Roman_Guskov authored 2 years ago and Nikolay Shostak committed 2 years ago

Decoder expects that 2 fields of one frame are stored in pcListPic next to each other.

But in case of Multilayer there will be some distance between field pictures. For example:

- POC0 (TopField) Layer0
- POC0 (TopField) Layer1
- POC0 (TopField) Layer2
- POC1 (BottomField) Layer0
- POC1 (BottomField) Layer1
- POC1 (BottomField) Layer2

Decoder needs to search for a second field picture.
It should belong to the same layer as the first field picture.

Also, POCs of first and second field are not necessary differ by one. So, I removed those checks

Changes include:
- use [[maybe_unused]] instead of cast to (void)
- use generic lambda function and remove related comment about C++11
- update MSVC version check
- don't condition '#pragma once' on MSVC version

Kirill Suverov authored 2 years ago and Nikolay Shostak committed 2 years ago

When decoder opens a file for reconstructed pictures of a particular layer
it should get bit-depth from first picture of the same layer

Kirill Suverov authored 2 years ago and Nikolay Shostak committed 2 years ago

PPS can be destroyed when picture is to be output.
So, instead of accessing pointer to PPS,
decoder should get conformance window from Picture object

Ticket #1563

Ticket #1554

Not final patch. TODO:
1) Do the same in few other similar places
2) Deal with "--upscaledOutput" case which uses even more fields from
potentially deleted SPS and PPS