Release notes
What's new in CUTFORGE for video creators.
[0.9.41] - 2026-08-27
Changed
- Refunds - a Paddle refund or chargeback now ends the matching PRO licence automatically.
- Privacy - site analytics store a hashed IP (not the raw address) and delete events after 90 days.
- Export - black-and-white and other color looks no longer fail when exporting.
[0.9.40] - 2026-08-26
Added
- SAM 2 tracking (PRO) - a far more reliable way to follow a person, animal, or object for 9:16 reframe. Instead of looking for the subject again in every frame, it remembers what it is tracking, so it holds on through movement and brief occlusions. Requires a PRO licence and a dedicated graphics card; CUTFORGE downloads a separate 162 MB module the first time you use it. You can also click the subject in the middle of the clip - it is then followed both forwards and backwards.
Removed
- YOLO26 tracking - removed, it was not accurate enough to keep. Use Track, or SAM 2 with a PRO licence. Existing projects that used it switch back to Track on their own.
Changed
- Tracking module 1.7.0 - smaller download, 75 MB instead of 84.
[0.9.39] - 2026-08-25
Added
- YOLO26 tracking - in Object / animal tracking, a YOLO26 button (newer detector) next to Track. Needs tracking module 1.6.0 (the app installs or updates it).
- Auto subject - pick the interesting region automatically (no click), then follow it for 9:16 reframe.
Fixed
- YOLO26 tracking - following a subject no longer fails right away with "No frames sampled". Reinstall tracking module 1.6.0 if it was already installed.
[0.9.38] - 2026-08-19
Fixed
- Cancel export is safer - stopping an export no longer risks killing the wrong Windows process.
- App logs stay tidy - old session log files are cleaned up automatically.
[0.9.37] - 2026-08-19
Improved
- User manual - reverse playback is now described correctly (preview shows reversed frames, REV badge). Tracking module size and 30 transitions match the app.
Fixed
- Replacing a video no longer keeps the old preview or face blur - if you overwrite a file with the same name, CUTFORGE rebuilds the preview and the export face mask.
- Exports with transitions no longer freeze at the final assembly step.
- Preview and face-mask caches stay bounded on disk.
- Cancel speech-to-text is more reliable.
[0.9.36] - 2026-08-13
Improved
- Face blur follows the face more closely in the preview - same detailed track as export, not a thinned copy.
- Advanced skin and "blur all faces" are real on export - skin uses a pixel mask; all faces blur each detected face (preview shows one oval per face).
- Reverse playback in the preview - you see the reversed frames while scrubbing and playing (a small REV badge). Export is still the full FFmpeg reverse.
- More transitions preview - fade to black/white, circles, zoom, and the other xfade looks now move in the preview, not only in the file.
[0.9.35] - 2026-08-13
Improved
- Clearer export errors - if export times out, the timeline is empty, or a clip is HEVC your FFmpeg cannot read, CUTFORGE now says so instead of a generic "FFmpeg failed".
- Wipe and slide transitions preview - those transitions now slide in the preview, not only in the exported file.
- Honest face-filter labels - Advanced skin is a softer blur on the tracked oval; "Blur all faces" uses one box around every face (including the gap between them).
Fixed
- Vertical recrop no longer falls back to letterbox if the animated crop file cannot be written.
- Speech-to-text will not start twice - wait for the current transcription to finish (or cancel it). Viral captions stay in one Undo.
- Mark beats is one Undo - no more hundreds of undo steps for a 3-minute track.
- Project name, frame rate, and aspect ratio can be undone.
- Reversed clips - face tracking uses the same time as export (the picture stays on the right face).
[0.9.34] - 2026-08-13
Improved
- Cancel transcription - you can now stop local speech-to-text while it is running (the overlay has a Cancel button). If it takes too long, CUTFORGE stops it instead of staying blocked.
- More precise face tracking on export - blur / recrop follow the face more closely on long clips.
Fixed
- Save no longer rewinds the project - Ctrl+S keeps your playhead, undo history, and selection.
- Old projects keep their sound on export - videos imported before audio detection was stored no longer export silent.
- Silence trim on moved clips - cutting silence now works even when the clip does not start at the beginning of the timeline, and the picture stays in sync.
