[AVTCORE] Query on Picture Abstraction and Implementation in RFC 9628

[Diptanu De <[e-mail]>]

Hello Authors,

I hope you are doing well.

I recently had the opportunity to go through RFC 9628 in detail, and I
wanted to sincerely appreciate the thought behind the design — particularly
the introduction of the “picture” abstraction for grouping frames across
spatial layers. It feels like a very clean and well-reasoned way to bridge
VP9’s frame-based structure with RTP’s transport and synchronization
requirements.

While studying the document, I was trying to relate this concept to
existing VP9 mechanisms such as superframes, because *from a high level, a
"picture" (which contains multiple subframes at different scalable layer)
seems identical to that of a superframe*. Although it is stated that the
concept of "picture" is different from a superframe (in the Picture ID
section of RFC 9628), I am particularly interested in understanding how the
detection of a "picture" will be implemented from a decoder PoV. *From VP9
spec, the superframe detection is quite explicit and deterministic from the
bitstream (via the last 34 bits of the superframe packet), whereas I could
not quite catch the underlying implementation behind picture detection from
a VP9 bitstream*.

So, it would be great *to understand your thought process around designing
"picture" detection, particularly how you see it being derived from the
bitstream* in practical implementations.

I find this design choice quite interesting, as it introduces a logical
abstraction without modifying the underlying codec behavior, and I was
curious to better understand the intended implementation perspective from
the authors themselves. Thank you again for your work on this
specification—it has been very insightful to study.

Regards,
Diptanu D