AI photo logging
Point your camera. Skip the search bar.
Most trackers make you hunt a database for “the green one in the photo.” MacroTrackr looks at the plate, names what's on it, and estimates the portions, so logging a meal takes one photo instead of five searches.
How photo logging works
Snap it
Photograph your plate. The model identifies each item and estimates grams.
Or describe it
“Two eggs, sourdough, half an avocado.” Plain English or voice dictation works just as well as a photo.
You approve it
Nothing lands in your log automatically. The AI drafts, you review and save. Every single time.
Confidence you can actually see
Each detected item carries a confidence score. Anything the model is unsure about is flagged before you save, so you know exactly what's worth a second look instead of trusting a hidden guess.
Corrections in two taps
Fix a portion, swap an item, or remove it entirely before it counts. The corrected meal is saved to your food list on the spot, so the next one logs in a single tap with the exact numbers you already confirmed.
What it costs you
Every account gets 3 AI analyses to try, shared across photo and text, no card required. Pro raises that to up to 15 photo or text logs per day, refilled at midnight. Barcode scans and re-logging a saved meal never spend a credit, on any plan, so the AI is there for the meals that actually need it.
Who reaches for this
Restaurant meals
You did not cook it, you cannot weigh it, and the menu lists no numbers. A photo gets you a defensible estimate in seconds, with the items broken out so you can correct the one that looks wrong rather than guessing at the whole plate.
The end of a long day
Motivation to log is lowest exactly when the day has been hardest. Pointing a camera survives tiredness in a way that five database searches does not, and a rough logged meal beats an accurate one you never entered.
Food you have no name for
Someone else's cooking, a dish from a cuisine you do not know, a bowl of several things at once. Searching requires knowing what to type; a photo does not, which is the whole point.
Photo logging questions
How accurate is the AI on a photo?
For common plated meals it usually gets the items right first time. Homemade dishes, mixed bowls and anything hidden under a sauce are harder, which is why every item carries a confidence score and nothing is saved until you approve it.
Does it save the meal automatically?
No. The AI produces a draft. You review it, fix any portion, remove anything wrong, and only then does it count toward your day. There is no path where a photo lands in your log without you seeing it.
Do I have to use the camera?
No. Describing the meal in plain English works just as well ("two eggs, sourdough, half an avocado"), and so does voice dictation. Both run through the same model and cost the same credit.
Does the AI learn my eating habits over time?
It does not. Each analysis is a fresh, stateless call: nothing about your history is used to train a model. Fast re-logging comes from your saved meals and pinned shortcuts, not from personalisation.
Does it work on a plate of leftovers?
Usually, though mixed and reheated food is where confidence scores drop. Treat a low-confidence item as a prompt to type the portion yourself rather than a reason to distrust the whole draft.
Do my meal photos get stored?
The photo is sent for analysis and the resulting meal is saved to your log. Your data is encrypted in transit and at rest, it is never sold, and you can export or delete your log from your profile at any time.
Log tonight's dinner
from a photo.
Free to start, no card. Three AI logs to try; barcode scanning and search stay free after that.