You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Screen capture reworked into an in-app "action capsule" flow: triggering 识图取字 or the OCR translate/explain/summarize shortcuts dims the screen instantly, dragging draws the selection with a blue border, and releasing captures it silently — no system-capture flash. The four actions (Recognize Text / Translate / Explain / Summarize) appear as a capsule at the selection's bottom-right and recognition runs only after you pick one; the selection can be dragged with the capsule following (actions re-capture the new area), and clicking outside the selection or an accidental tiny drag cancels and cleans up. The translate/explain/summarize shortcuts skip the capsule and run right after release, and the main window stays hidden throughout.
The capture selection can be resized from its edges and corners: an 8 px grab zone around the border shows a resize cursor, dragging resizes with the opposite edge anchored (8 px minimum, clamped to the screen), and the capsule follows while actions capture the resized area.
Changed
The 识图取字 shortcut description in Settings now reads "Select any area of the screen, pick an action, then recognize", matching the new flow.
The result window's original-text area now uses neutral highlight colours: text selection and the hover hint are a faint ink/paper tint with no blue-purple hue in either theme; selecting and correcting the original text work exactly as before.
Fixed
On multi-monitor setups the result window no longer lands on the wrong screen or gets cut in half at a screen edge: it now opens on the screen where the action happened and is clamped to that screen's visible area, with remembered positions tied to their screen.