Removes keying on pattern_idx inside TemporalLayers implementations for
several properties that are different between screencast temporal layers
and normal/default temporal layers.
This is a step towards sharing PopulateCodecSpecific between the layer
patterns and code deduplication to longer term be able to separate the
packetizer step from encoder settings, so that temporal patterns can be
used for asynchronous hardware encoders where there may be outstanding
frames.
BUG=chromium:702017, webrtc:7349
R=brandtr@webrtc.org
Review-Url: https://codereview.webrtc.org/2924993002
Cr-Commit-Position: refs/heads/master@{#19097}
I used a command like this to update the paths:
perl -pi -e "s/webrtc\/base/webrtc\/rtc_base/g" `find webrtc/rtc_base -name "*.cc" -o -name "*.h"`
The only manual edit is to add an include of webrtc/rtc_base/checks.h in
webrtc/modules/audio_device/android/opensles_common.h, which likely
was needed due to changed include paths due to 'git cl format'.
BUG=webrtc:7634
NOTRY=True
NOPRESUBMIT=True
Review-Url: https://codereview.webrtc.org/2969653002
Cr-Commit-Position: refs/heads/master@{#18871}
The previous limit leaved no margin for RTT.
BUG=webrtc:4172
Review-Url: https://codereview.webrtc.org/2911243002
Cr-Commit-Position: refs/heads/master@{#18333}
If a frame is dropped and re-encoded because it exceeded the target
bitrate by a large factor, the next frame will be encoded at max qp
(worst quality) in order to get a frame through in a timely manner. The
next frame after this will still have lower quality since the rate
controller essentially gets reset. In order to mitigate that we boost
the qp for that next frame, which brings the stream back to a good
quality quicker.
However, if the network conditions are _really_ bad, this boosted qp
may be too large, causing the frame again to be dropped an re-encoded.
This CL set's a minimum bitrate available in order to enabling the
boosting in the first place.
It also adjusts a timeout (max time between frames in TL0), since a
too small value and very difficult frames in conjunction with the
mentioned bad network could actually cause bad network over-utilization
in turn leading to packet loss and bad follow-on effects to that.
There was also some slop in the rate keeping for the two layers.
This has been tightened up and affected test cases have been fixed.
BUG=webrtc:7694
Review-Url: https://codereview.webrtc.org/2897983002
Cr-Commit-Position: refs/heads/master@{#18236}
This is instead derived from pattern_idx inside the frame config.
This also removes active_layer_ use from
ScreenshareLayers::PopulateCodecSpecific and instead ties the layer to
TemporalLayers::FrameConfig.
BUG=chromium:702017, webrtc:7349
R=sprang@webrtc.org
Review-Url: https://codereview.webrtc.org/2860063002
Cr-Commit-Position: refs/heads/master@{#18017}
Partially removes "current frame" state from TemporalLayers, aiming to
tie more state to the frame config being encoded. This is necessary for
having several outstanding frames being encoded.
Also renames several structs, since TemporalReferences contains more
than references it's renamed TemporalLayers::FrameConfig.
BUG=chromium:702017, webrtc:7349
R=sprang@webrtc.org
Review-Url: https://codereview.webrtc.org/2853073004
Cr-Commit-Position: refs/heads/master@{#18012}
TESTED=video_loopback and AppRTCMobile with forced encoder reinits every 30 frames.
BUG=webrtc:7475
Review-Url: https://codereview.webrtc.org/2833493003
Cr-Commit-Position: refs/heads/master@{#17984}
Decouples encode flags and calculates them the same for both default and
screencast temporal layers.
With this change encoders could start using TemporalReferences for
temporal-layers flags, but they can not be used by asynchronous encoders
(hardware encoders) yet.
Also removes 'timestamp' as a dead parameter to FrameEncoded().
BUG=chromium:702017, webrtc:7349
R=marpan@google.com, sprang@webrtc.org, marpan@webrtc.org
Review-Url: https://codereview.webrtc.org/2769263002 .
