Why do older WooCommerce photos stay blurry while new uploads look sharp?

Separate the Three Sizes

Inspect the affected product image in three places: the original attachment pixels, the generated size currently selected by the page, and the dimensions at which the browser renders that file. A sharp original does not prove that the displayed thumbnail is sharp; the page may be loading a smaller generated file and enlarging it. Record the image URL or filename, pixel dimensions, and displayed dimensions for both an old image and a new image.

A theme or image-sizing change can leave older thumbnail files in place. Also check whether the image appears soft because of cropping or because a small file is being stretched. Hard cropping removes edges to fit a target shape; it does not increase detail or make an undersized source sharper.

Build a Controlled Fixture

Use a non-destructive test copy of a product or a separate draft product. Choose two distinguishable assets before comparing them: Asset A is an existing older product photo that currently looks blurry, and Asset B is a new upload of a different photo whose original pixel dimensions are recorded before upload. Keep the display context identical so the comparison isolates the image-generation history rather than product layout differences.

The new asset should be large enough for the intended rendered dimensions, but do not assume a particular minimum or maximum. Record the original pixels for both assets, the generated sizes available after upload, and the rendered dimensions visible in the product view. Inspect the browser’s loaded image dimensions or equivalent media evidence rather than judging only from the editor preview.

Illustrative exercise

Illustrative exercise: define Asset A as the old blue-mug attachment, with an original of 1600 by 1600 pixels and a displayed size of 800 by 800 pixels. Define Asset B as a newly uploaded red-kettle attachment, with an original of 1600 by 1600 pixels and the same displayed size of 800 by 800 pixels. These dimensions and names are illustrative only; replace them with the values you actually observe.

Compare A and B in the same product-image position. If A loads a smaller generated file than the page needs while B loads a suitably sized generated file, the comparison passes as evidence of stale generated sizes. If both load suitably sized files but A is still soft, inspect A’s original pixels and any crop or source-quality difference. The fixture passes only when the suspected cause explains the observed file dimensions and the corrected test image is visibly sharp at the same rendered size.

Action Sequence and Branches

1. 1. Create the test copy, identify Asset A and Asset B, and record original pixels, generated sizes, loaded file dimensions, and rendered dimensions.
2. 2. Compare the loaded files. If the old image uses an undersized generated file, regenerate the relevant image sizes using the available regeneration process, then retest the same copy. Do not regenerate the entire catalog before this controlled check.
3. 3. If the old image’s generated file is suitably sized, inspect the original attachment. If its original pixels are already soft or smaller than the rendered requirement, replace that attachment with a sharper source; regeneration cannot add missing detail.
4. 4. If both files are suitably sized but only one looks wrong, compare crop behavior and the exact displayed source. If hard cropping removed important edges, adjust the intended crop or source composition and retest.
5. 5. Accept the fix only when the old asset loads an appropriately sized generated file, the displayed dimensions remain unchanged, and the product view is sharp without changing unrelated media.

Evidence to Keep

Keep a short before-and-after record for Asset A: original pixel dimensions, generated filename or dimensions, rendered dimensions, crop observation, and the result after regeneration or replacement. This distinguishes a thumbnail-generation problem from an undersized original or a crop mismatch.

If the test copy improves but the live product does not, compare the live page’s loaded image file and rendered dimensions before repeating the change. A cache, template variation, or different attachment reference may mean the two pages are not using the same generated size; treat that as a separate observation rather than assuming the regeneration failed.

Scope of the documentation

The official references below support tool behavior and feature boundaries. The test fixture, branching decisions and acceptance criteria are Itemlume editorial recommendations; citation does not imply endorsement. Illustrative sizes, names and timings are not your production specifications.