| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
| |
The 400x300 resolution used previously was triggering a sticky
situation with XO-4 graphics (http://dev.laptop.org/ticket/12693)
Tweak the resolution slightly to work around this problem, and land
on a much more standard resolution at the same time.
|
|
|
|
|
|
|
|
| |
We don't need save the preview in the metadata encoded,
doing it makes more difficult the interoperability with other
activities using it, and takes more space in the disk.
Signed-off-by: Gonzalo Odiard <gonzalo@laptop.org>
|
|
|
|
|
|
|
|
|
|
|
| |
This never worked right: we always managed to block, but the unblock
never happened.
Previously this didn't hurt anything, but on F17 it causes problems: since
the pad is blocked, we never take the photo.
Remove pad blocking to work around this. Seems to be working fine
with live-pipeline modification.
|
|
|
|
|
|
|
|
|
|
|
|
| |
When there are more than one parameter with the same type
gettext need named parameters to enable the translators to change
the order if needed.
Pootle server complain about this format strings in Record activity,
and you can see the same error doing "./setup.py genpot"
v2: Fixed typo
Signed-off-by: Gonzalo Odiard <gonzalo@laptop.org>
|
|
|
|
|
|
| |
The video feed is not displayed on screen while saving audio, so we can
stop the feed. This means that alt-tabbing away during an audio recording
releases the video feed much quicker.
|
|
|
|
|
|
| |
Introduce the relevant code catches to allow audio recording
when no camera is present. Still has some rough edges but is
now usable at least.
|
|
|
|
|
| |
Returning from video playback would continue using X instead of Xv for the
live camera feed.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a rework of the UI that uses a more standard GTK+ principles
than previously. There is still a small amount of black magic, kept
away inside mediaview.py.
The UI/model separation was also refined in places, and a lot of code
was simplified.
Overall the functionality remains identical, and I tried to keep the UI
the same as before (but it is not pixel perfect).
As this was quite big there may be some bugs to shake out.
|
|
|
|
|
| |
Avoids a segfault which is probably due to a gstreamer bug.
http://dev.laptop.org/ticket/10183
|
|
|
|
|
|
|
|
|
|
| |
Several seconds of silence were being prepended to the recording.
This looks like a gstreamer bug.
Work around it by stopping the whole pipeline while adjusting it.
http://bugs.sugarlabs.org/ticket/2040
http://dev.laptop.org/ticket/10175
|
|
|
|
|
|
|
|
| |
Recording in this mode is too taxing for the XO-1 and even the XO-1.5.
The A/V sync is lost.
Without significant changes, recording at this resolution/quality is
unlikely to be successful on anything except a high end system.
|
|
|
|
|
|
|
|
| |
On XO-1, we quickly meet the limits of the default queue limits.
Increase them both and add debugging to detect these conditions
in future.
This fixes A/V sync on XO-1.
|
|
|
|
|
|
|
| |
XO-1 capture was broken because the camera driver doesn't support
framerate changes right now.
Add some code to check if this capability is available before using it.
|
|
|
|
|
| |
This is certainly a good idea and was only removed earlier based on
a misunderstanding
|
|
|
|
|
|
|
|
|
| |
This should reduce resource usage a little (it'll display frames when is
most convenient rather than worrying about the clock) and might reduce
weirdness associated with trying to synchronize the video preview
with the audio.
Suggested and tested by Jon Nettleton
|
|
|
|
|
| |
Speeds up saving, theoretically improves quality, and maybe even helps
synchronization?
|
|
|
|
| |
High/best quality thumbnails weren't scaled properly
|
|
|
|
| |
We always record video with audio.
|
|
|
|
|
|
|
|
| |
OGG doesn't store timestamps of audio or video frames, so it's very
important to use a perfectly consistent stream.
Add the audiorate element and adjust the justification for the
videorate.
|
|
|
|
| |
Improves A/V sync, for reasons not entirely known.
|
|
|
|
| |
This is needed for good A/V sync (gstreamer bug?)
|
|
|
|
|
|
|
| |
But we also maintain the gstreamer-level videorate framerate filter.
This is because the XO-1.5 camera delivers a lot more frames than you
ask for, and this throws off gstreamer's timings.
http://dev.laptop.org/ticket/10137
|
|
|
|
|
|
|
|
|
|
| |
Remove queue element in videobin, this makes recorded videos smoother
Prefer fresh frames for A/V display
Add a buffer in audiobin to avoid overrun/underrun problems when
starting recording, avoiding some kind of bug where A/V sync is bad for
the whole video.
|
|
|
|
| |
Like it did before the rewrite
|
|
|
|
|
|
|
|
| |
Restore the v60 rewrite to use a single pipeline, tweaking at runtime
when necessary.
This improves responsivity of the application, and results in the
activity behaving reliably again.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
* Use uuid and do not md5 video files
* Fallback to ximagesink if xvimagesink is unaccessible
* Use one pass for encodings
* Remove glivex and add thumb_play to glive
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Record-59 is broken in joyride because we moved to a gstreamer, which
doesn't like these broken pipelines.
This commit reworks the whole pipeline system, making it easier to
read and generally better structured.
This is the work I did this summer, which didn't quite work for 8.2.
Now that we have upgraded to the newer gstreamer in F10, the bugs
have disappeared. There are a couple of quirks with video recording
but it works well.
|
|
|
|
|
| |
Revert to the old pipeline, which at least works 100% with the old
gstreamer versions that we are now shipping.
|
|
|
|
| |
Otherwise we get a really spooky live video lag
|
|
|
|
|
|
|
|
|
| |
When you add the video recording elements (theora + wav writing)
on-the-fly, there is a very visible freeze to the recorded video output
(several seconds).
This delay completely goes away if you stop the pipeline while doing the
reshuffling, and the visual glitch caused by this is minimal.
|
|
|
|
| |
Must have missed these hunks somehow
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Move to a model where the pipeline core is fixed, and we can plug in
photo and video bins to the tee element, and/or add an audiobin
separated from the rest for recording audio. Fixes problems caused by
not doing pad blocking and having unconnected elements lying around
in the pipeline.
Video recording is not quite working right... Audio and photo seem OK.
|
|
|
|
|
| |
It returns True on success, which results in the idle handler calling
the function again...
|
|
|
|
| |
A gstreamer assertion suggests this is the correct order of operations.
|
|
|
|
| |
git-svn-id: http://mediamods.com/public-svn/camera-activity/tags/51@1016 574bc980-5f2d-0410-acbc-c8f9f0eb14e0
|
|
|
|
| |
git-svn-id: http://mediamods.com/public-svn/camera-activity/tags/51@1013 574bc980-5f2d-0410-acbc-c8f9f0eb14e0
|