o It has been mostly rewriten and now uses OpenGL instead of software rendering, making it much faster on computers with video cards that don't belong to the stone age. (AMZ)
o Normal and SSA mode have been merged into one, all functionality available to both is now used at once. This is the same behavior as Sabbu has. (AMZ)
o Auto-commit will never change to the next line. (AMZ)
o Added an option to, when dragging line start/end markers, snap to keyframes or nearby subtitle start/end markers. This behavior can always be toggled by holding down shift. (AMZ)
o The "Play" shortcut will always play, regardless of whether it was already playing or not. There is a new shortcut for "Stop". (AMZ)
o Styles for line start/end markers have changed. (AMZ)
o Holding down ctrl lets you retime the original line while in Karaoke mode. (AMZ)
o Middle click will now jump the video to the time under the mouse. (AMZ)
o Added DirectSound audio output, to avoid using PortAudio under Win32. (AMZ)
o Switching to a new line in audio mode now centers the line on the grid. (AMZ)
o Flexibible audio cache names now allow you to have more than one copy of Aegisub open with audio loaded to HD cache. (AMZ)
o Added Medusa-style (numpad) global timing shortcuts, that can be enabled with the button under the audio display. This is disabled by default. (AMZ)
o Fixed drawing of inactive lines in audio display to prevent it from being drawn over selection. (AMZ)
o Added options to toggle the display of the audio timeline and of the time over the mouse cursor. (AMZ)
o Added option to disable automatic grabbing of times from selected lines on the audio display. Also, it will never pick the times from 0:00:00.00 -> 0:00:00.00 lines. (AMZ)
o The calculated FFT data are now cached, so things should be faster.
- The maximum size of the cache can be configured. Default is unlimited, which provides the best performance assuming enough memory is available.
o The quality of the spectrum can be easier configured, better quality requires more CPU and memory.
o Actual signal power is now more accurately represented.
o The palette is changed.
o The selection is no longer shown by ugly reverse colour but with a different palette instead.
o Use vertical zoom slider to amplify/dampen displayed signal strength. (Useful for better visualisation of quiet sections, or easier picking out the dominating frequencies in noisy sections.)
o Automatically select the style of the current line when Style Manager is opened. (jfs)
o Style storages with invalid characters are now caught, and the invalid characters replaced with safe ones. Previously such storages seemed to be created correctly but were never stored to disk. (jfs)
o Added long-missing Edit buttons. (jfs)
o Renaming a style will now ask you if you want to rename all instances of it on the script. (AMZ)
o Style name collisions (when moving from storage, creating or renaming) will now prompt the user before performing the action. (AMZ)
o Added an "Import from script..." button to import styles directly from other subtitle files. (AMZ)
o Added buttons to move styles up, down, to top or to bottom, on both storage and current script. (AMZ)
o Added sorting functionality for both storage and current script. (AMZ)
o Prevent creating multiple storages with the same names. (AMZ)
- Style editor changes:
o Prevent creating multiple styles with the same names. (AMZ)
o Instead of falling back to your local charset, Aegisub will now use the "universalchardet" library (the one used by the Mozilla project) to autodetect the character set of non-Unicode files. (AMZ)
- Added "Video Formats" to "Open Audio"'s file selector, so you can open audio directly from video files (regardless of whether they're loaded or not). This has always worked, but it was never explicit. (AMZ)
- Treat comments inside {}'s as plain text, not as overrides; Also, don't assume override blocks start with a backslash, even if they probably should (Dansolo)
o Added some debug-calls (which are disabled by default; aegisub.output_debug is replaced with a do-nothing function (you can change this in karaskel.lua)
o The regular aegisub.output_debug is aliased to aegisub.output_warning, in order to always be able to show real warnings, in case something goes wrong
o Fixed bug, triggered when a line had a style not defined in the subs. A warning is now shown instead.
- Fixed bug where \t tags (among others?) would lose their parameters when being rebuilt (happens with Framerate Transport export filter, Automation and during the karaoke timing improvements listed above, among other places) (jfs)
- 1,1+2 recombining lines where 1 is a substring of 2 no longer causes incorrect behavior. 1+2,2 similarly fixed. This fix also gives more sanity-checking in the recombining (jfs)
o The grid can optionally no longer receive focus (this behavior is disabled by default, change "Grid Allow Focus" to 0 in config.dat to prevent it from getting focus).
