When one photo isn't enough
Some meals defeat a single frame: the curry hiding its ingredients, the plate that's half leftovers, the protein shake with no visual identity at all. Describe mode — the toggle next to Snap on the photo screen — lets you hand the analysis more than one photo and your own words, together.
What you can provide
- Up to 3 photos — shoot them or multi-select from your gallery. Different angles, the packaging next to the plate, the second helping.
- A written description — anything the camera can't know: "the bowl is half quinoa half rice", "cooked in about a tablespoon of butter", "only ate half".
- Either alone also works — text with no photos behaves like voice logging, typed instead of spoken.
Your words outrank the pixels
The rule that makes Describe powerful: what you state is treated as truth, overriding what the image suggests. Say "that's seitan, not chicken" and it's logged as seitan; state a quantity and it's kept rather than re-estimated. And multiple photos of the same meal are understood as one eating occasion — the same dish from two angles doesn't log twice.
Same review, same rules
The result lands on the familiar review screen — everything editable before it enters your diary. Describe needs a connection like all AI logging, and the same honesty applies: estimates worth a glance, never a source for allergen decisions.