Skip to content

FreeVideo 0.4.32 - #643

Merged
KBRASK merged 11 commits into
mainfrom
release/0.4.32-public
Nov 4, 2026
Merged

KBRASK merged 11 commits into
mainfrom
release/0.4.32-public

Conversation

@KBRASK

@KBRASK KBRASK commented Nov 4, 2026

Copy link
Copy Markdown
Collaborator

0.4.32: A picture can be generated for a span of imported sound, and the top bar has a permanent Tutorial entry.

  • Sound-led video: drag a span over imported sound in the Edit view, describe the picture, and generate a video for that span while the original sound is kept.
  • Tutorial: the top bar has a permanent Tutorial entry. Three new feature examples cover making a first video, characters and the Edit view. These and some tutorial chapters have "Take me there", which highlights the matching control.
  • Audio descriptions: when a description asks to leave out music or other sounds, that request is removed before generation and, where needed, a sentence is added saying the sound comes from the action and the surroundings, so unwanted background music appears less often. Descriptions that ask for music or silence, and quoted lines, are unchanged.
  • Two GPUs: a machine with two GPUs now decides whether tasks can run together from the memory each task actually uses. On Linux, two 16 GB cards with 64 GB of RAM can run two tasks at once. Single-GPU machines are unaffected.
  • Editor: deleting an edit no longer keeps failing after a brief loss of contact with the background service.
  • Editor import: a video that is cut off or damaged part-way now keeps the part that can be read instead of failing the whole import.
  • Reference audio: generating with reference audio works again on Mac, and no longer fails on other systems when the model has to be loaded first.

KBRASK added 11 commits November 4, 2026 04:29
…ng and two-GPU memory sizing

- Tutorial: a permanent entry in the top bar, Take me there for each guided card, and three new cards
  (first video, characters, edit view).
- Editor: generate a picture that follows imported sound over a range that has no picture.
- Prompt: requests to leave out music or other sounds are worded as the sounds the scene has; requests for
  music, spoken lines and quoted speech stay unchanged.
- Two GPUs: the host-memory envelope is sized from the measured use of each task.
…acement, per-line reference...

Add the dubbing engine: CosyVoice3 pack, worker, exact-length line placement, per-line reference and failure, voice span and level reporting
…es again in the original...

Add voice change: recognise the lines of a clip and say the edited ones again in the original voice, placed at the original timing; editor panel with original/new switch, failure note and report button
…eeps finished lines of a...

Voice change: the dubbing service takes trusted reference files and keeps finished lines of a stopped run, as the voice job passes them
…s the service, not after...

Voice change: the job notes its own dubbing folder while it still owns the service, not after the wait
… renewal from blocking the...

Report which step of deleting an edit failed, and keep a failed lease renewal from blocking the delete
…utput line count on the...

Launcher: a failed setup inspection names its action, exit code and output line count on the first line

When the inspection child prints no plan, the error now starts with one line such as "Setup inspection failed (plan, exit 1, 0 output lines)". The runner records how many lines the child printed and the class of the exception that stopped the run; a timeout is named, and a Windows crash code also appears in hex. The line holds only integers, fixed words and a class name, so a shortened problem report keeps it whole.
A video cut off part-way or damaged in the middle failed the whole import and showed as unreadable. The importer now skips packets the decoder rejects, stops at damage in the container, and keeps the frames read so far; a file with nothing readable still fails as before. The class of a read failure (never its message or path) is kept in the import record and logged, and the Report a problem button beside an unreadable import sends it as the Kind line.
…ader again

The 0.4.31 latent cache moved the video encoder import into the video branch, leaving the audio branch without its loader: reference audio encoded outside a resident engine, which is every Mac request, raised NameError. A test now checks that no name in media_encoding.py is undefined.
@KBRASK
KBRASK merged commit 7ac5f34 into main Nov 4, 2026
5 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant