PixWit

🖼️ In pictures

Every screenshot below is the real PixWit interface — open the tutorial for the full details

🔍Capture and recognize in one step

Press Ctrl+O and the screenshot is recognized straight away, without opening the editor first.

• A tiny model ships with the app; download the higher-accuracy small model once and it works offline
• You can also use PixWit AI or your own Baidu OCR credentials
• The whole result is copied to the clipboard automatically, with a "✓ Copied N lines" toast

Read the text recognition guide →

🌐Image translation drawn back onto the picture

10 target languages and three styles: literal, idiomatic or localized.

• Text mode shows the source and the translation side by side, both directly editable
• Image mode overlays each translation on the original; every box can be dragged and resized
• Saving the image renders the translations into the file, named Translation_Screenshot…

Read the translation guide →

▦⤴Tables straight into Excel

Press Ctrl+E, and the table lands in your clipboard.

• Both plain text and HTML clipboard formats are written, so Ctrl+V in Excel just works
• Rows, columns and merged cells are preserved — no Paste Special needed
• A failed recognition leaves your clipboard untouched and explains what went wrong

Read the table extraction guide →

💡Let the AI read the screenshot for you

OCR extracts the text first, then a language model turns it into a structured reading.

• Summary, key points and suggestions — useful for long documents, contracts and papers
• Your own service comes first, then PixWit AI; when the cloud quota runs out it falls back to the recognized text
• The result window matches the translation one: both panes are editable and copyable

Read the AI insights guide (Chinese) →

🎥Recording with a picture-in-picture camera

Press F3 to start, F3 again to stop, F4 to pause and resume at any moment.

• A 3-second countdown starts the recording; press F3 during it to cancel
• Annotate live while recording — until you pick a tool, the cursor passes straight through
• System audio and microphone can be recorded together with separate volume, output as MP4 / MKV / GIF

Read the screen recording guide →

🎬Post-production without leaving the app

Trim, compress, change speed and dub in the same window you recorded in.

• Zoom the timeline up to 1000%, cut out middle segments, and single-segment trims are lossless
• Three compression levels save up to about 70%, and 0.5x–3.0x speed changes leave frames untouched
• Also convert to GIF, dub the video, or split the audio into fixed-length files

Read the video guide →

📖 First steps: the toolbar and the tray ​

PixWit starts as a slim toolbar pinned to the screen (60px wide, always on top) while the program keeps running in the system tray:

• The toolbar: a drag handle ⋮⋮⋮ on top, then Region capture, Open image and Record screen. You can also drop an image or video file straight onto the toolbar. The × at the bottom only hides the toolbar — the app keeps running in the tray.
• Auto-hide at the screen edge (on by default): drag the toolbar to an edge and it snaps there; after the cursor leaves it retracts to a thin strip, and sliding back over that strip brings it out again. Dragging it to the middle of the desktop disables the behaviour. Hidden is not the same as exited — the F1 and F3 hotkeys keep working.
• Tray menu: right-click the tray icon for every entry point — Screenshot history (browse / open folder / change save folder / clear), Settings, Show main window, Start with Windows, dark / light theme, language, Help (User Guide / Check for updates / Feedback / Open log file / Website) and Exit.
• Single instance: launching PixWit again does not start a second copy — it brings the existing toolbar up and on top.
• Start with Windows: tick the tray entry. Administrators get a scheduled task, standard accounts get a startup entry.

Full details in the User Guide and the installation guide (Chinese).

🧭 The full feature list ​

Everything PixWit can do, grouped the way the user guide is organised. Each line links to the detailed tutorial.

📸 Capture & annotate ​

• Three selection modes: free selection (default), fixed aspect ratio (1:1 / 4:3 / 9:16 / 16:9) and fixed-size "stamp" mode (800×600 / 1024×768 / 1280×720 / 1920×1080 / 500×500 / custom 10–9999 px) — Screenshots
• UI element detection: window-only (fast) and element-only (precise) modes; press Tab to cycle and scroll to step through overlapping windows and controls, and the level you picked sticks — Screenshots
• Annotation tools: arrow (optionally with a caption), pen, rectangle, ellipse, line, polyline, text, numbering and mosaic, with stroke/fill, solid/dashed, opacity and spotlight — Screenshots
• Behaviour switches: mask opacity, full-screen capture, cross-monitor capture, double-click to copy and close, auto-finish after dragging, show on-screen hints — Screenshots
• Scrolling screenshot: automatic stitching every 0.5 s or manual scrolling, pauses when the cursor leaves the region, and the result becomes a sticky note — Scrolling screenshot (Chinese)

