Preliminary Notes

245 views
Skip to first unread message

Derek Buitenhuis

unread,
May 12, 2016, 5:00:27 PM5/12/16
to spatial-media-discuss
Hi,

I've just read through the draft, and have a few small notes (since I'm more of a container guy than a 3D/VR guy).

  • Why are all the box names capitalized? This may be some convention I'm not aware of, but usually box names are lower case.
  • Under the semantics section for the PRHD box, it states "The values 0 to 255 are reserved for future layouts." but defines 0, 1, and 2. Should be changed to say "current and future layouts" or something. Same for the CBMP box.
  • Starting from the definition of the CBMP box, the "Mandatory" and "Quantity" sections are missing. I assume this is because it is supposed to be inferred by the fact that they must subclass PROJ?
  • Under semantics for the EQUI box it states "crop_right is the amount from the right of the frame to crop; must be less than 0xFFFFFFFF - crop_right". I assume this must be an error, since it's using itself in its own definition.
  • In general, these sorts of specs do not call the format MP4, but ISOBMFF.
  • For the example, is it perhaps useful for a more illustrative format, e.g. something like what boxdumper outputs (http://pastie.org/10834707). I have no strong feelings about it, though.

My coworker Vittorio should be able to provide more in-depth comments.


Cheers,

- Derek

Robert Suderman

unread,
May 12, 2016, 6:06:06 PM5/12/16
to spatial-media-discuss
Thanks for the quick feedback! Just some thoughts on your feedback and interested in hearing others thoughts.


On Thursday, May 12, 2016 at 2:00:27 PM UTC-7, Derek Buitenhuis wrote:
Hi,

I've just read through the draft, and have a few small notes (since I'm more of a container guy than a 3D/VR guy).

  • Why are all the box names capitalized? This may be some convention I'm not aware of, but usually box names are lower case.
We used capitalized box names to match another Spatial Audio spec  we published. If it is more typical to use lower case identifiers we should be able to change.
  • Under the semantics section for the PRHD box, it states "The values 0 to 255 are reserved for future layouts." but defines 0, 1, and 2. Should be changed to say "current and future layouts" or something. Same for the CBMP box.
 Sounds good.
  • Starting from the definition of the CBMP box, the "Mandatory" and "Quantity" sections are missing. I assume this is because it is supposed to be inferred by the fact that they must subclass PROJ?
 They subclass the Projection Data Box with a single mandatory type. Any suggestions on a clear way to indicate a single subclass of ProjectionDataBox is mandatory while not specifying the exact subclass?
  • Under semantics for the EQUI box it states "crop_right is the amount from the right of the frame to crop; must be less than 0xFFFFFFFF - crop_right". I assume this must be an error, since it's using itself in its own definition.
Correct, it should read crop_left. 
  • In general, these sorts of specs do not call the format MP4, but ISOBMFF.
Renaming to only ISOBMFF could be confusing for some users less familiar with the specification. Would referring to it as "MP4 (ISOBMFF)" be reasonable?
  • For the example, is it perhaps useful for a more illustrative format, e.g. something like what boxdumper outputs (http://pastie.org/10834707). I have no strong feelings about it, though.
I'll discuss with others. I'm more inclined to stick to the current format for brevity but we may be able to modify to the boxdumper format.

Derek Buitenhuis

unread,
May 17, 2016, 3:24:57 PM5/17/16
to spatial-media-discuss
On Thursday, May 12, 2016 at 11:06:06 PM UTC+1, Robert Suderman wrote:
Thanks for the quick feedback! Just some thoughts on your feedback and interested in hearing others thoughts.

[...]
 
We used capitalized box names to match another Spatial Audio spec  we published. If it is more typical to use lower case identifiers we should be able to change.
 
OK, I was merely curious.

 They subclass the Projection Data Box with a single mandatory type. Any suggestions on a clear way to indicate a single subclass of ProjectionDataBox is mandatory while not specifying the exact subclass?

It could just be a note in the PROJ definition or something, perhaps?
 
Renaming to only ISOBMFF could be confusing for some users less familiar with the specification. Would referring to it as "MP4 (ISOBMFF)" be reasonable?

Yep, sounds reasonable.
 
  • For the example, is it perhaps useful for a more illustrative format, e.g. something like what boxdumper outputs (http://pastie.org/10834707). I have no strong feelings about it, though.
I'll discuss with others. I'm more inclined to stick to the current format for brevity but we may be able to modify to the boxdumper format.

Perfectly fine. I suppose it's a matter of preference, and you can't please everyone.

Cheers,
- Derek

Robert Suderman

unread,
May 18, 2016, 5:21:51 PM5/18/16
to spatial-media-discuss
I have an attached draft with a few changes to see your thoughts. Tell me what you think,


On Tuesday, May 17, 2016 at 12:24:57 PM UTC-7, Derek Buitenhuis wrote:
On Thursday, May 12, 2016 at 11:06:06 PM UTC+1, Robert Suderman wrote:
Thanks for the quick feedback! Just some thoughts on your feedback and interested in hearing others thoughts.

[...]
 
We used capitalized box names to match another Spatial Audio spec  we published. If it is more typical to use lower case identifiers we should be able to change.
 
OK, I was merely curious.


I'll keep the possibility of using lower case identifiers open as it does fit the MP4 spec better.
 
 They subclass the Projection Data Box with a single mandatory type. Any suggestions on a clear way to indicate a single subclass of ProjectionDataBox is mandatory while not specifying the exact subclass?

It could just be a note in the PROJ definition or something, perhaps?
 

I've added some comments to PROJ to describe the content requirements. See what you think.
  
  • For the example, is it perhaps useful for a more illustrative format, e.g. something like what boxdumper outputs (http://pastie.org/10834707). I have no strong feelings about it, though.
I'll discuss with others. I'm more inclined to stick to the current format for brevity but we may be able to modify to the boxdumper format.

Perfectly fine. I suppose it's a matter of preference, and you can't please everyone.

I've added a layout closer to boxdumper that feels like an improvement over the bullet points. I removed the more verbose details so keep succinct.
 
spherical-video-rfc-2.md
Reply all
Reply to author
Forward
0 new messages