The Compare plugin uses an AI vision model to read your photos and explain a verdict. Here is what that means, what it does well, and when to double-check.
After you add your photos and continue to the Result step, the images are sent securely from your signed-in account to an AI vision model. The model reads the text and details in each image and returns a structured answer, so the plugin can show a consistent result every time: a winner or tie with a summary for price tags, or ratings and notes for wines.
The model reads prices, sizes, producers, vintages, regions, and other visible text.
For price tags it calculates unit prices; for wines it weighs the quality indicators it can see.
You see the winner, ratings, and the reasoning behind them.
It extracts price and size from a clear tag and calculates a unit price faster than mental math in an aisle.
When a unit price is printed on a tag and disagrees with its own figure, the printed one is preferred.
If unit prices are effectively equal, or neither tag is readable, the result is a tie rather than a forced answer.
Each result includes reasons, so you can verify them against the physical tag, label, or menu.
| Limit | What you can do |
|---|---|
| Blurry, dark, or glare-covered photos | Retake it. See photo tips. |
| Cut-off or hidden text | Make sure price, size, and unit price are all inside the frame. |
| Unusual or very small print | Move closer and confirm the figures yourself on the tag. |
| Wine ratings from a label only | Treat as guidance; taste and budget still matter. |
| Unreadable labels | The notes say so and the rating is conservative. Try another photo. |
Comparisons require you to be signed in, and each account has a daily comparison limit to keep the service fair. If you reach it, the plugin asks you to try again the next day. For how photos are handled and stored, see the privacy information available in RoomTopics.