Low resolutions when using H264

[fho...@nanocosmos.de]

Hello,

we are experiencing low resolutions beeing sent when using H264 codec in chrome.

We use the newest adapter.js from here: https://github.com/webrtc/adapter/tree/master/release

Those are the settings we use:

Resolution: 720p

Framerate: 30

videosendinitialbitrate: 1000 (changing does not change behaviour)

videosendbitrate: 1500 (changing does not change behaviour)

We could verify that all settings are beeing adopted in chrome://webrtc-internals, all but the resolution. Bitrate is very low (100k - 200k) but

when shaking the camera the video bitrate reaches a maximum value of videosendbitrate. 

Additionally we checked googBandwidthLimitedResolution, googCpuLimitedResolution and googViewLimitedResolution in webrtc-interals which all 

are false all the time. Resolution looks like this:

mediaType : video

googCodecName : H264

googFrameWidthInput : 1280

googFrameWidthSent : 320

googFrameHeightInput : 720

googFrameHeightSent : 180

googFrameRateInput : 29

googFrameRateSent : 30

If we lower the framerate (to about 10 fps) this issue disappears but thats not sufficient for our use case. Sometimes we can trick the

sent resolution to reach the target resolution by shaking the camera.

When changing the codec back to VP8 we see that initially the resolution drops (for about 3 seconds) but then reaches the desired 720p resolution.

VP9 works without the initial drop.

We would need the codec to be H264 in the desired resolution to match our use case. Occasional frame drops 

and varying birates would not be a big problem.

Did anyone experience similar issues with H264? Is there a way guarantee a fixed resolution?

Best regards

Finn Hoeck

[Niklas Enbom]

Hi, which chrome version and OS is this on? I'd like to know if this is using hardware encoding or not.

--

---
You received this message because you are subscribed to the Google Groups "discuss-webrtc" group.
To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrtc+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/41cb4f2d-3211-4204-bb05-2e0da86dbb36%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[fho...@nanocosmos.de]

Sorry, forgot to mention.

This is:

win 10

chrome Version 53.0.2785.116 m

in chrome://flags "WebRTC H.264 software video encoder/decoder " is enabled.

To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrt...@googlegroups.com.

[Niklas Enbom]

OK, that's software encoding then. Can you file a Chrome bug and include a screen shot of the video send graphs under webrtc-internals? 

To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrtc+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/3a1b9eb1-0954-45f9-b2f1-6c7e7cecb9c7%40googlegroups.com.

[fho...@nanocosmos.de]

Ok, I filed a report:

https://bugs.chromium.org/p/chromium/issues/detail?id=651342

Thank you.

Best regards

Finn Hoeck

[fho...@nanocosmos.de]

It seems to be somehow confirmed that H264 codec causes adaptation of resolution because of cpu usage (openH264 is not multithreaded and still has issues, see answer in https://bugs.chromium.org/p/chromium/issues/detail?id=651342). 

But my original question still remains: Is there a way to prefer bitrate adaptation (and/or framerate) over resolution adaptation, maybe as SDP patch or some other constraint? I could not find any info about degradation preferences beeing implemented (https://www.w3.org/TR/webrtc/#idl-def-rtcdegradationpreference). Is that planned?

Best regards,

Finn Hoeck