How to translate text on screen that you cannot copy

Text in video subtitles, game quests, and images often cannot be selected. Dragging the mouse over it does nothing, so you cannot copy it into a translation tool.

When this happens, start by deciding what result you need.

• To look up one word or sentence under the pointer, use Cursor Lookup.
• To translate an area of the screen directly, use Screenshot Translation.
• To extract, copy, or organize the original text, use Screenshot OCR.
• If the text can be selected normally, use Selection Translation.

Some on-screen text is only part of an image. The application itself has no characters that can be copied. Cursor Lookup, Screenshot Translation, and Screenshot OCR all recognize text from the image first. The difference is whether you receive a lookup result, a translation, or editable source text.

Look up one word or sentence with Cursor Lookup

Move the pointer over the text and trigger Cursor Lookup. Float Translate reads the word or sentence near the pointer and displays the corresponding result.

It works well for short content such as an item name in a game or a word in an image. Because the source text does not need to be selectable, it can handle text that ordinary Selection Translation cannot access.

For a full quest description or a larger block of text, Screenshot Translation is more direct.

Translate an area directly with Screenshot Translation

Screenshot Translation is suitable for longer text that cannot be selected, such as video subtitles and game quests.

Trigger Screenshot Translation, then draw a box around the area containing the text. Float Translate recognizes the text inside the box and shows the translation in the translation panel. The entire process runs as one action: you do not need to open Screenshot OCR first or configure a workflow yourself.

Keep the selection close to the text you need. Extra icons, buttons, and complex backgrounds can interfere with recognition. If the screen contains chat messages or account information, avoid including unrelated details.

Use Screenshot OCR when you need the original text

The main output of Screenshot OCR is the recognized source text, not an immediate translation.

Use it when you need to copy text from an image, inspect recognition line by line, keep the current screenshot, or continue editing and organizing the content.

If you only want to understand the screen as quickly as possible, Screenshot Translation removes the extra steps of checking and passing along the OCR result.

Handling text in images, videos, and games

All three situations may require screen recognition, but the best way to select the area differs.

Text in images

First zoom in until the edges of the characters are clear, then select only the part you want to translate. Use Screenshot Translation when you only need to understand it. Use Screenshot OCR when you need to copy or reorganize the original text.

If an image contains body text, buttons, and decorative lettering, do not select everything at once. Process it by paragraph or region to preserve the reading order more reliably.

Video frames and subtitles

Pause the video and wait for the subtitle and animation to become still before selecting the area. Subtitles are usually near the bottom of the frame; select only the current subtitle rather than capturing the whole picture.

If the player provides selectable subtitles or a subtitle file, use the original subtitles first. This avoids recognition errors caused by image quality, compression, and moving backgrounds.

Game dialog, menus, and prompts

Cursor Lookup works well for short content such as item and skill names. Screenshot Translation is better for quest descriptions, dialog, and longer prompts.

Wait until all of the text is visible, and avoid changing images or unrelated content. If the game provides a text log, select the text from the log first.

For other application interfaces, also try selecting the text normally before using OCR. Use Cursor Lookup, Screenshot Translation, or Screenshot OCR only when the current text truly cannot be selected.

If the text is selectable, use Selection Translation

Text in webpages, documents, and chat applications does not need OCR as long as it can be selected normally.

Select the text and trigger Selection Translation to view the translation directly. Because the current application provides the source text itself, this is usually faster and does not introduce OCR recognition errors.

What to do when recognition is inaccurate

Screenshot Translation and Cursor Lookup both depend on screen text recognition. If characters are missing, incorrect, or split into the wrong sentences, try these four steps first.

• Enlarge the text. Zoom in on the video, image, or application window before recognizing it again. Similar characters can merge when the type is too small.
• Use a smaller area. Select one complete block of text at a time and avoid icons, buttons, and borders.
• Wait for a still image. Pause the video, wait for text animation to finish, and use a clear image when possible.
• Check special content. A single incorrect character can change the meaning of numbers and proper names.

When a single Screenshot Translation is not suitable

Screenshot Translation works best for still images or text that remains on screen for a while.

Continuously scrolling subtitles, rapidly changing game dialog, and an entire video require continuous processing. Use the video’s own subtitles, a subtitle file, or a dedicated live translation and transcription tool instead.

Remember this order

If the text can be selected, use Selection Translation.

If it cannot be selected, use Cursor Lookup for one word or sentence, Screenshot Translation for a whole area, and Screenshot OCR when you need to copy or organize the original text.