- Fixed some internal workings, which should make Aegisub use considerably less RAM (especially for large Karaoke files, 8-10x less RAM usage was measured with a 9 MB file). (AMZ)
- The font override button in the subtitles edit box can now modify font size, italics, bold and underline as well as font face (no strikeout, as wx does not provide an interface to access that data). (AMZ)
- Now requires Avisynth 2.5.6a or later, added an option ("allow ancient avisynth") to override it to config.dat. (Myrsloik)
- Added reading of keyframe and timecode data from Matroska files (which are still not recommended, due to their dependency on DirectShowSource). (AMZ)
- Added a bias slider to the "make times continuous" function in timing post-processor, which sets the point between the subs where the two lines will meet. (AMZ)
- The manual has been extensively re-written and extended to include documentation about a lot of new (and some not so new) features. It's reasonably complete... for the moment. (TheFluff, Motoko-chan, jfs)
- Aegisub now supports file attachments, which are stored decoded (to save memory) and are not part of the undo stack (for the same reason). Previously, they were simply left ignored in the file as unknown lines. (AMZ)
- Implemented Timing Post-Processor dialog to automatically add lead-in/lead-out, scene timing and adjascence adjustment for sharp-timed subs (AMZ)
- Aegisub is now translateable, with proper localization files. (AMZ) Available now are:
o Brazilian Portuguese
o French (thanks to TechNiko and Kurasu)
o Russian (thanks to ]thrash_sensei[)
o German (thanks to equinox)
- Fixed saving of hotkeys (AMZ)
- Fixed behavior of Insert key in Translation Assistant (AMZ)
- Ctrl+Enter now previews in translation assistant (AMZ)
- Fixed "make times continuous" bug (AMZ)
- Checks if there's enough free space when using disk cache and asks if disk cache should be used when there isn't enough ram available (Myrsloik)
- Always enforces the set memory limit for avisynth even if no video is opened (Myrsloik)
- Fixed crashes when not enough memory could be allocated for ram cache (Myrsloik)
- Properties dialog will no longer flag file as modified if nothing changed (AMZ)
- The config.dat file is now fully documented again, for the first time since beta 1.00... let's see how long it lasts. (TheFluff)
- Selected line is now properly enabled on audio display when audio is loaded (AMZ)
- Cosmetic improvements to toolbar (AMZ)
- Improved command line and drag-and-drop opening of files (extended to allow video and audio at the same time as well) (AMZ)
- Fixed audio glitches when using disk cache and scrolling the audio display while playing (Myrsloik)
- Improved the ram cache allocation so it should work for really large files if enough ram is available (Myrsloik)
- Fixed exporting of SSA styles (AMZ)
- Automation: The precalculation step in karaskel.lua is now in a function by itself, precalc_syllable_data (jfs)
- Automation: New function added to utils.lua, string.headtail(s): treats s as a list of words separated by spaces, returns two strings, the first word (the head) and the rest of the string (the tail), like car/cdr (jfs)
- Fixed behavior of dialog that asks you to change script resolution based on video resolution (AMZ)
- Video and audio displays can now be hidden via the View menu (AMZ)
- Translation assistant now shows proper text color (AMZ)
- Translation assistant now shows total line count (AMZ)
- Hotkeys for translation and styling assistants are now customizeable (AMZ)
- Default character encoding for saving can now be set via the evil "save charset" in config.dat. (AMZ) A few examples:
o "save charset=UTF-8" - Default setting. RECOMMENDED.
o "save charset=UTF-16" - Saves as UTF-16 instead of UTF-8. This solves issues if you must deal with a program that can't support UTF-8, such as ancient textsub or wordpad.
o "save charset=local" - Kills unicode support in saving subtitles, and saves as local charset. THIS OPTION IS NOT RECOMMENDED! It will, however, make scripts compatible with Medusa and other non-unicode programs.
o "save charset=Shift_JIS" - Same as above, but saves as Shift_JIS (japanese script). NOT RECOMMENDED!
o "save charset=ISO-8859-1" - Same as above, but saves as ISO-8859-1 (western script). NOT RECOMMENDED!
Remember: You can always save to another charset via the File->Export option. This is much safer!
