Why a photo estimate is never exact — and why that's fine

Point your phone at a bowl of pasta and Calerio gives you a number in about three seconds. That number is a guess — an informed one, but a guess. No camera can tell you exactly how much olive oil is under the sauce, or whether the cheese is regular or reduced-fat. Anyone who tells you a photo can produce a lab-grade calorie count is selling you something.

What the model is actually doing

Calerio's model looks at portion size, likely ingredients, and typical preparation for that kind of dish, then compares it against a large database of similar meals. It's closer to how an experienced dietitian would eyeball a plate than to a kitchen scale. That's a genuinely useful estimate — it's just not a measurement.

Why the gap closes over time

The first estimate on a new dish is the least accurate one you'll get. Correct a portion size once — tell it the chicken breast was smaller than it guessed — and Calerio remembers that correction the next time you log something similar. Over a few weeks, the numbers for your actual, repeated meals get noticeably tighter.

The honest way to use an estimate is the way you'd use a bathroom scale: not perfect, but consistent enough that the trend means something. One meal being off by 15% barely moves your week. A pattern of consistently misjudged portions is worth correcting, and that's exactly what editing a result does.