- Undo after face tracking - other edits you make while a long tracking job runs are no longer undone together with the tracking result.
[0.9.33] - 2026-08-07
New
- Music tempo (BPM) detection - CUTFORGE now detects the tempo of your music automatically when you import an audio file or a video with audio. The BPM appears right away in the media library, and a "Mark beats" button places a marker on every beat of the track on the timeline - ideal for cutting to the rhythm.
- Automatic music fade - music placed on an audio track now fades in and out smoothly at the start and end of the video (0.5 s), instead of starting/stopping abruptly. You can disable it in the export settings.
- Machine specs - Help -> Specs shows your computer's hardware (processor, cores/threads, memory, graphics card) with a simple explanation, all local.
Improved
- More transitions - the transition list now includes fade to black/white, dissolve, wipes and slides in all directions, circle open/close, pixelize, radial and zoom in (20 transitions total). Added in 0.9.33: fade to grayscale, smooth up/down, circle/rectangle crop, horizontal/vertical slices, horizontal blur and vertical/horizontal squeeze (30 total).
- Color presets - five new ready-made looks: Cool, Black & white, Vintage, Teal & orange and Cinematic, applied with one click and adjustable.
- Intro/outro templates - the Timeline menu can insert a ready title at the start ("Add intro") and a "Subscribe!" outro at the end of your video.
- Info hints - a small "ⓘ" icon next to the main buttons explains in one line what each button does, in plain language.
Fixed
- Export reliability - exports no longer hang when a video has no audio track, when a render step stalls, or when a tiny fragment of the timeline is shorter than a single frame; the export now fails cleanly with a clear message instead of stopping forever.
[0.9.32] - 2026-08-06
New
- Super Track - a third button in Object/Animal tracking runs the template rectangle engine (ZNCC match) on the clicked subject; the tracked box outline turns on automatically after a successful run, with honest per-sample scores (real correlation, not model confidence). Ideal for subjects with a stable appearance (buildings, products, animals), while the GPU "Pick subject" stays the best choice for highly deformable subjects.
Improved
- Match score on the tracked box - with "Show the tracked subject outline" on, a live score badge shows the template-match percentage at the playhead, and the tracking section displays an overall tracking score /100 (mean of real matches, ignoring Kalman-predicted frames).
- Adaptive subject tracking - the classic rectangle/template tracking is now much more robust for deformable or moving subjects (swimming duck, birds): the tracking template refreshes after strong matches while the pristine anchor template stays as a fallback, the box size adapts softly to scale changes (±10 % per sample, clamped ±30 %), and the search window widens after misses to catch fast motion.
- Lighting-robust template matching - the subject tracker now scores matches with ZNCC (zero-mean normalized cross-correlation) instead of plain pixel difference, so shadows, glares or a global brightness drift no longer drop the match score; the best of two templates (live + anchor) is used at every frame, and each refresh blends the new patch with the current template (70/30) so the box never jumps on a single noisy frame.
- Fast Super Track analysis - template matching now runs on a downscaled frame with an integral-image mean (O(1) per position) and a narrower adaptive search window: a 1080p clip analyzes in roughly real time (~150x faster than the initial implementation), and cancelling/timing out a tracking run now kills the background process so it cannot keep writing phantom progress logs.
- Robust Super Track on moving subjects - the click template is tighter (mostly the subject, less background), the living template refreshes on every accepted match so it rides through posture changes, and the continuity grace accepts dips down to 0.45 - a full 44 s geese clip now tracks end-to-end (previously lost after ~2 s), with honest per-sample scores.
- Multi-core Super Track - the per-frame template search (2 templates × 3 scales) runs in parallel on `available CPUs − 1` workers (capped at 6, `CUTFORGE_TRACK_THREADS` overrides): a 44 s 1080p clip analyzes in ~8 s (was ~15 s), and one core stays free so the PC remains responsive.
- Original-format export - a new "Original (source format)" aspect option in the export settings keeps the video's native aspect ratio (no forced 9:16); subject tracking no longer switches the project to 9:16 automatically, so the source format is preserved by default. Smart-reframe cropping still applies when a 9:16 export is chosen.