📌 Sticky notes, history and personalisation ​

• Sticky notes: borderless, always on top, 10%–500% zoom, Ctrl+wheel for opacity, double-click to copy and close, plus crop, rotate and mirror — Sticky notes (Chinese)
• Clipboard sticky: Ctrl+Shift+V pastes the clipboard image under the cursor, and pressing it again walks back through the clipboard history; plain text is rendered as a note card first — Sticky notes (Chinese)
• History: grouped by today / yesterday / this week / earlier, starrable and filterable, searchable by the text inside a screenshot (requires the OCR index), with a configurable folder and retention policy — History (Chinese)
• Color picker: an 8x magnifier follows the cursor and copies the color as RGB / #HEX / HEX — Color picker
• Watermark: text, size, color, font, position (including a 45° tiled pattern), margin and date-time, previewed live and burned into the output — Watermark
• Personalisation: 8 interface languages, dark and light themes, custom tray icon and tray click behaviour — Settings

🎥 Screen recording ​

• Region and full-screen recording: a 3-second countdown, pause and resume at any time, and paused time is not counted — Screen recording
• Live annotation: until you pick a tool the cursor passes straight through, then Alt+I / Alt+J / Alt+O / Alt+K draw lines, arrows, ellipses and rectangles — Screen recording
• Camera picture-in-picture (full-screen only): the camera feed sits in the bottom-right corner and can be dragged wherever you want it in the final video — Screen recording
• Audio: silent / system audio / microphone / both, with independent volume for each channel when recording both — Screen recording
• Output and quality: MP4 / MKV / GIF, three quality presets, 1–120 fps and hardware encoding; long recordings are split into 10-minute segments automatically — Screen recording

🤖 AI capabilities ​

• OCR: local models (built-in tiny, downloadable small) and cloud services can be combined; the result is copied automatically and clicking any line copies just that line — Text recognition
• Image translation: 10 target languages and literal / idiomatic / localized styles; image mode overlays the translation on the original and saves them together — Image translation
• Table extraction: the result goes to the clipboard in plain text and HTML, so pasting into Excel keeps rows, columns and merged cells — Table extraction
• LaTeX formulas: Ctrl+M extracts formula source with a multimodal model, one formula per line — Text recognition
• AI insights: OCR first, then a language model writes a summary, key points and suggestions — AI insights (Chinese)
• AI chat: multi-turn questions with the screenshot attached, switchable models and up to 20 prompt templates — AI chat (Chinese)
• QR codes and barcodes: recognized instantly and the first result is copied to the clipboard — Text recognition
• Selectable providers: local OCR only (offline), PixWit AI (works out of the box) or your own AI service and Baidu OCR, routed by a fixed priority — Settings

🖼️ Images and video ​

• Image viewer: 10%–1000% zoom, fit-to-window or actual size, full screen, previous and next — Images (Chinese)
• Crop: rectangle or ellipse, free or 1:1, plus ID-photo presets (1-inch, 2-inch, small 1-inch, ID card, passport) — Images (Chinese)
• Perspective correction: click the four corners to straighten a skewed document or whiteboard — Images (Chinese)
• Resize and quality: 1–10000 px, 0.01–10x scaling and JPEG quality compression — Images (Chinese)
• Save, convert and print: PNG / JPEG / BMP / GIF / WebP / PDF / ICO conversion, and print with automatic paper orientation — Images (Chinese)
• Video editor: trim (including deleting middle segments), compress, change speed, convert formats, dub and split audio — six tabs — Video
• GIF editor: crop, change speed and resize, plus frame-by-frame editing, drag-to-reorder and duplicate-frame detection — Video

⚙️ Settings and hotkeys ​

• Seven settings pages: General / Screenshot / Annotation / Recording / Hotkey / Saving / AI Service Config — Settings
• 16 global hotkeys, all remappable or disableable, with a tray warning when a combination is taken — Shortcuts
• Saving: default format, image quality, file name prefix and separator, date format and the screenshot retention policy — Saving
• Updates and feedback: automatic update checks, both GitHub and Gitee download sources, and three ways to deal with a download blocked by Windows or an antivirus — Updates and feedback (Chinese)

Want the per-module walkthrough with every default value? See the feature overview (Chinese). For hands-on efficiency tips see tips and tricks (Chinese), and the FAQ answers the most common questions.

💬 Join the community

📱 QQ group: 1091085813

GitHub Gitee