Phone snap,
studio piece.

Get near a window, turn off the overhead lights, and shoot from about 45 degrees with the dish filling most of the frame. Wipe the plate rim. Shoot within a minute or two of plating, while everything still looks alive, and take five frames instead of one. That's 80% of it. The last 20% — clean studio light, true colour — is the part software can add afterwards.

Two angles cover nearly everything. A 45-degree tilt — camera looking down at the front of the dish — suits most plates, because it shows height and depth the way a seated diner sees them. Straight top-down works for flat food: pizza, bowls, spreads. Straight-on at table level is for tall builds like burgers and layered desserts. Pick one angle per menu section and stick to it; the consistency looks more professional than any single hero shot.

Restaurant bulbs. Most dining rooms run warm tungsten light around 2700K, and phone cameras only partly correct for it — so rice turns yellow, white plates go cream, greens drift toward olive. Shoot near a window in daylight when you can. When you can't, the cast is fixable afterwards; colour correction is one of the main things PlateRefresh does on every photo.

No. A phone flash is a small, hard light firing straight at the plate — it flattens texture, throws a harsh shadow behind the dish, and puts hot glare on anything wet or glazed. Soft window light beats it every time. If the room is genuinely too dark, carry the plate somewhere brighter; nobody will ever know the pasta was photographed by the front window.

Your eyes handle a far wider range of light than a phone sensor. Standing at the pass, you see the sheen on the sauce and the detail inside a dark bowl at the same time; the camera has to choose one, so highlights burn out or shadows turn to mud, and colour drifts with the room lighting. The dish is fine. The exposure isn't — which is why fixing the light fixes the photo.

Light, mostly. A pro shoot uses one big soft light source, a touch of glow from behind the dish so it reads fresh, and an exposure that keeps sheen on the sauces without burning them white. Add a plate straight from the kitchen — not one that sat twenty minutes — and honest styling like a brush of oil or a crack of pepper. That's the whole recipe. Good food photography involves less trickery than people assume.

A phone. That's the list. No ring light, no lightbox, no reflector boards, no editing apps — the missing ingredient in phone photos is studio lighting, and that gets added after the fact. As for props, you already own the best ones: the plates and tables your customers actually eat off make the menu look like your restaurant instead of a stock library. What matters more than any prop is consistency — same surfaces, same light, same angle across the whole menu — and brand kits exist to lock exactly that.

Yes — what platforms prohibit is photos that misrepresent the dish, and PlateRefresh is built specifically to avoid that. A real photo of your real burger with studio lighting applied is the same thing every restaurant with a professional photographer has done for decades, just faster. We never add ingredients, change portions, or swap the dish.

Details vary by platform, but they all want the same things: a sharp, well-lit photo of the actual item, landscape or square, at least around 1,200 pixels wide, with the food filling most of the frame on a clean background. All of them reject watermarks, logos, collages, and photos with people in them. Exact pixel specs shift often enough that the merchant guides are worth checking — but a clean, high-resolution shot passes everywhere.

The usual offenders, in order: a watermark or logo on the image, blur, a photo too dark to read, a collage or border, the item not filling the frame, or the photo showing something other than what's sold. Platforms also bounce obvious stock photos. A sharp, straightforward shot of the real dish on a clean surface almost never gets refused.

Indirectly, but powerfully. The apps rank partly on conversion — how many people who see your listing go on to order — and tiles with appetising photos convert far better than grey placeholders. Complete, recently updated listings also score better than half-filled ones. A fully photographed menu lifts you twice: more taps per impression, and better placement because of those taps.

Yes, and most restaurants under-use it. Google has long reported that businesses with photos get significantly more direction requests and website clicks, and food photos are exactly what people scroll through when choosing where to eat. Upload your best plates, refresh a few every month, and make sure the first photos are dishes — not the sign, not the empty dining room.

If you own the rights, yes — and that's worth checking. Photographer contracts often license images for specific uses, so shots you paid to put on Wolt may not be cleared for a billboard. Everything made with PlateRefresh carries a full commercial license: menus, apps, ads, print, social, forever. For print, take the 4K download — 2K is plenty for screens, but big print wants the extra resolution.

No, because the food in the photo is your food. PlateRefresh doesn't generate a dish from a text prompt — it takes the photo you shot and rebuilds the lighting, color, and background around the exact plate you cooked. What changes is what a full studio day would change: light, contrast, surface, mood.

Changed: lighting, exposure, colour balance, contrast, the surface and background around the plate. Never touched: the food itself. Same ingredients, same garnish, same portion, same plating, same angle. The rule is simple — a customer holding the plate next to the photo should see the same dish — and it's enforced in the pipeline, not just promised on this page.

Midjourney invents a dish; we relight yours. Your photo goes in first, and the model is constrained to preserve the dish — lifting its textures, never reinventing it — while improving everything around it. Plus the practical stuff a chatbot won't do: locked brand plates across your menu, batch processing of 50 photos, delivery-platform crop ratios, and automatic refunds for failed shots.

Probably. The minimum we need: the dish in focus, reasonably lit, and fully in frame. From there the pipeline handles dull light, yellow cast, ugly backgrounds, harsh shadows and flat contrast. What it can't fix — and be suspicious of anything that claims otherwise — is a blurred photo, a half-eaten dish, or a frame so dark there's no detail left: that information is gone, and "recovering" it means inventing a dish that isn't yours. Retake those; everything else is fair game.

A brand kit locks your actual plates, tabletops and surfaces so every photo across your menu looks shot in the same restaurant. Upload photos of the ceramics and tables you really serve on, and every generated shot uses those exact props. Even Test Mode, the cheapest way in, includes 1 brand kit; most single-location restaurants need exactly one.

Yes — upload up to 50 photos in one batch and they run through the pipeline together, so a full menu takes a coffee break rather than an afternoon. Each dish is still handled individually, and you get an email when the batch is done. Most batches come back within about ten minutes.

When a generation fails outright, the credit comes back automatically. When the result just isn't right, run it again with different settings. The honest expectation: most dishes look right on the first or second try. Test Mode exists so you can confirm that on your own food before putting a whole menu through.

You do, permanently. Every paid plan — Test Mode included — comes with a full commercial license covering menus, delivery platforms, advertising, social and print. Cancel whenever you like; the photos you've already made stay yours to download and keep.

They're stored encrypted, used only to produce your finished shots, and never used to train AI models. PlateRefresh is an EU company and takes GDPR seriously: you can export everything or request full deletion at any time. Nothing you upload feeds anyone's training data.

Yes, under the same rule that has always governed food advertising: the photo must honestly represent what's served. Adjusting light and colour is standard practice — every professionally shot menu photo you've ever seen was edited. PlateRefresh also embeds the machine-readable AI disclosure the EU AI Act calls for into each file's metadata, so that box is ticked without a visible watermark.