Like video_decoder.cc, a call to Encode that returns
WEBRTC_VIDEO_CODEC_FALLBACK_SOFTWARE will trigger an attempted fallback
to a built-in software encoder. Initialization information, along with
any rate and channel parameter info, will be replayed on the software
encoder and then the frame (that cause the fallback) will be immediately
replayed for the software encoder.
Also modified the existing behavior to Release() the "real" encoder even
if a fallback encoder exists. That seems like the correct behavior.
BUG=webrtc:2920
Review URL: https://codereview.webrtc.org/1328863002
Cr-Commit-Position: refs/heads/master@{#10368}
- "WebRTC.Video.BandwidthLimitedResolutionInPercent"
If the frame is bandwidth limited, the average number of disabled resolutions is logged:
- "WebRTC.Video.BandwidthLimitedResolutionsDisabled"
BUG=
Review URL: https://codereview.webrtc.org/1311533012
Cr-Commit-Position: refs/heads/master@{#10333}
Reason for revert:
Reverting to see of this resolves build bot failures (Nexus 7.2, especially debug) which now seems to sometimes fail to start tests altogether.
Original issue's description:
> Add screenshare perf tests with lossy links
>
> BUG=
>
> Committed: https://crrev.com/4af0f1a098bc908cfe825981b825687673d5134a
> Cr-Commit-Position: refs/heads/master@{#10290}
TBR=pbos@webrtc.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=
Review URL: https://codereview.webrtc.org/1415603002
Cr-Commit-Position: refs/heads/master@{#10322}
- "WebRTC.Video.QualityLimitedResolutionInPercent"
and if a frame is downscaled, the average number of times the frame is downscaled:
- "WebRTC.Video.QualityLimitedResolutionDownscales"
BUG=
Review URL: https://codereview.webrtc.org/1325153009
Cr-Commit-Position: refs/heads/master@{#10319}
This CL changes as little as possible and I'll follow up later with
ownership of the other members in ChannelGroup.
The next step is to remove the id used for channels.
BUG=webrtc:5079
Review URL: https://codereview.webrtc.org/1411723002
Cr-Commit-Position: refs/heads/master@{#10318}
This is the first CL to get ready for adapting audio bitrate based on
BWE. I've kept this CL as small as possible and had to add a few getters
to ChannelManager. The next CL will do the same for receive ViEChannels.
The getters are a bit uggly, but is an in-between-state. Let's discuss
future ownership of the different modules and what do do with
ChannelGroup.
BUG=5079
Review URL: https://codereview.webrtc.org/1394243006
Cr-Commit-Position: refs/heads/master@{#10298}
Also updated the RTPPayloadRegistry::RestoreOriginalPacket signature to not take the first arg as a **, since it isn't modified.
Review URL: https://codereview.webrtc.org/1394573004
Cr-Commit-Position: refs/heads/master@{#10276}
Reason for revert:
Temporarily reverting as this causes some issues with perf tests. Especially tests with packet loss no longer works.
Original issue's description:
> Adding support for simulcast and spatial layers into VideoQualityTest
>
> The CL includes several changes:
> - Adding flags describing the streams and spatial layers.
> - Reorganizing the order of the flags, to make them easier to maintain.
> - Adding a member .params_ to VideoQualityAnalyzer.
> (instead of passing it to every member function manually)
> - Updating VideoAnalyzer to support simulcast.
> (select appropriate ssrc and fix timestamps which are sometimes increased by 1)
> - VP9EncoderImpl already had code for automatic calculation of bitrate for each layer.
> Changing to first read bitrates and resolution ratios from the flags, if specified.
> If not specified, reverting to the old code are setting the values automatically.
> - Changing the parameters in LayerFilteringTransport, replacing
> xx_discard_thresholds with selected_xx, to make it easier to use for the end user.
>
> Committed: https://crrev.com/87f83a9a27d657731ccb54025bc04ccad0da136e
> Cr-Commit-Position: refs/heads/master@{#10215}
TBR=pbos@webrtc.org,mflodman@webrtc.org,ivica@webrtc.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Review URL: https://codereview.webrtc.org/1397363002
Cr-Commit-Position: refs/heads/master@{#10252}
The CL includes several changes:
- Adding flags describing the streams and spatial layers.
- Reorganizing the order of the flags, to make them easier to maintain.
- Adding a member .params_ to VideoQualityAnalyzer.
(instead of passing it to every member function manually)
- Updating VideoAnalyzer to support simulcast.
