You didn’t become a photographer to spend your evenings deciding whether frame 412 is better than frame 413. But here you are. Sorting out 847 images clicking at each back and forth. Coffee going cold. The editing hasn’t started yet, and you already know it’ll be midnight before it’s done.
Why culling takes so long (and it’s not your speed)
The bottleneck in manual culling isn’t how fast you can click. It’s the volume of mandatory decisions — most of which have objectively correct answers that don’t need a human eye.
Closed eyes: reject. Out of focus: reject. Mid-blink, mid-word, cropped face: reject. These aren’t editorial decisions. They’re technical filters. And they’re what fills the first hour of every cull.
AI culling doesn’t replace your eye. It clears the queue so your eye only has to work on what’s left.
What AI culling actually detects
Arams analyzes your images by sending a small JPEG preview — not your RAW file — to the cloud. The full-resolution file never leaves your machine. Previews are deleted within 24 hours.
The analysis tags each image with everything a technical first-pass needs to know:
- Closed eyes and bad eyes (the mid-blink state — not fully closed, not open)
- Bad mouth — lips caught between expressions, mid-word, or awkwardly open
- Out of focus — when autofocus hunted and missed
- Overexposed / underexposed
- Cropped face — subject partially outside the frame
- Looking down / up gaze — useful for editorial portrait work
Every face in the frame is analyzed individually. A shot where the bride looks great but the groom blinked gets flagged on the groom’s face — you see exactly what triggered it.
What it doesn’t flag: the moment where everyone’s laughing and eyes are half-closed. The blurry shot that captures something irreplaceable. The technically imperfect frame that’s the most emotionally alive image in the whole gallery. That’s yours. That’s what you’re paid for.
Step-by-step: culling a wedding gallery with AI
1. Import and set up your project
Open Arams, create a new project. A naming convention worth using: YYYY-MM-DD_ClientName. Drag your image folder in — RAW, JPEG, TIFF all supported. For RAW files, Arams uses the embedded JPEG preview from your camera (Settings → General → “Use embedded JPEG preview”), which keeps the color rendering you’re used to from your camera’s in-body processing. Toggle it off if you prefer Arams to generate its own preview.
2. Run analysis — then go do something else
Select all images and click Analyze. Arams sends small previews to the cloud and tags every image. For 300 images, this takes about 3 minutes. For 800, plan 8–10 minutes. Start it and walk away. You don’t watch analysis. You just collect the results.
3. Choose your culling approach
Smart Cull — if you know your delivery count: Set a target number — say, 400 from 800 frames. Smart Cull selects the highest-rated images automatically. From 286 images in a demo, it selected 50 in under five seconds. Review the shortlist, override any pick with keyboard shortcuts 1–5, and you’re done with your first pass.
Feature filters — if you want to go image-by-image on the remainder: Open the face feature filters in the left sidebar. Stack the filters you care about: no closed eyes, no bad eyes, no bad mouth, no out of focus, no bad light. These five filters alone removed 457 images from a 2,841-image event gallery without a single manual review. What’s left is technically sound. Your editorial pass is now on a fraction of the original gallery.
Both approaches work. Smart Cull is faster when you have a fixed delivery count. Feature filters give you more control when you’re shooting with complex editorial criteria.
4. Find every shot of a specific person
Face identity filtering is completely free — no credits required. Enable automatic face search in Settings, and Arams indexes every face in your gallery. Click a face thumbnail to filter by that person. Adjust the similarity threshold (30% works for most wedding scenarios). To exclude someone who kept wandering into frame, hold Option and click their face — they’re excluded from the filtered view.
For wedding delivery, this means finding every image of the couple together, every shot of each family group, every usable candid of a specific guest — without scrolling through the whole gallery.
5. Do your editorial pass in Caps Lock mode
After AI filtering, your shortlist is the people who survived the technical cut. Now you make the editorial decisions. Enable Caps Lock in Arams: when you press P (pick) or X (reject), focus auto-advances to the next image. It’s faster than it sounds. Press the key, the image moves. Your eyes never leave the content.
Override any AI rating with 1–5 on the keyboard. When you manually set a rating, the indicator turns orange — that’s your override, not the AI’s score.
The honest time comparison
| Approach | Gallery size | Outcome |
| Manual review — no tools | 300 images | 45–90 min of clicking |
| Smart Cull | 286 images | 50 selected in under 5 seconds |
| Feature filters | 2,841 images | 457 removed, no manual review |
| Full AI cull + editorial pass | 800-image wedding | 15–20 min total active time |
Ready to cull your next gallery differently?
Download Arams free to help you speed up culling.
When you’re ready to unlock AI analysis and Smart Cull, the Starter Pack is $20: 100 professional cloud retouches and 4,500+ Smart Cull credits. That’s a full wedding gallery covered — enough to see whether the time math changes for your shoots before you commit to anything.
Want to learn more?
Watch our YouTube series with step-by-step guidance on organising your projects, culling and batch retouching from our team. Or if you prefer long-form content, you can watch the webinar replay on how to process thousands of photos.