Removed
- SAM subject segmentation removed - the SAM point-prompt mode (models, tracker mode, preview silhouette, blur-by-mask) has been removed; the tracking pack returns to 1.4.0 (no SAM models) and `--track-mode sam` is rejected. Existing projects with a SAM face filter load cleanly and fall back to the standard engine.
[0.9.31]
-
2026-08-02
New
- One-click viral subtitles - a new "1-click" menu in the top-right adds subtitles to your clip in one click: local transcription, viral karaoke style, smart placement (bottom, or top when the face is low in the frame), then the 9:16 export window opens ready to go.
- Faster STT startup - the speech-to-text module check is now near-instant, and the progress messages show each step (preparing, extracting audio, loading model, transcribing, aligning words) instead of a single stuck status.
[0.9.30]
-
2026-08-02
Security & Reliability
- Faster, safer smart-reframe exports
-
fixed an intermittent export failure/corrupt crop when rendering several smart-reframe segments in parallel.
- Subtitle fixes
-
more robust ASS subtitle escaping (carriage returns, special characters in file names).
- Import hardening
-
subtitle timestamps are now clamped to sane bounds; project files with invalid numbers are rejected instead of breaking the timeline.
- Autosave improvements
-
edits are no longer lost if you close the window right after making them; overlapping saves keep the newest version.
- Face tracking
-
analysis inputs are validated before the tracker starts; deleting a clip also removes its linked captions.
- Cleaner undo history
-
no more empty undo steps from tiny drag gestures.
[0.9.29]
-
2026-08-02
Security & Reliability
- Better error handling during export
-
clearer messages when something goes wrong (missing FFmpeg, empty timeline, timeout).
- Rate limiting on the operations dashboard
-
prevents abuse.
- Log sanitization
-
sensitive tokens are now masked in log files, keeping your private builds safe.
- Media path hardening
-
additional validation to prevent path traversal attacks.
- Project file size guard
-
byte-accurate checks for project files with special characters.
Performance & Stability
- Export PID tracking
-
lock-free mutex for faster, smoother export cancellation.
- Tracker timeout
-
face tracking now times out after 10 minutes instead of hanging forever on a stuck model.
- Single-pass export warning
-
a log warning now appears when using single-pass mode on timelines longer than 30 seconds, reminding you that segmented export is faster.
- Audio filter chain
-
extracted to its own function, reducing compositor complexity.
Developer Improvements
- 8 new error codes
-
stable IPC codes for export errors, making debugging easier.
- STT module reorganized
-
transcription code split into smaller modules for easier maintenance.
- Targeted clippy lints
-
15 specific functions now have allow attributes instead of a project-wide suppression.
- Tracker JSON protocol
-
groundwork for structured communication between the tracker worker and the main app.
- WhisperX standalone roadmap
-
plan to bundle WhisperX as a standalone executable in a future STT pack update.
[0.9.15] - 2026-07-09
Fixed
- Trim undo/redo - undoing and redoing a clip trim keeps the timeline and preview in sync (no more odd jumps or missing waveforms)
- Left trim handle at 0s - dragging the start handle left when a clip is already at the beginning of the timeline no longer stretches or pushes the clip to the right
[0.9.14] - 2026-07-09
Fixed
- FFmpeg setup - after automatic download finishes, CUTFORGE now recognizes FFmpeg immediately (no more return to the setup screen)
[0.9.13] - 2026-07-09
Fixed
- FFmpeg setup - first-launch download works again (previous download link had expired)
Improved
- Private beta updates - installers and optional packs (face tracking, speech) download via a faster, more reliable path
[0.9.12] - 2026-07-09
Fixed
- Phone videos filmed upright - face pick, tracking overlay, and blur stay aligned when the file has rotation metadata (common on phones)
- More reliable cancel - cancelling export or face analysis mid-run recovers cleanly without leaving the app stuck
[0.9.11] - 2026-07-09
Improved
- Face tracking vs privacy blur - SMART DETECT / Analyze track the face without applying blur; turn on privacy blur explicitly in the Face filter panel when you need it
