sudacode/SubMiner
1
0
Fork 0
mirror of https://github.com/ksyasuda/SubMiner.git synced 2026-08-27 12:15:25 -07:00
Code Issues Packages Projects Releases Wiki Activity

fix(subtitles): recover positioned ASS word spacing and drop control debris (#217)

Browse Source
This commit is contained in:
sudacode
2026-08-25 02:22:21 -07:00
committed by GitHub
parent ed7d3f4c3d
commit e816b3b371
26 changed files with 2934 additions and 105 deletions
Download Patch File Download Diff File Expand all files Collapse all files
changes/ass-canonical-animation-lines.md
+7 -2
View File
@@ -1,5 +1,10 @@
type: fixed
area: subtitles
- Typeset ASS karaoke and animated signs no longer flood the primary overlay, subtitle sidebar, immersion history, or sentence mining with repeated glyph fragments or full-line color phases. Matching timed comments and full-line boundary events recover the complete authored line without merging ordinary repeated dialogue or separately positioned signs, and dialogue spoken while a song's animation is on screen is kept intact instead of being replaced by the lyric. Entrance and exit frames that run past the authored line timing still resolve to the clean line during lyric transitions, and dialogue spoken while a song's animation is on screen enters immersion and subtitle history without the fragment lines beside it.
- The secondary subtitle overlay drops layered duplicate lines from animated tracks, so a short stack of repeated words collapses to its distinct lines even when the full karaoke heuristic does not apply.
- Typeset ASS karaoke and animated signs no longer flood the primary overlay, subtitle sidebar, immersion history, or sentence mining with glyph fragments and per-frame color phases. The authored line is reconstructed once and shown from its first sung syllable until its hold ends, so a lyric's invisible lead-in and fade tail no longer surface it beside the line actually being sung.
- Reconstructed lines keep their authored word spacing, including gaps encoded only by fragment positions or hidden behind wide syllable chunks, so translations stop running words together.
- Ordinary repeated dialogue, separately positioned signs, wrapped lyric rows, and multi-row CC-style dialogue blocks are still published, and dialogue spoken while a song's animation is on screen stays intact instead of being replaced by the lyric.
- Decorative layers stay out of the published text: highlight sweeps, drop-shadow and glow copies, symbol-font glyph decoration, particle-build glyph swarms, sign walls, near-invisible texture strings, and hidden zero-scaled or zero-clipped text.
- On the live overlay, a finished lyric whose exit ghosts still linger in mpv's text keeps explaining those fragments instead of reappearing beside the line that has already taken over.
- Event-heavy karaoke files that stalled subtitle loading for several seconds now parse in well under a second.
- The secondary subtitle overlay collapses layered duplicate lines from animated tracks even when the full karaoke heuristic does not apply.