Comments on: The Challenging Path to WebRTC H.264 Video Codec Hardware Support https://bloggeek.me/webrtc-h264-video-codec-hardware-support/ The leading authority on WebRTC Mon, 27 Apr 2020 13:37:58 +0000 hourly 1 By: Tsahi Levent-Levi https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119218 Mon, 01 Oct 2018 04:53:20 +0000 https://bloggeek.me/?p=12600#comment-119218 In reply to Suzanne Spears.

Safari is still at its infancy when it comes to WebRTC, so things tend to break there more in other places.
Who fixes stuff when things break? Should it be Safari or the other browser vendors? That’s not easy to answer.

As a developer, the question then becomes who do you care about more at the moment? Safari users or Firefox/other users?

Assuming iOS (not inside an app) is of high priority for you, then you’ll be opting for H.264 and make do with whatever quirks and limits Safari imposes on you. Otherwise, I’d just focus on VP8 if I were you.

]]>
By: Suzanne Spears https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119217 Sun, 30 Sep 2018 20:27:02 +0000 https://bloggeek.me/?p=12600#comment-119217 How do these pieces of video support software and specifications for VP8, H.264 and WebRTC relate to the SpectrumTV website in Safari (it currently works – does not use VP8) and in Firefox (it does not work, since, v62.0, ~9/5/18 – uses VP8)

]]>
By: Lennart Grahl https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119215 Tue, 10 Jul 2018 21:49:12 +0000 https://bloggeek.me/?p=12600#comment-119215 Apparently the Edge team also doesn’t understand the meaning of “mandatory”. The underdog of WebRTC: Data Channels! 😉

]]>
By: Tsahi Levent-Levi https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119214 Mon, 09 Jul 2018 20:13:05 +0000 https://bloggeek.me/?p=12600#comment-119214 In reply to Arik.

Arik,

Thanks. Do note you are making the assumption all devices everywhere have GPU that is accessible to you with all the building blocks needed to get that codec implemented and that implementation is uniform in its feature set across all devices. Which is where things fall apart, even without the fact that many of these devices don’t accelerate video coding using a GPU in the first place.

]]>
By: Arik https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119213 Mon, 09 Jul 2018 18:31:56 +0000 https://bloggeek.me/?p=12600#comment-119213 As far as I understood, codec hardware acceleration today is manly accelerated components, algorithms and methods rather than complete codecs. So your GPU could do efficient transforms, motion vector estimation, etc which then can be used for acceleration of any codec (VP8/9 as well as H.).
With this methodology you also get the advantage of being able to update the CPU codec logic without replacing the accelerating hardware and without falling back to software-only processing.

]]>
By: Tsahi Levent-Levi https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119212 Mon, 09 Jul 2018 15:11:38 +0000 https://bloggeek.me/?p=12600#comment-119212 In reply to Philipp Hancke.

Thanks Fippo 🙂

I think VP8 is a lot simpler. If you add it to your hardware, there’s only one reason for you to do so (or a main one) – WebRTC. And the way you test WebRTC today is by running it against Chrome.

H.264 is the swiss army knife of the current video codec generation, which means it gets pitted against many different use cases where WebRTC is but a minor niche.

]]>
By: Philipp Hancke https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119211 Mon, 09 Jul 2018 14:41:36 +0000 https://bloggeek.me/?p=12600#comment-119211 In reply to Philipp Hancke.

and VP8 hardware accelleration is around the corner for Intels Kaby lake processors: https://groups.google.com/a/chromium.org/forum/?utm_medium=email&utm_source=footer#!msg/blink-dev/vbYCDv5ve5w/pq-uV_QRAwAJ

]]>
By: Philipp Hancke https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119210 Mon, 09 Jul 2018 14:39:23 +0000 https://bloggeek.me/?p=12600#comment-119210 Mind you that the VP8 realtime hardware requirements haven’t really changed since 2012:
https://www.webmproject.org/hardware/rtc-coding-requirements/
The page doesn’t say “simulcast” anywhere but the requirements ensure the encoder and decoder are capable of decoding what Chrome produces with simulcast including temporal scalability.

]]>
By: Tsahi Levent-Levi https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119209 Mon, 09 Jul 2018 12:54:39 +0000 https://bloggeek.me/?p=12600#comment-119209 In reply to Gustavo Garcia.

As with the case of John, your 2c are worth their weight in pure gold 😉

]]>
By: Gustavo Garcia https://bloggeek.me/webrtc-h264-video-codec-hardware-support/#comment-119208 Mon, 09 Jul 2018 12:52:01 +0000 https://bloggeek.me/?p=12600#comment-119208 Nice post Tsahi. The thing I found more interesting about H.264 is using it for 1:1 iOS calls. In that case you don’t need to worry about Android and OpenH264 and you get access to High Profile.

The fact that it is (supposed to be) used by Duo in that specific setup makes me feel it is worth the effort and it is tested/stable enough in that setup. Still it is not trivial to use and the battery savings can be not as big as they were supposed to be.

Just my 2c

]]>