Fixed
- Face blur preview - blur oval stays on the face; mask no longer animates while the video frame is still loading; blur stays in sync with playback
[0.9.10] - 2026-07-09
New
- Export quality checks - automated preview↔export parity tests for color, opacity, reverse, dual layout, and face filters so final MP4 stays closer to what you expect
[0.9.5] - 2026-07-08
Fixed
- WebM import - importing some VP9/WebM files no longer crashes the app during waveform analysis
- Face blur preview - blur now covers the full tracked face (ellipse mask) and stays in sync during playback
- Blur all faces - clearer message when the face-tracking pack is too old for crowd blur
- Preview playback - fewer double-audio glitches after pause; overlay videos stay in sync
- Caption trim - word timings stay aligned when you trim a subtitle clip to the playhead
[0.9.4] - 2026-07-07
Fixed
- Speech-to-text module download - private (CUTFORGE-TEST) builds download the Whisper pack again from the official proxy instead of a broken direct link
[0.9.3] - 2026-07-07
New
- Smarter captions - subtitles stay in sync when you split, trim, or run silence cut on the source clip
- YouTube chapters - generate chapter markers from your subtitles and export a ready-to-paste chapter file
- Speaker zoom - automatic punch-in on talking segments (Pro panel)
- Blur all faces - anonymize every face in the frame for crowd or GDPR-style exports
[0.9.2] - 2026-07-07
New
- Caption word tools - merge or split words (text and timing) in the Text inspector
- Timeline silence markers - quiet sections highlighted on the ruler; click to jump
- Dual versus - top/bottom split layout for reaction or comparison videos
- Transcript editing - delete spoken ranges from the transcript; timeline ripples automatically
- Frame fit - contain or pan when clip aspect differs from the project frame
- New project templates - Shorts (9:16), YouTube (16:9), or podcast audio when creating a project
Improvements
- Karaoke / word-pop export - bounce animation on highlighted words matches preview
- Speech-to-text - STT pack downloads on public builds without a private beta gate
Fixes
- Stability - dependency security updates (Rust advisory fixes)
[0.9.1] - 2026-07-05
New
- Custom window (Windows) - title bar with minimize, maximize, close, and drag (frameless app chrome)
Improvements
- Smart reframe - turning off "Follow face on 9:16 export" restores 16:9 preview and export when no other clip uses smart reframe
- Subtitle generation - clearer step-by-step progress (STT module, audio extract, WhisperX transcription and word alignment)
Fixes
- Faster startup - avoids double FFmpeg setup on launch
- Speech-to-text - fixes a bug where the STT module was re-downloaded on every run after the first transcription attempt (WhisperX pack)
[0.9.0] - 2026-07-04
New
- Appearance - View menu: Dark, Light, or Auto (follows Windows light/dark mode); your choice is remembered
Improvements
- Light theme - softer editor palette for daytime editing (timeline, menus, inspector, media bin)
- Stickers in light mode - clearer badge icons; drop shadow on preview stickers
- Sticker placement - drag aligns with export frame; stickers stay movable on image-only timelines
Fixes
- Security hardening - safer app startup, license checks, and optional pack downloads (no change to normal editing workflow)
[0.8.84] - 2026-07-03
Improvements
- Tracking quality - more reliable face and animal/object tracking on full clips
- Export progress - percentage shown while preparing face blur masks before the final encode
Fixes
- Export vs tracking - cancelling an export no longer stops an in-progress face or object analysis
- Face filter - switching mask mode clears outdated tracking until you re-analyze
- Tracking jobs - prevents overlapping analyze runs; progress overlay stays accurate
[0.8.83] - 2026-07-02
Improvements
- Clearer tracking UI - Face tracking (SMART DETECT, pick, analyze) stays in Face filter; Object / animal tracking moved to Montage with its own 9:16 reframe block
- 9:16 preview - crop applies right after tracking finishes; face detection % badge stays readable in vertical preview
[0.8.82] - 2026-07-01
New
- Track animals & objects (GPU) - in Face filter, Choose subject then Start tracking uses the updated tracking module (pack 1.4.0) for birds, pets, etc.
