Proposal: align video scaling with JSEP

All,

sorry for a long message, but I've been looking at how to resolve Issue 
#1283 [a] 'centering, scaling, cropping'. The Issue itself says that 
JSEP and webrtc-pc are not aligned and was reported by Ekr as part of 
his full read through of webrtc-pc in May.

Long story short: I think webrtc-pc should be updated to follow what 
JSEP says, and think we should merge the PR [b].

The reason why is that the current webrtc-pc spec (in my opinion) is 
vague and that it assumes knowledge the sender does not have.

Webrtc-pc [c] talks about a 'desired' video size, but what is desired is 
not defined. If the video track at the receiving end was displayed in a 
single video element, the dimensions of that element would probably be 
the 'desired' size, but first of all such dimensions are not signaled to 
the sender, and secondly it is easy to imagine situations where the 
video at the receiving end is displayed on several video elements of 
different dimensions.

What _is_ signaled from receiver to sender (using the 'imageattr' SDP 
attribute) is video dimensions that the receiver supports [d], but that 
is not the same as the size of a video element but rather the 
capabilities of the receiver.

The current text in webrtc-pc [c] is said to be made to "minimize 
occurrence of images with with letter box or or pillow boxing.", but as 
explained above that can't be made without knowledge of the dimensions 
of the rendering video element.

Therefore I propose we remove the current text on 'centering, scaling, 
cropping', and instead reference JSEP section 3.6.2.

Thoughts?

//Stefan

[a] https://github.com/w3c/webrtc-pc/issues/1283
[b] https://github.com/w3c/webrtc-pc/pull/1570
[c] https://w3c.github.io/webrtc-pc/#rtp-media-api
[d] http://rtcweb-wg.github.io/jsep/#rfc.section.3.6.2

Received on Tuesday, 26 September 2017 11:57:07 UTC