How do I describe a before-and-after image without implying more than it shows?

Separate the job into two parts: identify the comparison, then say what is visible in each panel. W3C’s example for related images uses a shared caption to identify the group and individual descriptions for each image. That structure can help readers follow the comparison without treating it as proof of a cause or outcome. [S1]

First, check that “before” and “after” are supported labels, not guesses based on appearance. If you do not know the timing, use “Panel A” and “Panel B.” Then name concrete, observable differences: position, color, presence, shape, or visible text. Avoid interpreting those details as improvement, damage, treatment, or success unless other reliable information supports the claim.

For example, assuming the panels actually show these details: “Panel A shows a small mark near the upper edge. Panel B shows no visible mark in that area.” This describes the images; it does not say what caused the difference or whether the mark is gone outside the pictured view.

Yes, if “improved” adds a judgment the images do not establish. A visible difference can be described directly, but improvement depends on what counts as better and may require information beyond the pictures. Likewise, “the process removed the mark” makes a causal claim that the panels alone may not support.

Try replacing the interpretation with a specific observation:

1. Instead of “The surface improved,” write “The second panel shows a more even-looking surface.” Use this only if that is plainly visible.
2. Instead of “The repair fixed the crack,” write “A crack is visible in Panel A; no crack is visible in the corresponding area of Panel B.”
3. Instead of “The treatment worked,” state only the visible difference, or add a separately supported explanation with its source.

If the feature is not clear, partly hidden, or shown at different angles, say so or leave the comparison out. A caption cannot establish what happened between images.

Give the pair a shared comparison label, then distinguish each image with its own short description. For example:

“Two views of the [object or subject]. Panel A: [observable detail]. Panel B: [observable detail].”

If the images have verified dates or other useful context, include those; do not infer dates from their order. W3C’s related-image example follows this general structure by identifying the group and providing an individual description for each image. [S1] That example supports the structure, not particular wording for your images or a method for proving what caused a difference.

Keep descriptions focused on details needed to understand this comparison. Avoid repeating the same full caption for both panels, and do not use an image description to smuggle in a claim—such as “successful result”—that the visible details do not establish.

Read each sentence and ask: “Could a reader verify this from the panels or confirmed accompanying information?” If not, remove it, qualify it, or attribute it to supporting information. Then check that the panels are identified consistently and that your wording does not turn a visible difference into a cause, judgment, or promise.

Copyable template:

“Comparison: [what the two panels show]. Panel A ([confirmed label/date, if available]): [specific visible detail]. Panel B ([confirmed label/date, if available]): [specific visible detail]. [Optional limitation: e.g., ‘The images show the two views but do not establish what caused the difference.’]”

Keep the optional limitation only when it helps prevent a likely misunderstanding. The template cannot verify the images or their labels for you; those details must come from the material you have.