- Smart reframe in Face filter - Smart reframe 9:16 (follow face or tracked subject on vertical export) is now in the Face filter panel, below SMART DETECT
Improvements
- Subject pick - aim the target in the preview, reposition with another click, then launch tracking; Start over resets tracking and returns the project to 16:9 landscape preview when appropriate
- Progress - separate Object tracking vs Face tracking labels during analysis
Fixes
- Subject pick marker - correct size after Start over (preview zoom no longer stuck from the previous 9:16 crop)
[0.8.81] - 2026-07-01
New
- SMART DETECT - in Face filter, analyze the main face automatically (no click in the preview); manual pick is still available
Improvements
- Face / STT modules - after a local worker update, the app fixes the pack checksum instead of re-downloading the full module every time
Fixes
- SMART DETECT - clearer message when the tracking module is too old
[0.8.80] - 2026-07-01
New
- Smart Reframe 9:16 - turn landscape clips into vertical Shorts/Reels/TikTok exports that follow a tracked face (analyze face first, then enable Smart Reframe in Montage)
Improvements
- Inspector - Clip frame rate shows the probed FPS of the selected video (separate from project FPS)
- Face tracking - progress stays visible through the full analyze step; UI responds as soon as tracking finishes
Fixes
- Face tracking - no more multi-minute freeze after analyze completes
- Smart Reframe export - exported vertical video matches the blue preview crop smoothly (single continuous file)
[0.8.74] - 2026-06-30
New
- Karaoke-grade subtitles - optional WhisperX STT pack (v2) aligns words to speech for tighter karaoke and word-pop highlighting; download the updated speech module from Pro -> Generate subtitles
- Word editor - edit each subtitle word and its start/end time in the inspector after transcription
Improvements
- Animated subtitles - preview sync follows the actual audio/video clock; export matches box fill, shadow, and word-pop scale
- STT pack 2.0.1 - fixes speech-module install on Windows (replaces broken 2.0.0)
Fixes
- Subtitle timing - more accurate word timings when trimming the source clip before transcription
[0.8.73] - 2026-06-29
New
- Animated subtitles - Karaoke and word-pop presets highlight words in sync with speech (uses Whisper word timings)
- Inspector - drag section blocks (Montage, Pro, Text, ...) to reorder panels; order is saved on this PC
Improvements
- Toolbar - customize quick and timeline buttons (Minimal / Standard / Full presets or per-action checkboxes); default shows Export and Split only (zoom always visible); Edit -> Customize toolbar...; layout saved on this PC; all actions stay in menus and shortcuts
- Export speed - on our internal 60 s DJI benchmark (FX max, NVENC): export ~104 s -> ~99 s (~5-6 % faster). Your projects and hardware may differ
- Hardware encoders - Intel QSV and AMD AMF support; Auto tries NVENC -> QSV -> AMF -> software x264
- Multi-GPU NVIDIA - pick which NVIDIA GPU when several are installed
- Export settings - hardware summary (CPU cores, detected encoders) at the top of the dialog
- Face filter - faster skin-mask preparation; more reliable export when face blur combines with overlay fades
- Subtitles on one video track - fewer FFmpeg jobs when captions sit on a single V1 timeline (depends on your edit)
Fixes
- Export progress - correct percentages when multiple segments render in parallel (completed count vs active segment)
[0.8.72] - 2026-06-19
_Thank you to ṪëṄäḲïṄṾḞẌ for the beta feedback._
Improvements
- Inspector - click anywhere on a section title to expand or collapse (Pro, Montage, Text, and other clip panels)
- Pro & Montage - related controls grouped in clear visual cards (keyframes, effects, subtitles, speed, transitions, etc.)
