Aria
Scan a Recipe, Type the Numbers, Adjust the Grams
A plate photo is still the fastest way to log lunch. Plenty of meals never sit in front of the camera that way. A recipe card does. A package panel does. And a scan that read 140 grams of rice when you served 90 needs a correction that does not start the photo over. Fuel now has two more ways to log, and a tighter review on AI Meal Scan.
Scan Recipe
On Log a meal, Scan Recipe sits beside Text Search, Scan Barcode, and Favorites. Take a photo, or choose one from your gallery. A cookbook page, a recipe on a screen, and a handwritten list all work. Three small habits make the read cleaner: fill the frame with the ingredient list, keep the page flat, and include the “serves” line.
The scan captures the card itself:
- Recipe name, when one is printed.
- Servings or yield — the “serves 4” line, so the log can start at one portion.
- Each ingredient’s name, plus a search label for the nutrition databases.
- Quantity and unit — grams, ounces, pounds, milliliters, teaspoons, tablespoons, cups, and the other common kitchen measures.
- Edible grams for the whole recipe, which the app divides by the number of servings before you review.
Review opens on one serving. You also see an approximate calorie total for the whole recipe, and a servings control if you ate two portions. Calories, protein, carbohydrates, fat, and micronutrients then come from USDA FoodData Central and the Canadian Nutrient File (CNF) — the same reference path AI Meal Scan uses. The photo pass reads ingredients and amounts. The databases supply the nutrition.
A line that says “salt, to taste” or “a handful of herbs” stays open. Those amounts sit at zero grams until you set them, so a vague phrase is never turned into a silent weight. If an ingredient has no match in either database, its macros can still be estimated and labeled as an AI estimate, and micronutrients stay blank on that row. Ingredients with no usable match stay out of the totals.
Manual entry, when you already have the numbers
Manual entry sits under the logging grid. It is the additional logging option for greater accuracy when a photo would only be guessing: a package panel, a restaurant nutrition sheet, or a recipe you have already worked out. You type the figures you trust, and those figures are what the day records.
The form captures:
- Meal name
- Calories
- Protein, in grams
- Carbohydrates, in grams
- Fat, in grams
The name and the calories are required. After the meal is added, AuraBase looks that name up in USDA or CNF and scales micronutrients from the matching reference row, using the calories and protein you entered. The form asks for macros. The log can still carry vitamins and minerals, marked at medium confidence when the match is direct. A name that does not match a reference food keeps the macros you typed.
Adjust each ingredient on AI Meal Scan
AI Meal Scan still starts with a photo of the plate. The new part is the review, after the ingredients are matched. Each detected ingredient is its own row: the name, the USDA or CNF food it matched, and a gram amount. Plus and minus move that ingredient in five-gram steps. When the amounts look right, Apply and Update Totals rebuilds the meal.
The rebuild stays on the same database match. Calories, protein, carbohydrates, fat, and micronutrients scale with the new grams. The photo is not sent again, and the food is not looked up a second time. Set the chicken from 120 grams to 180, and that row’s nutrients move with the weight. Rows labeled as an AI estimate scale their macros the same way and stay without micronutrients. Rows with no database match stay out of the totals.
The same gram controls appear after a Scan Recipe review, so a cup that converted a little high can be corrected before you confirm. For how the match itself is built, see Inside AuraBase’s Hybrid AI Meal Scanner V2 and What Are AI Calories?
Which path to use
- Snap the plate when the food is in front of you, then fix any ingredient that looks off.
- Scan the recipe when you are cooking from a card and want one serving, with the option to log more.
- Type the numbers when the label, or your own math, is the source you want the day to remember.
All three land on the same day: calories, macros, and, where a reference row exists, micronutrients you can open later.
Product notes
- AuraBase Scan Recipe: a photo of a cookbook page, screen, or handwritten list. The scan captures the recipe name, servings, ingredient names, quantities, and units, then divides whole-recipe grams by servings. Calories, macros, and micronutrients come from USDA FoodData Central and the Canadian Nutrient File.
- AuraBase Manual entry: meal name, calories, protein, carbohydrates, and fat. Micronutrients are filled from a USDA or CNF lookup of that name, scaled from the calories and protein you entered.
- AuraBase AI Meal Scan review: each ingredient has a gram control in five-gram steps. Apply and Update Totals rescales calories, macros, and database micronutrients on the existing match.
- See also: /blogs/aria-ai-meal-scanner-v2 — hybrid scanner architecture.
- See also: /blogs/what-are-ai-calories — how AI Calories are calculated.