Privapixel

What was fixed:

🎯 Speaker tracking on stage
Framing no longer stays locked in one position and now follows gradual movement, keeping the active speaker centered in the vertical clip.

🖼️ Static projection protection
Improved logic that distinguishes the live speaker from faces printed on slides or projections, preventing jumps to the screen without blocking natural movement.

📹 Low-resolution video detection
Better frame analysis and tracking when the face disappears for a few moments, with position estimation to avoid losing framing.

⚙️ More robust export
If an individual clip fails during rendering, the others continue to be generated normally, with automatic fallback when needed.

📁 Output folder dialog
When choosing the output folder, the dialog opens centered relative to the main window and no longer shifts the form when canceled.

What was implemented:

🎯 Camera that follows the speaker
In vertical 9:16 clips, framing is no longer a fixed center crop — it automatically follows whoever is speaking at each moment.

🧠 Local AI face detection
The system analyzes video frames, locates faces, tracks each participant in the scene, and organizes speakers by position (left, right, etc.).

🗣️ Active speaker identification
Combines Whisper transcription, mouth movement, and speech segments to decide who is speaking — with smooth switching when the conversation changes participants.

📱 Ready for Reels, TikTok, and Shorts
The clip starts framed on the right speaker, avoids focusing on the table or the wrong participant, and delivers a professional-looking result for social media.

💻 100% local
All visual analysis runs on your PC — your video does not need to be uploaded to external servers for speaker framing.

What was implemented:

🎬 Long video → short clips
Upload podcasts, lessons, live streams, interviews, or any long video — Privapixel analyzes the content and suggests the best cuts for social media.

🔥 Viral potential score
Each clip gets a viral potential score and appears ranked from most to least promising — you choose which ones to export.

📱 Export formats
Export in Vertical 9:16 (Reels, TikTok, Shorts), Horizontal 16:9, or keep the original video format.

📝 Synced subtitles
Generate burned-in subtitles on exported clips, with real word-level timestamps via local Whisper — ready to publish without manual editing.

✨ Word-by-word highlight
Option to highlight the spoken word in the caption in real time (karaoke style), one phrase at a time — ideal for holding attention in the first seconds.

⚙️ Full control
Choose language (Portuguese, Auto, English, Spanish), clip duration (short, medium, or long), maximum count (5 to 12 clips), and enable turbo mode for long videos (~5× faster).

💻 100% local
All analysis, transcription, and export runs on your PC — no uploading videos to external servers.

What was implemented:

🔗 Download by link
Paste the video link — Privapixel analyzes it, adds it to the queue, and downloads it straight to your PC.

💻 Local processing
Everything runs on your computer — without relying on ad-filled websites and without uploading files to external servers.

🎬 Quality and format
Choose the desired quality (up to 4K) and export as MP4, MKV, or WebM, or extract audio only as MP3, M4A, or OGG.

⚡ Smart mode
When you paste a link, the video is added to the queue automatically — just paste and download.

📊 Queue with progress
Track each download status in real time, with thumbnail, quality, and file size.

What was implemented:

Transcribe Audio

🎙️ Everyday audio
Support for MP3, OGG, OPUS, M4A, AAC, WAV, AMR, FLAC, and WMA — including audio exported from WhatsApp and Telegram.

🤖 Local Whisper
Speech recognition with AI on your PC; the model downloads once (~142 MB) and is ready for offline use.

🌍 Languages
Portuguese, automatic detection, English, and Spanish — choose before transcribing.

📄 Text or subtitles
Export .txt for reading, archiving, and quick copy, or .SRT with timestamps for use as captions.

👁️ Transcription preview
View transcribed text on screen before and after saving to the output folder.

Transcribe Video

🎬 Popular video formats
MP4, MKV, AVI, MOV, WEBM, WMV, FLV, and M4V — the app extracts audio and transcribes locally.

📝 Ready .SRT subtitles
Generate standard subtitle files with start and end time for each line — import into CapCut, Premiere, DaVinci Resolve, and other editors without typing manually.

📱 Built for creators
Speed up content production for online courses, training, and internal materials with automatic captions.

📋 Batch processing
Transcribe multiple audio or video files at once using the same queue as other tools.

🔒 100% local
Your audio and video are not uploaded to cloud servers — total privacy on your PC.

What was implemented:

Ready for Sale Photo

🛒 Marketplace-ready
Ideal for Mercado Livre, Shopee, catalogs, and online stores — clean photo, product in focus, and professional look.

🤖 Local AI (U²-Net)
Background removal on your PC, with no images uploaded to external servers.

🎨 White or premium background
Choose neutral white for marketplaces or a premium studio-style background for a more sophisticated look.

🌓 Soft shadow
Optional gentle shadow under the product for depth and a studio-photo feel.

📐 Preset sizes
1000×1000 (Mercado Livre), 1200×1200 (recommended), 1600×1600 (high quality), and 2000×2000 (premium).

👁️ Real-time preview
Compare the original photo with the estimated result before generating.

Offer Art

📣 Complete promotional art
Product photo highlighted at the top and a red band with name, prices, highlight, and call-to-action button at the bottom.

💰 Old and new price
Shows the previous price struck through and the promotional price in highlight — ideal for sales posts.

📱 WhatsApp CTA
White rounded button with green WhatsApp text; optional official WhatsApp icon before the word WhatsApp.

🏷️ Highlight and logo
Customizable highlight line (e.g. Limited-time offer) and optional store logo in the corner of the art.

👁️ Instant preview
See the estimated art while filling in the fields — perfect for adjusting copy before batch generation.

📋 Batch processing
Generate art for dozens of products at once using the same queue as other tools.

What was implemented:

Enhance image

✨ Combinable adjustments
Select one or more options: sharpness, noise reduction, contrast, brightness, color correction, and brightening dark photos.

👁️ Real-time preview
Compare the original and estimated result side by side when you select each image in the queue.

📋 Batch processing
Apply the same enhancements to dozens or hundreds of photos with one click — ideal for phone photos, catalogs, and social media.

🔒 100% local
All image processing runs on your PC with no upload to external servers.

What was implemented:

Rounded corners

🔲 Corner radius
Slider from 0% to 50% to set rounding, with pixel value shown alongside and live preview.

🖼️ Transparent background (PNG)
Checkbox to export with transparency in the corners; preview uses a checkerboard when transparent areas are present.

📋 Batch processing
Use the same queue as other tools: apply rounded corners to dozens or hundreds of images at once.

Convert formats

🔄 Smart conversion
Source format is detected automatically from the selected image; choose the target format and process in batch.

⚙️ Quality and metadata
Quality control for compressed formats (e.g. JPEG and WebP) and option to strip EXIF metadata on export.

📁 Broad compatibility
Support for many popular formats — all processed locally on your PC.