Crop the photo to the part you are asking about
A photo of a whole sign, board or menu makes the AI hunt for your line in small print. Crop to the part you are asking about, keep the column names, and the answer gets easier to check.
Why it matters
When you send a photo of a full departures board, a menu or a receipt, the AI has to find your one line among dozens.
Anthropic's prompting guide says testing showed "consistent uplift on image evaluations when Claude is able to 'zoom' in on relevant regions of an image."
You can do the zoom yourself before you send the photo: crop it.
Do this
- Take the photo as usual.
- Crop it to the row or area you are asking about.
- Keep the headers or column names in the crop, so the AI knows which number is which.
- Ask your question, then check the answer against the real sign.
A line to send with the cropped photo
This photo is cropped to the row I care about. Read only this row and tell me the value under each column name.
One thing to keep in mind: Crop too tight and you cut off the column names. Then the AI has to guess which number is the gate and which is the flight.
In Anthropic's docs the zoom is a crop tool the model uses. Cropping the photo yourself before you send it applies the same idea. The airport story in the video is an example. Source: Anthropic: Prompting best practices, "Improved vision capabilities".
Ask me anything about AI and cloudA paid call on your own setup: your agents, your bill, your plan.Book a call →