Cr-Commit-Position: refs/heads/master@{#17397}
Use default temporal layers instead of the RealtimeTemporalLayers one.
When using low fps, having a single stream with much higher bitrate than
the lower simulcast stream causes poor rampup behavior.
Tested manually.
BUG=webrtc:4172
Review-Url: https://codereview.webrtc.org/2723983002
Cr-Commit-Position: refs/heads/master@{#16934}
The original CL was reverted because of a bug discovered by the
chromium bots. Description of that CL:
> Review-Url: https://codereview.webrtc.org/2636443002
> Cr-Commit-Position: refs/heads/master@{#16135}
> Committed: a28e971e3b
The first patch set of this CL is the same as r16135.
Subsequence patch sets are the fixes applied.
Some new test cases have been added, which reveal a few more bugs that
have also been fixed.
BUG=webrtc:4172
Review-Url: https://codereview.webrtc.org/2641133002
Cr-Commit-Position: refs/heads/master@{#16299}
Avoids confusion about the meaning of "incoming".
BUG=webrtc:6897
Review-Url: https://codereview.webrtc.org/2624073003
Cr-Commit-Position: refs/heads/master@{#16007}
Drop frames if incoming frame rate is higher than the configured max
framerate.
BUG=webrtc:6897
Review-Url: https://codereview.webrtc.org/2578993002
Cr-Commit-Position: refs/heads/master@{#15819}
A recent cl (https://codereview.webrtc.org/2510583002) introduced an
issue where temporal layers may return incorrect bitrates, given that
they are stateful and that the GetPreferredBitrateBps is called.
The fix is to use a temporary simulcast rate allocator instance, without
temporal layers, and get the preferred bitrate from that.
Additionally, some regression in bitrate allocated stems from overly
often reconfiguring the encoder, which yields suboptimal rate control.
The fix here is to limit encoder updates to when values have actually
changed.
As a bonus, dchecks added by this cl found a bug in the (unused) RealtimeTemporalLayers implementation. Fixed that as well.
BUG=webrtc:6301, chromium:666654
Review-Url: https://codereview.webrtc.org/2529073003
Cr-Commit-Position: refs/heads/master@{#15250}
This is yet another reland of https://codereview.webrtc.org/2434073003/
including two fixes:
1. SimulcastRateAllocator did not handle the screenshare settings properly for numSimulcastStreams = 1. Additional test case was added for that.
2. In VideoSender, when rate allocation is updated after setting a new VideoCodec config, only update the state of the EncoderParameters, but don't actually run SetRateAllocation on the encoder itself. This caused some problems upstreams.
Please review only the changes after patch set 1.
Original description:
Extract bitrate allocation of spatial/temporal layers out of codec impl.
This CL makes a number of intervowen changes:
* Add BitrateAllocation struct, that contains a codec independent view
of how the target bitrate is distributed over spatial and temporal
layers.
* Adds the BitrateAllocator interface, which takes a bitrate and frame
rate and produces a BitrateAllocation.
* A default (non layered) implementation is added, and
SimulcastRateAllocator is extended to fully handle VP8 allocation.
This includes capturing TemporalLayer instances created by the
encoder.
* ViEEncoder now owns both the bitrate allocator and the temporal layer
factories for VP8. This allows allocation to happen fully outside of
the encoder implementation.
This refactoring will make it possible for ViEEncoder to signal the
full picture of target bitrates to the RTCP module.
BUG=webrtc:6301
R=stefan@webrtc.org
Review URL: https://codereview.webrtc.org/2510583002 .
Cr-Commit-Position: refs/heads/master@{#15105}
Reason for revert:
Seems to be causing flakiness in perf test:
FullStackTest.ScreenshareSlidesVP8_2TL_LossyNet
Original issue's description:
> Reland of Issue 2434073003: Extract bitrate allocation ...