- Timeline - lighter toolbar with grouped actions and shortcut hints; zoom slider (5%-400%) centered in the bar
- Timeline tracks - + / − on the last overlay, audio, and text track to add or remove empty tracks
- Timeline layout - text tracks shown above the video stack (subtitles on top, like pro NLEs)
[0.8.71] - 2026-06-25
_Thank you to ṪëṄäḲïṄṾḞẌ for the beta feedback._
New
- Subtitles - generate captions locally from a clip (optional Whisper pack download); audio never leaves your PC
- Subtitle styles - presets (box, bold, neon, minimal, pop) with live preview and matching export burn-in
- Multi-export - export one project to YouTube, Shorts, TikTok, and Instagram in a single batch
- Podcast assistant - guided wizard: template, import, silence cut, ducking, captions, and batch export (each step skippable)
Improvements
- CUTFORGE PRIVÉ - STT pack download and subtitle generation require the private test build; clear message on the public app
- STT pack 1.0.1 - fixes Windows transcription (includes required whisper.cpp DLLs)
Fixes
- Subtitle sync - captions stay aligned with speech when using Short/Medium/Long length presets (word-level Whisper timing)
[0.8.70] - 2026-06-19
_Thank you to ṪëṄäḲïṄṾḞẌ for the beta feedback._
Improvements
- Timeline - scrubbing the ruler or playhead no longer selects timecode text
- Inspector - type exact numbers on all sliders (audio, image, text, stickers, montage, chroma, effects, face filter, silence trim)
- Montage - rotate clips up to ±360° in preview and export
- Keyframes - clearer labels, click a keyframe in the list or on the timeline to jump the playhead
- FFmpeg - on first launch, CUTFORGE retries an automatic download if FFmpeg is missing; manual gyan.dev setup stays as a fallback (pinned BtbN build updated after upstream removed the previous download)
[0.8.69] - 2026-06-19
Improvements
- Preview - multiple audio tracks (A1, A2, ...) play together; export frame keeps the correct 16/9 ratio when you resize the preview panel
- Import - waveforms appear sooner after media probe (peaks computed at import)
- Performance - lighter undo history, smoother media library during playback, faster timeline lookups while playing
- Face filter - mask cache prepared right after face analysis for quicker export
- Accessibility - keyboard control of menus, timeline scrubbing (arrows, Home/End), and clip selection; English screen-reader label for the inspector
Security
- Downloads - tracking pack and desktop telemetry use private gate tokens
- Stickers & paths - stricter file path checks; reduced IPC exposure on the splash screen
[0.8.68] - 2026-06-19
_Thank you to Likuello for the beta feedback._
Improvements
- Media library - create bins to organize your assets; new imports go into the active bin
- Edit menu - customize keyboard shortcuts (saved on this device)
- Effects - vignette and warm LUT are stronger at the same slider value in preview and export
[0.8.67] - 2026-06-19
_Thank you to Likuello for the beta feedback._
Improvements
- Menu bar - File, Edit, Timeline, View, and Help menus with keyboard shortcuts shown inline; quick access bar for Save, Import, and Export
- Import - drag files from Explorer anywhere on the app window to place them on timeline tracks or in the media library
- Playback - smoother preview audio when switching between video tracks
[0.8.66] - 2026-06-18
Improvements
- Timeline - video tracks follow standard editor order (main track V1 at the bottom, overlays on top)
- Import - drag files from Explorer onto a timeline track to import and place clips; drop elsewhere to add to the media library
- Navigation - arrow keys move the playhead one frame at a time; Shift ±1 second, Ctrl ±5 seconds
[0.8.65] - 2026-06-17
Fixes
- Face filter - update the tracking module to 1.3.2 from Face filters if analyze fails on some videos (FFmpeg compatibility)
[0.8.64] - 2026-06-17
Fixes
- Inspector - Text and Stickers sections are always available in the right panel again (add titles and stickers without selecting a clip first)
[0.8.63] - 2026-06-17
Improvements
- Crash recovery - if a recovered autosave is found at startup, choose Restore or Ignore instead of an automatic reload
- Undo - history uses less memory on large projects
- Accessibility - dismiss toasts with a screen-reader label; resize the preview panel with keyboard arrows
Fixes
- Missing media - relink and scope errors show a clear message instead of failing silently
- Security - stricter validation of file paths used by stickers and media import
[0.8.62] - 2026-06-16
New
- Silence cut - remove internal silences across all unlocked timeline tracks in one action (with ripple)
Improvements
- Face filter - smoother preview while the sequence is playing; score badge updates less distractingly
- Preview - text overlays follow playback more smoothly
