How to localize help-center screenshots
A help-center screenshot has a specific job: show the reader where to click. A visually polished translation can still confuse people if its labels do not match the live product or the surrounding instructions. Start with the reader’s interface language, not with the image editor.
Match the interface your reader uses
If the product is localized, capture its actual localized interface when possible. If it is only available in English, replacing every English menu label with translated text may make the instructions harder to follow. Keep the real label visible and explain its meaning in the surrounding article instead.
Decide which words are interface labels and which are editorial annotations. Translating an annotation such as “Choose your workspace” is different from inventing a localized button label that does not exist in the product.
Prepare a safe source image
Use a demonstration account. Remove customer names, email addresses, API keys and private project information before sending an image to a cloud service. Opening a file locally in ImageL10n is separate from choosing cloud translation, which uploads it for processing.
Capture the relevant panel at a readable resolution. Keep enough surrounding context for readers to locate the control. Cropping away every landmark can make a clean-looking screenshot less useful in a step-by-step guide.
Keep a small terminology checklist
Write down product names and recurring labels before editing: Workspace, Team, Invite, Save changes. Choose translations that agree with the actual product and the help article. Review each screenshot against that list.
ImageL10n does not currently provide a saved glossary or team review workflow. For the first release, maintain the checklist outside the editor and apply it during manual review. Do not rely on independent generation requests to always pick identical wording.
Translate and review one image at a time
Open the source PNG or JPG, choose a target language and use Translate image when cloud processing is available. Review the resulting text, then adjust text boxes, fill colors and font sizes. The current AI workflow processes one image into one language per request.
Compare the result with the article’s numbered instructions. If the article says “Click Save changes”, use the corresponding label in the screenshot. Check that arrows, step markers and emphasis still point to the correct control after text changes.
Publish with context and a maintenance plan
Export the image at its original dimensions and inspect it inside the actual help article. Add alt text that describes the relevant action or state, rather than a list of search keywords. The instructions should still make sense when the image is unavailable.
Use a consistent filename and record the product version or capture date in your documentation workflow. When the interface changes, revisit both the source screenshot and its localized variants. Cloud Save edits can preserve an ImageL10n document, but it is not a documentation version-control system.