Re: [avtext] Frame marking & VP9 SVC

["Mo Zanaty (mzanaty)" <[e-mail]>]

Hi Sergio,

Good catch. The VP9 payload format has changed since the Frame Marking LID extensions were defined. The VP9 authors expect it may change further. Based on discussion during today's AVT session, we will remove VP9 from the Frame Marking draft (which will help avoid MIS-REF later), and move that information to the VP9 payload format so it can evolve without diverging.

The discussion on U and P bits was more nuanced. The main point is Frame Marking should help identify when layer refresh has occurred and when layer up switching is possible. We will add some text on this for simple, fixed scalability structures, and note that complex, non-fixed structures require payload-specific inspection and logic that no extra bits will completely eliminate.

Cheers,
Mo

From: avtext <[e-mail]<mailto:[e-mail]>> on behalf of Sergio Murillo <[e-mail]<mailto:[e-mail]>>
Date: Tuesday, March 28, 2017 at 9:54 AM
To: "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>
Subject: [avtext] Frame marking & VP9 SVC


Hi,

I have been implementing the framemarking for PERC and I have some doubts regarding the VP9 LID mapping.

First, is that according to https://tools.ietf.org/html/draft-ietf-payload-vp9-03 there are no quality layers on VP9 SVC they are considered as case of spatial layers:

   VP9 supports quality layers as spatial layers without any resolution
   changes; hereinafter, the term "spatial layer" is used to represent
   both spatial and quality layers.

So, shouldn't the LID reference to the spatial layer ID only  and omit the quality layer id completely? Also, the spatial layer id is 3 bits on that draft.

Also, in order to be able to implement VP9 SVC layer selection, the SFU needs the information of the P and U bits of the VP9 header description:

   P: Inter-picture predicted layer frame.  When set to zero, the layer
      frame does not utilize inter-picture prediction.  In this case,
      up-switching to current spatial layer's frame is possible from
      directly lower spatial layer frame.  P SHOULD also be set to zero
      when encoding a layer synchronization frame in response to an LRR
      [I-D.ietf-avtext-lrr<https://tools.ietf.org/html/draft-ietf-payload-vp9-03#ref-I-D.ietf-avtext-lrr>] message (see Section 5.4<https://tools.ietf.org/html/draft-ietf-payload-vp9-03#section-5.4>).  When P is set to
      zero, the T bit (described below) MUST also be set to 0 (if
      present).



   U: Switching up point.  If this bit is set to 1 for the current
      frame with temporal layer ID equal to T, then "switch up" to a
      higher frame rate is possible as subsequent higher temporal
      layer frames will not depend on any frame before the current
      frame (in coding time) with temporal layer ID greater than T.


Without that info, the SFU won't be able to upscale, so it would be required to add that information on the VP9 LID.
Including that changes, it would be something like this:


    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  ID=2 |  L=2  |S|E|I|D|B| TID |0|0|0|P|U| SID |    TL0PICIDX  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


Best regards
Sergio