- Timeline - ruler ticks adapt on long projects; trimming clips feels more responsive during drag
- Face filter export - progress shows which clip is being processed
- Inspector - opens only the panels you need for the current selection
Fixes
- Face filter - face-tracking model download works without a GitHub account (served via cutforge.app)
- FREE tier - Advanced inspector tools (keyframes, effects, chroma, proxy, SRT) stay available; PRO still removes export watermark only
- FREE tier - watermark position in preview matches the exported file more closely
- License checkout - more reliable activation after purchase
Website
- Windows download - temporarily shown as Coming soon on cutforge.app (same as macOS)
[0.8.61] - 2026-06-14
Improvements
- Error reporting - when you opt in, handled export, import, and project failures are reported so we can fix issues faster
- Editor - timeline, inspector, and export refactored under the hood (no intentional UI change)
Fixes
- Website changelog - release notes on cutforge.app display with correct encoding and punctuation
[0.8.60] - 2026-06-14
Improvements
- Face filter (Advanced skin) - smoother, less flickery blur masks during motion
- Face filters - optional tracking pack 1.3.1 (update from Face filters if you use Advanced skin)
[0.8.59] - 2026-06-14
New
- Face filter - choose Standard or Advanced skin for a tighter skin-only blur mask
- Face filter - switching engine prompts you to re-analyze the clip
Requirements
- Advanced skin requires tracking pack 1.3.0 or newer (install or update from Face filters)
[0.8.58] - 2026-06-13
Improvements
- Face filter - analysis progress shows a real percentage instead of a fixed value
- Face filter - you can cancel while the optional module is downloading; a confirmation message appears when analysis is cancelled
- Face filter - clicking a face in the preview stays accurate when the clip is zoomed or cropped
- Help - note about Windows SmartScreen when `cutforge-tracker.exe` is first run
[0.8.57] - 2026-06-13
Improvements
- Face filter export - more reliable on long videos; cancel responds faster and uses less memory
- Dialogs - improved keyboard and screen reader support (focus stays inside, Escape to close)
Security & privacy
- Additional hardening for optional downloads, usage statistics, and error reports
- PRO license - shorter offline grace (3 days) if your license is revoked while you were online; helps detect invalid system dates
[0.8.55] - 2026-06-12
Fixed
- Face filter - export no longer crashes on installed builds when the face blur filter is enabled
[0.8.54] - 2026-06-12
Improvements
- Face filter - face tracking no longer flashes a black terminal window on Windows during analysis or mask export
- Face filter - progress overlay messages are translated (French, Russian, English)
[0.8.53] - 2026-06-12
Improvements
- Face filter - the Intensity slider (0-100 %) now clearly changes blur strength in both preview and export
- Face filter - preview blur is aligned with the exported video (same mask shape and intensity curve)
- FREE tier - watermark position in the sequence preview matches the exported file
[0.8.51] - 2026-06-12
Improvements
- FREE export - updated watermark with Cutforge.app branding
[0.8.50] - 2026-06-12
Improvements
- Privacy - the optional usage statistics prompt appears after your first successful export, so you can try exporting before deciding
- Changelog - release notes on the website are available in English, French, and Russian (user view vs technical view)
[0.8.38] - 2026-06-11
Fixes
- Face filter - export no longer shows a black background or a drifting white shape; skin blur and warm tint apply correctly to your video
- Face filter - clicking a face in the preview stays accurate when you resize the window or go fullscreen
- Face filter - analysis errors are shown clearly in the panel
Improvements
- Face filter - detection confidence (%) shown after analysis
[0.8.37] - 2026-06-11
Fixes
- Face filter - the optional tracking module downloads correctly from the website again
- Face filter - you can choose skin blur or warm tint and adjust strength before running analysis
[0.8.36] - 2026-06-11
New
- Face filter - click a face in the preview, run analysis, then apply skin blur or a warm tint; effects are included in the exported video
- Optional face tracking module (~17 MB) - downloaded on first use
[0.8.33] - 2026-06-11
Fixes
- Stickers - stickers export correctly again when cached images are used
[0.8.30] - 2026-06-10
New
- On-screen text - add titles and captions on the timeline, with fonts and styles in the inspector
- Stickers - emoji and shapes on clips, visible in preview and export
Improvements
- Export - faster on long projects thanks to segment caching
_For older versions, open the Technical view for the full developer changelog._