admin

Held submissions land in the event's Submissions tab and the Live tab's "Photo & video reviews" panel; the review drawer shows the photo, the AI verdict panel (confidence % and reasoning), a Points to award input, and Approve (A) / Reject (R) actions. See Running an Event.

Tips

• Write criteria like you're grading, not inspiring. "Team members visible in front of the fountain, water visible" validates far better than "a fun team moment" — the AI compares the image against your literal description.
• The AI is deliberately lenient. For competitive events where a borderline photo shouldn't auto-score, turn on Review Before Points and let the AI verdict pre-sort your queue instead of deciding it.
• With both toggles off, photos are participation trophies — every submission auto-accepts. That's a valid, zero-effort mode for casual hunts; just know that's what you're choosing.