- Setting "Use nonstandard milisecond times" in config.dat makes Aegisub use miliseconds instead of centiseconds for ASS times. DO NOT use it unless you know EXACTLY why you should! NORMAL VSFILTER CAN'T HANDLE THIS. (AMZ)
- Resolution resampler now affects drawing commands as well (AMZ)
- Added Undo/Cut/Copy/Paste/Select All to subtitles edit box context menu (AMZ)
- Added a "OK" button to style editor dialog, and now "Apply" applies changes without closing it (AMZ)
- Paths to video/audio/timecodes files are now stored relative to subtitles (instead of as absolute paths) in [Script Info] (AMZ)
- Audio display now shows previous line in grey (this feature can be disabled (0) or extended to every line (2) with "Audio Inactive Lines Display Mode" on config.dat) (AMZ)
- Audio display height can now be adjusted by drawing the sash under it (AMZ)
- Holding shift while clicking on video seek bar now snaps to keyframes (AMZ)
- Fixed positioning of subs on video while shift is held down (AMZ)
- Aegisub is now Open Source under the 3-clause BSD license
- Bugfix: Crash in Styles Manager related to copying styles (jfs)
- Automation: Fixed progress-reporting in karaskel.lua (jfs)
- Installer now shows the icon in add/remove programs (Myrsloik)
- Installer defaults to the directory of the previous install (Myrsloik)
- Fixed behavior of "Don't create start menu shortcuts" in installer (Myrsloik)
- Bugfix: The actor dropdown list didn't always work, it's hopefully fixed now (jfs)
- Automation: Yet another attempt at fixing the slowdown bug (jfs)
- Added option "Audio Draw Boundary Lines", if disabled, no vertical lines (except for current selection) are drawn on the audio waveform. Karaoke syllable boundaries are always drawn (jfs)
- Added option "Audio Draw Selection Background", if disabled, the background and waveform color isn't changed for the selection in waveform display mode (jfs)
- SSA Mode: Committing changes now switches to next line and sets the selection to the part following the old selection (jfs)
- SSA Mode: Selecting another line in the grid no longer changes the audio selection (jfs)
- SSA Mode: "Play current line" sets the selection to the current line (jfs)
- SSA Mode: Committing works even when no changes were made (just go to next row) and negative timings are accepted (jfs)
- When loading video, if the video and subs resolution don't match, ask to change the subs resolution (jfs)
- New config option to control the above, "Video Check Script Res" (jfs), it can take the following values
(0) Never change subs resolution to match video
(1) Ask to change subs resolution on mismatch
(2) Always change subs resolution on mismatch (never ask)
- New subs have default resolution 640x480 (to match most common videos) (jfs)
- New config option: "Audio Line boundaries Thickness" controls the thickness of the lines marking the start/end of the current selection (jfs)
- Bugfix: Loading audio files of different sample rates without closing the program resulted in audio playing too slow/fast (jfs)
- New color picker dialog (jfs)
o Graphically pick colors from different colorspaces (RGB, PC-YUV, HSL, HSV)
o Manually enter color values in those colorspaces
o ASS and HTML hex input
o 24 recently-used colors (automatically managed)
o "Screen eyedropper" to make a screengrab and pick a color from that
- The following windows remember their last used position (not across program restarts, though) (jfs)
o Styles Manager
o Styling Assistant
o Translation Assistant
- Most dialog windows now center on the window behind them (jfs)
- The start and end markers for audio selection can now have different colors, use the "Audio Line boundary start" and "Audio Line boundary end" options to change them, the old "Audio Line boundaries" option is no longer used (jfs)
- Blank script resolution will now be interpreted as 384x288 (VSFilter's default) if script res is blank (AMZ)
- Audio playing of selection will now play until new end time when it's changed during playback (AMZ)
- Aegisub will now display a proper error message if it attempts to load an outdated avisynth.dll, instead of just crashing (Myrsloik)
- Automation: Changed the result array from process_lines to use one-based indexes, instead of zero-based indexes, to follow Lua standards. Zero-based indexes still work. The input arrays are still zero-based. (jfs)
- Automation: Added new factorybrew script, line-per-syllable.lua. Allows you to make line-per-syllable karaoke effects by using variable substitution and calculation. Built off karaskel-adv.lua (jfs)
- Automation: Updated simple-k-replacer.lua to also allow for calculations, and it is based on karaskel.lua now (jfs)
- Fixed count of matches in Replace All dialog (AMZ)
- Fixed extra line breaks bug in importing from txt (AMZ)
- The automation samples are now always installed (Myrsloik)
- Current audio selection is now shown in inverse colors when using spectrum analyser mode. Can be disabled with "Audio Spectrum invert selection" option (jfs)
- Bugfix: Clean Script Info export filter errorneously removed the PlayResY line (jfs)
- Fixed critical bug in installer (thank movax for that) that could corrupt Windows's Uninstaller information (Myrsloik)
- Installer will no longer always ask you if you want to keep the styles dir (Myrsloik)
- Installer now saves the avisynth installer/source to the desktop, so you can proceed if downloading fails (Myrsloik)
- Fixed size of last character in editbox with syntax highlight off (AMZ)
- Fixed audio stopping issue (e.g. forcing you to press 'S' or 'Space' twice) (AMZ)
- Five splash screens are now available. Set "Splash number" to 0 in config.dat to get a random one. Defaults to Akari. (AMZ)
(1) Misuzu from Air TV
(2) Lucy from Elfen Lied
(3) Kozue from Mahoraba
(4) Lamhirh (aka Lafiel) from Crest of the Stars
(5) Akari from Aria
- Added function (Ctrl+6) to shift selection so first line goes to current frame (AMZ)
- Automation now loads scripts both with and without UTF-8 BOM. They are still required to be UTF-8 encoded. (jfs)
- Fixed crash-bug in Automation related to internal item-count in tables used as arrays (jfs)
- Automation now displays a proper progress-bar during processing
- New function in Automation: aegisub.set_status(text), displays a status message below the progress bar during processing. aegisub.output_debug(text) also does that, but additionally adds the message to the debug log. Debug log is only shown if output_debug was actually called. (jfs)
- New functions in Automation: aegisub.frame_from_ms(ms) and aegisub.ms_from_frame(frame) for doing per-frame effects (jfs)
- Modified some of the example scripts for Automation. Scripts based on karaskel.lua might break now! (jfs)
- Added some UTF-8 handling functions to utils.lua (jfs)