Log a meal from a photo

What it does

You take or upload a photo of a meal. NutrientOS identifies the likely components, estimates portions, and matches each one to a food record. The result opens as an editable draft: every item, portion, and match can be changed, removed, or replaced before you log it.

The point is to remove the slowest part of tracking. Typing out a mixed plate — chicken, rice, a sauce, some vegetables — takes four searches and four portion decisions. A scan turns that into one review pass.

Who it is useful for

• •People who quit tracking apps because entering a home-cooked plate item by item is tedious enough to skip, and skipped days end the habit.
• •Anyone logging mixed or composite meals stir-fries, curries, bowls, and restaurant plates that do not map to a single database entry.
• •People who already know their foods and just want the entry pre-filled so they can adjust one number and move on.

What NutrientOS shows you after a scan

• •Each detected item with its estimated portion and the food record it matched to.
• •Calories and the protein, carbohydrate, and fat split for the meal and for each item.
• •Micronutrients carried by the matched records, with unmeasured ones marked unknown.
• •Food Grade and a satiety estimate for the meal, computed from the matched nutrition.
• •A correction path — "Fix result" — that lets you describe what was wrong in plain language.

Known limitations

Photo estimation is genuinely hard, and this feature is wrong often enough that you should treat it as a first draft rather than a measurement.

• •Portions are the weakest part. A photo has no scale reference. Depth, bowl size, and camera angle all distort volume, and dense foods hide inside a serving.
• •Hidden ingredients are invisible. Cooking oil, butter, sugar in a sauce, and dressing do not appear in a photo and can move a meal by hundreds of calories.
• •Similar-looking foods get confused. White rice and cauliflower rice, full-fat and low-fat dairy, and different cuts of the same meat can look identical.
• •It cannot read a specific brand from a plate. For packaged food, the barcode scanner is the accurate path — use it whenever a package is in front of you.

If accuracy matters for a particular meal — a weigh-in week, a competition cut, a clinical conversation — weigh the components and log them directly. A scan is for consistency on ordinary days.

How missing data is handled

When a nutrient is not present in the underlying record, NutrientOS shows it as unknown rather than as a zero. A blank is honest; a zero would read as "this food contains none of it", which is a different and usually false claim.

How it connects to the rest of NutrientOS

A scanned meal is an ordinary diary entry once you confirm it. It counts toward daily calories and macros, contributes to micronutrient coverage, receives a Food Grade and a satiety estimate, and appears in the same daily and weekly views as anything you typed by hand or scanned by barcode.

Common questions

Does the meal scanner replace weighing food?

No. It estimates portions from a photo, which has no scale reference, so it will not match a kitchen scale. Use it to keep logging consistent on ordinary days, and weigh food when a specific number needs to be right.

Is the meal scanner free?

The free plan includes a small monthly allowance of AI meal scans so you can try it before paying. Paid plans raise that allowance. Manual logging, barcode scanning, Food Grade, satiety, and micronutrient tracking are not limited by the AI allowance.

What happens to my meal photos?

Photos are processed to produce the food estimate. What is retained, for how long, and which processing provider is involved are described in the Privacy Policy, and AI processing requires your consent inside the app.

Can I correct a scan that got it wrong?

Yes. Every detected item can be edited, swapped, or deleted before you log it, and the "Fix result" flow lets you describe the correction in plain language rather than rebuilding the meal by hand.