What’s new#

Changes in each Wunjo Make release, newest first. Wunjo Make is built on Kdenlive; what changed in the parts of the editor the two share is listed in Kdenlive release history.

Wunjo 3.0.1#

  • Fixed macOS launch failure caused by a std::pmr symbol from macOS 14 in a framework built for 13.3

  • Shipped the application icon theme inside the macOS bundle and told Qt where to find it

  • Drawn the interface with the Breeze widget style on macOS instead of the native one

  • Named the macOS bundle and its menu entry Wunjo Make

  • Found the bundled Python interpreter, so plugins work on macOS

  • Loaded the assistant’s model on Metal without the flags meant for discrete cards

  • Let agents outside the editor find its scripting socket on macOS

Wunjo 3.0.0#

  • Rebuilt as a desktop nonlinear editor on the MLT framework, replacing the web interface

  • Multi-track timeline with transitions, effects, keyframes and time remapping

  • Colour correction, audio post-production, subtitles and titling

  • Rendering to practically any format through FFmpeg

  • Built-in face detection for tracking or hiding a face on a clip

  • Plugin platform: installable Artificial Intelligence plugins in isolated Python environments

  • Plugins for speech recognition, voice cloning, voice separation and content cutting

  • Chat assistant that drives the editor through its scripting interface

  • Update notifications for installations made from a file

Wunjo 2.0.8#

  • Smart Crop a feature that automatically adapts any video to any aspect ratio

  • Loading and processing times are about twice as quick

  • Word-based editing with Smart Cut

  • Memory optimization and fix bugs

Wunjo 2.0.7#

  • History day tasks

  • Automatic video editing based on visual content and speech (e.g. for podcasts or long videos)

  • New video generator methods (only for flash-attn version)

  • More generators parameters

  • Metadata

Wunjo 2.0.6#

  • Identity-preserving generation

  • Multi-language interface: English, Russian, Spanish, Chinese, Korean

  • Multi-GPU support

  • API for developers

  • Zoom for images

  • Fast start

Wunjo 2.0.5#

  • Quick tips

  • Smart search

  • Revamped interface

  • Unified login system

  • Bug fixes

  • Translation tooltip for multilanguage

Wunjo 2.0.4#

  • Speed boost 1.5x

  • Manual and automatic portrait animation

  • Retarget face

  • Enhanced video generation

  • Transformation in portable version

  • Anonymous login

  • Clearer model settings

  • Face detection fix

  • Faster small model loading

  • Unlimited file uploads

  • Automatic face interface detection

  • Progress bar about status

Wunjo 2.0.3#

  • Added control restyling (This work with 8 GB VRAM)

  • Added generation text-to-video, image-to-video, text-to-image, outpaint and inpaint image

  • Improved quality of restyling

Wunjo 2.0.2#

  • Added functionality to clone voice in any language from text and audio

  • Improved audio separator to remove noise before clone voice

  • Enhanced processing speed

Wunjo 2.0.1#

  • Fix bugs

Wunjo 2.0.0#

  • New generation of Wunjo v2

  • Big update features https://youtu.be/-MGow5zCZUQ

  • Completely redesigned interface for desktop and mobile devices

  • Enhanced performance with simultaneous task handling, stability, and faster processing

  • Improved face swap functionality now supports multiple face swap at different timings via API

  • Introducing generator of non-existent people face

  • Deepfake analyzer added to discern authentic content from manipulated media

  • Enhanced lip sync for smoother animations and elimination of chin defects

  • Real-time mask selection for seamless editing experience

  • Added select area to remove text

  • Now you can create chroma key video

  • Custom file manager introduced for efficient file management within the application

  • Improved interaction with neural network models, including seamless downloading from the internet

  • Easy switch from client to server application and back

  • Multi-user support with customizable limits on simultaneous tasks and memory usage

  • Internal tips provided for optimizing content quality

  • Smart generator of prompts for video and object restyling

Wunjo 1.6.2#

  • Custom browser for WebGUI to solve trouble with opening additional tabs

  • Added select browser to run

  • Added check between offline and online mode

  • Added a message for the user about missing models and how to download manually

Wunjo 1.6.1#

  • Resolved issues with the enhancer and upgraded its capabilities for videos and drawings

  • Refined vocoder performance for voice cloning

  • Introduced variable speech cloning speeds

  • Added models for isolating background sound in deepfakes and purifying voice tracks

  • Incorporated a feature to extract background noise from media

  • Enhanced the encoder for more accurate voice cloning

  • Improved overall voice cloning fidelity

  • Implemented multilingual speech synthesis from a single text input

  • Added a new interactive tool for user-defined mask creation that tracks and moves with segmented objects

  • Launched a music generation feature

Wunjo 1.6.0#

  • Advanced the object removal tool for images and videos

  • Enhanced video editing capabilities

  • Implemented automatic segmentation mask saving

  • Introduced Video2Video and InpaintVideoMask2Video with ControlNet via text prompts

  • Optimized memory usage for face swapping and object removal in lengthy videos

Wunjo 1.5.5#

  • Enabled voice cloning in Russian

  • Optimized facial animation and improved error handling for GIF processing

  • Improved log display within the console

  • Added a disk space indicator for output from speech synthesis and deepfakes

Wunjo 1.5.4#

  • Corrected the Russian voice model training from checkpoint issue

  • Introduced a similarity coefficient parameter for Face Swapping and Mouth Animation

  • Enhanced English voice cloning for extended phrases

Wunjo 1.5.3#

  • Automated ffmpeg installation for Windows

Wunjo 1.5.2#

  • Resolved the model auto-loading issue on Windows

  • Accelerated application startup

  • Enhanced performance of speech synthesis and deepfakes on CPU

  • Fixed GIF processing errors in face replacement and retouching

  • Corrected the frame-audio merging issue

Wunjo 1.5.1#

  • Added voice translation features with encoder and vocoder support for English and Russian

  • Implemented voice cloning synthesis from text

  • Established a centralized hub for voice cloning language models

  • Introduced face and background enchantment for photos and videos without deepfake technology

  • Released a comprehensive version including all extensions

  • Enhanced real-time progress indication and translation

  • Integrated a debug module with Python console

  • Transitioned to native language translation for real-time updates

  • Improved GPU availability notifications for users

  • Launched a deepfake emotion module for experimental research

  • Updated the application guide

  • Implemented AI-powered retouching of video frames

  • Enhanced Chinese language support for voice cloning

  • Addressed Windows-specific bugs in Wunjo AI

Wunjo 1.4.0#

  • Introduced deepfake synchronization with source video for lip movements

  • Implemented video segment selection for deepfake based on audio track length

  • Enabled customization of .wunjo cache directory

  • Added multi-language support for application interface

  • Trained and integrated TTS models for English

  • Implemented model language identification in the user interface

  • Enabled cross-language speech synthesis for Russian and English TTS models

  • Included update notifications within the application

  • Integrated training videos and wiki page links