>
> This is a reland of https://codereview.webrtc.org/2434073003/ including
> some fixes for failing test cases.
>
> Original description:
>
> Extract bitrate allocation of spatial/temporal layers out of codec impl.
>
> This CL makes a number of intervowen changes:
>
> * Add BitrateAllocation struct, that contains a codec independent view
> of how the target bitrate is distributed over spatial and temporal
> layers.
>
> * Adds the BitrateAllocator interface, which takes a bitrate and frame
> rate and produces a BitrateAllocation.
>
> * A default (non layered) implementation is added, and
> SimulcastRateAllocator is extended to fully handle VP8 allocation.
> This includes capturing TemporalLayer instances created by the
> encoder.
>
> * ViEEncoder now owns both the bitrate allocator and the temporal layer
> factories for VP8. This allows allocation to happen fully outside of
> the encoder implementation.
>
> This refactoring will make it possible for ViEEncoder to signal the
> full picture of target bitrates to the RTCP module.
>
> BUG=webrtc:6301
>
> Committed: https://crrev.com/647bf43dcb2fd16fccf276bd94dc4400728bb405
> Cr-Commit-Position: refs/heads/master@{#15023}
TBR=mflodman@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=webrtc:6301
Review-Url: https://codereview.webrtc.org/2491393002
Cr-Commit-Position: refs/heads/master@{#15026}
This is a reland of https://codereview.webrtc.org/2434073003/ including
some fixes for failing test cases.
Original description:
Extract bitrate allocation of spatial/temporal layers out of codec impl.
This CL makes a number of intervowen changes:
* Add BitrateAllocation struct, that contains a codec independent view
of how the target bitrate is distributed over spatial and temporal
layers.
* Adds the BitrateAllocator interface, which takes a bitrate and frame
rate and produces a BitrateAllocation.
* A default (non layered) implementation is added, and
SimulcastRateAllocator is extended to fully handle VP8 allocation.
This includes capturing TemporalLayer instances created by the
encoder.
* ViEEncoder now owns both the bitrate allocator and the temporal layer
factories for VP8. This allows allocation to happen fully outside of
the encoder implementation.
This refactoring will make it possible for ViEEncoder to signal the
full picture of target bitrates to the RTCP module.
BUG=webrtc:6301
Review-Url: https://codereview.webrtc.org/2488833004
Cr-Commit-Position: refs/heads/master@{#15023}
Reason for revert:
Breaks perf tests.
Original issue's description:
> Extract bitrate allocation of spatial/temporal layers out of codec impl.
>
> This CL makes a number of intervowen changes:
>
> * Add BitrateAllocation struct, that contains a codec independent view
> of how the target bitrate is distributed over spatial and temporal
> layers.
>
> * Adds the BitrateAllocator interface, which takes a bitrate and frame
> rate and produces a BitrateAllocation.
>
> * A default (non layered) implementation is added, and
> SimulcastRateAllocator is extended to fully handle VP8 allocation.
> This includes capturing TemporalLayer instances created by the
> encoder.
>
> * ViEEncoder now owns both the bitrate allocator and the temporal layer
> factories for VP8. This allows allocation to happen fully outside of
> the encoder implementation.
>
> This refactoring will make it possible for ViEEncoder to signal the
> full picture of target bitrates to the RTCP module.
>
> BUG=webrtc:6301
>
> Committed: https://crrev.com/8f46c679d24a05b3f08e02c6d91ec9637f34e24f
> Cr-Commit-Position: refs/heads/master@{#14998}
TBR=stefan@webrtc.org,perkj@webrtc.org,mflodman@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=webrtc:6301
Review-Url: https://codereview.webrtc.org/2489843002
Cr-Commit-Position: refs/heads/master@{#15001}
This CL makes a number of intervowen changes:
* Add BitrateAllocation struct, that contains a codec independent view
of how the target bitrate is distributed over spatial and temporal
layers.
* Adds the BitrateAllocator interface, which takes a bitrate and frame
rate and produces a BitrateAllocation.
