Identify which layer contains the words
A public post can contain text in three places: its written caption, the image pixels and an added image description. Before using a phrase to search for the original graphic, record which layer supplied it. These layers can be related without saying exactly the same thing.
For example, an illustrative poster might display an event title while its description explains the layout. Searching the descriptive wording and calling it a quotation from the poster would blur that distinction.
Read the description through the badge
X explains that an ALT badge indicates an added image description. Select it to display the text, then dismiss the panel when finished. A GIF label is different: it signals moving media rather than opening a description. A GIF can also have a separate ALT badge. Source
X’s author instructions describe entering the description through +ALT and adding descriptions separately to multiple images. Source Check the particular image you are reviewing; do not assume one description applies to every image in the post.
Create a three-layer transcription note
| Layer | What to record | What to avoid |
|---|---|---|
| Post text | Relevant wording outside the image | Attributing it to a document pictured inside |
| Visible image text | Words you can actually read | Completing cropped or illegible words by guesswork |
| ALT description | Relevant text shown by the badge | Treating a descriptive sentence as automatic OCR |
You do not need to copy an entire post into the note. Keep the minimum wording necessary for your question and the actual post link. Mark uncertainties where they occur, such as an unreadable character or a description that refers to something outside the crop.
Choose the next search clue
Use clearly visible wording when your task is locating the graphic that contains those words. Use descriptive details as a separate search attempt when the visible wording is insufficient. Label the attempt accordingly so that a later match does not retroactively turn the description into a verified quotation.
If the layers disagree, retain both observations. An author’s description is useful source context, but its presence does not resolve the discrepancy. Open any resulting source page and compare the actual item before drawing a conclusion.
The screenshot-text workflow covers uncertain characters and cropped context. The image-source workflow covers checking a candidate destination. Neither a matching description nor a similar picture independently establishes a person’s identity.