--
---
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-webrt...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/14a2e1fa-98fb-4b05-80ba-6ed35b001fc1%40googlegroups.com.

You received this message because you are subscribed to a topic in the Google Groups "discuss-webrtc" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/discuss-webrtc/ZUigX7cSrdA/unsubscribe.
To unsubscribe from this group and all its topics, send an email to discuss-webrt...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/CAHgZEq7eZSAob6Re-w4cdWRo%3Di%2BZ7EZ_jj0hSzFeXDSP3A-VJw%40mail.gmail.com.
Regarding this sentence: As a side note, please bear in mind that 1080p@30fps requires 6 to 7 Mbps even with H265 in two-pass best quality modeHow do you explain Netflix and google are using a bitrate close to 2.5Mbps for their full HD stream ?
Regarding this diagram found here (https://blogs.gnome.org/rbultje/2015/09/28/vp9-encodingdecoding-performance-vs-hevch-264/)Bitrate vs SSIM2.5Mbps seems to be a good tradeoff between quality and bitrate, isn’t it ?
Second question: Why are you talking about H265 as this codec is not supported by browsers in webrtc (or maybe I miss something) ?
Hi ju,those are great questions, thank you for asking.
Regarding this sentence: As a side note, please bear in mind that 1080p@30fps requires 6 to 7 Mbps even with H265 in two-pass best quality modeHow do you explain Netflix and google are using a bitrate close to 2.5Mbps for their full HD stream ?By google you mean ..... hangout? duo? stadia? youtube? meet? All of those google products use different solutions with different compromises and without knowing which one you refer to, I can't answer your question.
As for Netflix, the answer is easy, they are using two pass encoding, with both per-title (since 2015) and per-segment ML-optimized encoding. It takes forever to encode, but they don t care since, as a pre-recorded content provider, they have infinite time budget to encode. Not only that, but they switched to VP9 many years ago (with optimised implementation and GPu encoding), and more recently AV1, giving them an additional 25 to 30% (average across all the resolutions and content of their media library) improvement each time.
If you want to learn more about per-title and per-segment/GOP encoding, you can start by those two blog post:
Ane Aaron (Netflix director for encoding) presentations at live streaming east / west or IBC2019 INTEL event, where we also presented are also providing interesting insights:All of that is just a google search away....Regarding this diagram found here (https://blogs.gnome.org/rbultje/2015/09/28/vp9-encodingdecoding-performance-vs-hevch-264/)Bitrate vs SSIM2.5Mbps seems to be a good tradeoff between quality and bitrate, isn’t it ?1. Both axis use logarithmic scales, which makes the 2.5mbps point difficult to visualise.
2. Without a reference point, you can compare the curves (the one on top is better), but you still cannot collude on wether it s good or not.
3. Libvpx is an entire library including several codecs like Vp8, VP9, ... I don't know which one has been used here.
Equally, x264 and x265 references *implementations* and not codecs themselves. If we go down this path, you should compare all the H264 implementations against the one provided by chrome, and equally all VP8 or Vp9 implementations.the only thing I can conclude from your drawing is: x265 seems better than h264 everything else being equal.
2.5 does not seem to be a good tradeoff. Moreover SSIM has but a poor correlation with perceived quality.
H.265 is not yet supported in browser, even though INTEL and Apple came together in November last year with our kind mediation, and implemented H.265 HW acceleration support in libwebrtc, webkit, and safari.Second question: Why are you talking about H265 as this codec is not supported by browsers in webrtc (or maybe I miss something) ?
The reasons why I took this one as a reference point are simple.1. it's a newer/better codec than H.264 or VP8, the two codecs available for webrtc, especially for high resolution for which it has been especially designed (with larger maximum tiles size among other things), So if you cannot even achieve your target bitrate with that better codec, you should not expect to be able to do it with H.264/VP8. This is a "sanity-check". If you cannot drive that fast with a ferrari, do not even try with a fiat punt.2. There is an industry standard reference table provided by Apple, used by everybody streaming to and from any apple product including apple TV, which is dependable, as opposed to some random blog post.
Hope this helps.