* A default (non layered) implementation is added, and
SimulcastRateAllocator is extended to fully handle VP8 allocation.
This includes capturing TemporalLayer instances created by the
encoder.
* ViEEncoder now owns both the bitrate allocator and the temporal layer
factories for VP8. This allows allocation to happen fully outside of
the encoder implementation.
This refactoring will make it possible for ViEEncoder to signal the
full picture of target bitrates to the RTCP module.
BUG=webrtc:6301
Review-Url: https://codereview.webrtc.org/2434073003
Cr-Commit-Position: refs/heads/master@{#14998}
If a very large frame is sent (high res slide change) when the available
send bitrate is very low, the it might take many seconds before any new
frames are emitted as the accrued debt will take time to pay off.
Add a bailout, so that if a frame hasn't been sent for 2 seconds, cancel
the debt immediately, even if the target bitrate is then exceeded.
BUG=webrtc:5750
Review URL: https://codereview.webrtc.org/1869003002
Cr-Commit-Position: refs/heads/master@{#12328}
Adds logging of:
- video stats that are recorded when a stream is removed
- bitrate stats that are recorded at the end of a call
- initial bwe rampup stats
BUG=
Review URL: https://codereview.webrtc.org/1788783002
Cr-Commit-Position: refs/heads/master@{#12133}
active_layer_ could be dereferenced while being -1...
Also added som DCHECKs
BUG=webrtc:5490
Review URL: https://codereview.webrtc.org/1656233002
Cr-Commit-Position: refs/heads/master@{#11486}
The main purpose was the interface-> include rename, but other files
were also moved, eliminating the "main" dir.
To avoid breaking downstream, the "interface" directories were copied
into a new "video_coding/include" dir. The old headers got pragma
warnings added about deprecation (a very short deprecation since I plan
to remove them as soon downstream is updated).
Other files also moved:
video_coding/main/source -> video_coding
video_coding/main/test -> video_coding/test
BUG=webrtc:5095
TESTED=Passing compile-trybots with --clobber flag:
git cl try --clobber --bot=win_compile_rel --bot=linux_compile_rel --bot=android_compile_rel --bot=mac_compile_rel --bot=ios_rel --bot=linux_gn_rel --bot=win_x64_gn_rel --bot=mac_x64_gn_rel --bot=android_gn_rel -m tryserver.webrtc
R=stefan@webrtc.org, tommi@webrtc.org
Review URL: https://codereview.webrtc.org/1417283007 .
Cr-Commit-Position: refs/heads/master@{#10694}
We must remove dependency on Chromium, i.e. we can't use Chromium's base/logging.h. That means we need to define these macros in WebRTC also when doing Chromium builds. And this causes redefinition.
Alternative solutions:
* Check if we already have defined e.g. CHECK, and don't define them in that case. This makes us depend on include order in Chromium, which is not acceptable.
* Don't allow using the macros in WebRTC headers. Error prone since if someone adds it there by mistake it may compile fine, but later break if a header in added or order is changed in Chromium. That will be confusing and hard to enforce.
* Ensure that headers that are included by an embedder don't include our macros. This would require some heavy refactoring to be maintainable and enforcable.
* Changes in Chromium for this is obviously not an option.
BUG=chromium:468375
NOTRY=true
Review URL: https://codereview.webrtc.org/1335923002
Cr-Commit-Position: refs/heads/master@{#9964}
This change includes several improvements:
* VP8 configured with new rate control
* Detection of frame dropping, with qp bump for next frame
* Increased target and TL0 bitrates
* Reworked rate control (TL allocation) in screenshare_layers
A note on performance: PSNR and SSIM is expected to get slightly worse with this cl. Frame drops and delays should however improve.
BUG=4171
R=pbos@webrtc.org, stefan@webrtc.org
Review URL: https://codereview.webrtc.org/1193513006.
Cr-Commit-Position: refs/heads/master@{#9495}