(select appropriate ssrc and fix timestamps which are sometimes increased by 1)
- VP9EncoderImpl already had code for automatic calculation of bitrate for each layer.
Changing to first read bitrates and resolution ratios from the flags, if specified.
If not specified, reverting to the old code are setting the values automatically.
- Changing the parameters in LayerFilteringTransport, replacing
xx_discard_thresholds with selected_xx, to make it easier to use for the end user.
Review URL: https://codereview.webrtc.org/1353263005
Cr-Commit-Position: refs/heads/master@{#10215}
Since the pacer is always enabled, removing enable/disable which makes
all packet queueing succeed. Also renaming one of the ::SendPackets
::InsertPacket to avoid confusion.
BUG=webrtc:1695, webrtc:2629
R=stefan@webrtc.org
Review URL: https://codereview.webrtc.org/1392513002 .
Cr-Commit-Position: refs/heads/master@{#10211}
In video_sender.cc, properly read the number of temporal layers for VP9 too.
Also, some cleanup in video_loopback.cc and video_quality_test.h.
Review URL: https://codereview.webrtc.org/1351693005
Cr-Commit-Position: refs/heads/master@{#10201}
Also, in Sample struct, replacing double with the original type.
It makes more sense to save the original data as truthful as possible, and then
convert it to double later if necessary (in the plot script).
Review URL: https://codereview.webrtc.org/1374233002
Cr-Commit-Position: refs/heads/master@{#10184}
When fetching a packet from the rtp packet history, cuased by a
retransmission, the transport seq extension header is enabled but the
sequence number is set to 0. A new transport seq should be assigned in
this case.
BUG=
Review URL: https://codereview.webrtc.org/1385563005
Cr-Commit-Position: refs/heads/master@{#10183}
Since padding is no longer sent on Encoded() callbacks, dummy callbacks
aren't required to generate padding. This skip-frame behavior can then
be removed to get rid of dummy callbacks though nothing was encoded. As
frames don't have to be generated for frames that don't have to be sent
we skip encoding frames that aren't intended to be sent either, reducing
CPU load.
BUG=
R=mflodman@webrtc.org, stefan@webrtc.org
Review URL: https://codereview.webrtc.org/1369923005 .
Cr-Commit-Position: refs/heads/master@{#10181}
Poller thread is currently started in the constructor, so the first call
to PollStats() may happen even before the streams have been configured.
This will blow up on RTC_DCHECK_GT(expected_bitrate_bps_, 0);
Thread should instead be started on PerformTest() call.
BUG=
Review URL: https://codereview.webrtc.org/1378303004
Cr-Commit-Position: refs/heads/master@{#10149}
This allows us to pass packet meta data, such as transport sequence
number, to libjingle and further down to the socket implementation. A
similar struct already exist in libjingle, see rtc::PacketOptions in asyncpacketsocket.h.
BUG=4173
Review URL: https://codereview.webrtc.org/1376673004
Cr-Commit-Position: refs/heads/master@{#10144}
In addition to this the ramp-up tests are refactored to use a receive call instead of only a remote bitrate estimator, and to make use of BaseTest.
BUG=webrtc:4836
Review URL: https://codereview.webrtc.org/1368943002
Cr-Commit-Position: refs/heads/master@{#10087}
The BWE expects arrival timestamps in ms, while the audio path delivered
them in us, causing the BWE to break down under the combined audio/video
BWE experiment. This was introduced in r9892 (68786d2040).
BUG=webrtc:4758
R=mflodman@webrtc.org, sprang@webrtc.org
Review URL: https://codereview.webrtc.org/1360913004 .
Cr-Commit-Position: refs/heads/master@{#10032}
In the middle of refactoring, I replaced the VideoCapturer with
FrameGeneratorCapturer, to reuse the code, and with that disabled the camera.
Now adding capturer_ element to VideoQualityTest and ignoring
frame_generator_capturer_ from the parent class test::CallTest.
Review URL: https://codereview.webrtc.org/1356933005
Cr-Commit-Position: refs/heads/master@{#10023}
In screensharing full stack tests, instead of using YuvFileGenerator by default
when no scrolling is used, I always used ScrollingImageFileGenerator.
That possibly slowed down the test a little bit, at least for the slowed
devices, as it unnecessarily copied few MBs per frame.
BUG=chromium:534220
Review URL: https://codereview.webrtc.org/1359783002
Cr-Commit-Position: refs/heads/master@{#10014}
