The Restaurant Photo Spec Sheet

Pull up the photo most operators upload as their hero: the dim phone shot of the dining room, taken at 7 PM under tungsten cans, portrait orientation because that’s how a phone sits in the hand. Now read it against the surface it landed on. The Google panel wanted 16:9 and cropped a grey-barred sliver out of the middle; the menu page wanted one consistent shape and got a portrait that broke the grid; the file was 1200 px on the long edge, so the hero looks soft and the thumbnail looks fine. Nothing is wrong with the room. Everything is wrong with the fit. This is a teardown of that photo — what the spec actually demands of it, row by row, and the one fix each row triggers.

Every restaurant ends up needing the same dozen image sizes, but almost no one shoots them at once. The result: a Google Business Profile cover that's the wrong aspect ratio and got cropped weirdly, a Yelp photo that's portrait when Yelp wants landscape, a menu page where every dish photo is a different shape and the layout breaks.

The spec sheet below is the document that closes that gap — the one I hand a photographer the morning of a shoot, or run myself off a phone. One shoot, the right specs decided in advance, and the result drops cleanly into every surface instead of getting cropped into one.

The 10 image sizes you actually need

That table is the artifact. Here is how to read it the way it gets used — not top to bottom as a reference, but as four lines the dim phone photo above failed, each one naming the single fix it triggers.

1. Aspect ratio — the line that crops you. The cover wants 16:9, the gallery 4:3, the menu thumbnail one shape held consistent, Instagram portrait at 4:5. The phone photo was portrait, so every landscape surface clipped it to a center sliver and every square one padded grey bars top and bottom. The fix this row triggers: tap the phone’s aspect picker before the shot, or frame the subject inside the centre 60% so the wings can be cropped without losing it.
2. Resolution — the line you can’t walk back. The largest surface on the list is the 2400 px site hero; the smallest floor is the 720 px GBP profile. The phone file was 1200 px, so it was already too small for the hero the day it was shot. The fix this row triggers: shoot at the top-of-list width plus 30%, because you can always crop down and you can never crop up.
3. File weight — the line nobody reads until the page is slow. The same hero ships to a phone over cell data and to a desktop on fibre, and the JPG straight off the camera is heavier than either needs. The fix this row triggers: export site images as WebP at 80% quality before they ever touch the page.
4. Lighting — the line the camera can’t fix in post. The phone photo was shot at 7 PM under tungsten, which reads orange and dingy on a phone sensor no matter how good the phone is. The fix this row triggers: shoot 11 AM to 2 PM near a window; daylight is the one variable a better camera can’t buy back.

Resolution rules

Always shoot at the resolution recommended above PLUS roughly 30% extra. You can always crop down; you can never crop up. A hero shot at 2400×1024 lets you also produce the 1200×630 OG card from the same file; a hero shot at 1200×630 leaves you stuck.

For images that ship to your site, export as CDN-friendly WebP at 80% quality — it's typically 60-70% smaller than JPG without visible loss.

Alt text — the most-skipped part of the spec sheet

Every image needs alt text. Not "IMG_4521.jpg," not the menu name, not "delicious food." A specific descriptive sentence: "Plate of carnitas with cilantro, white onion, and lime on a wood board." Costs 5 seconds per photo and earns you SEO + accessibility + AI-search citation.

For a 30-photo restaurant shoot, that's 150 seconds — or two minutes — of writing. Skip it and you've left every photo's SEO value on the floor.

So this is the artifact to put under the glass. Two versions of the same carnitas photo’s alt attribute — what a parser throws away on the left, what it keeps on the right:

Alt text is the part of the spec sheet nobody reads. It's also the part Google reads first.

The shot list to hand a photographer

Run Photo Brief. It generates a per-surface shot list with the spec sheet attached — exactly the document you'd hand a photographer the morning of the shoot. Save the brief to your Workshop so you can re-run it for the next seasonal shoot.

The shoot itself is usually 2-3 hours for a 30-photo deliverable: the hero, 10-15 dishes, 4-6 room shots, 2-3 detail shots (a hand pouring wine, smoke off a grill, the chef plating). Done in one go, you have material for every surface for the next 12 months.

The 6-second pre-publish gut check

Before any photo lands on the site, run it through this check. Six questions, one click each, takes about as long as it sounds. If you can't answer "yes" to all six, the photo isn't ready — not because the photo is bad, but because something around the photo is.

The "shoot it yourself with a phone" version

If hiring a photographer isn't in budget, the phone version still works. Modern phones in good light routinely beat low-effort DSLR work — the variable that matters most isn't the sensor, it's the time of day and the angle to the window. Three rules:

1. Daylight only. Shoot 11 AM to 2 PM near a window. Restaurant overhead lighting reads orange and dingy on the phone sensor.
2. One angle per dish, repeated. Pick top-down OR 3/4 front, and use the same angle for every dish so the menu page reads as one set.
3. Edit lightly. Increase exposure +0.3, boost shadows. Skip filters.

How often to re-shoot

Twice a year. Once before peak season (Memorial Day for summer, mid-October for the holidays); once when the menu changes substantially. If you re-shoot less than once a year, the photos start looking dated to regulars even if they're still technically accurate.

Save the dates in your Workshop alongside your Margin Math and Restaurant Audit scenarios. The compounding argument: photo cadence is one of the few maintenance loops that visibly affects average review rating, because the GBP cover and the menu thumbnails set the expectation guests bring through the door — and reviews are written against that expectation, not against the food alone.

Bad photos don't just fail to help. They actively drag the average down by setting the wrong expectation before anyone walks in the door.