generate tests for HEVC

[jpiesing]

I think this is the correct sparse matrix for HEVC.

https://docs.google.com/spreadsheets/d/1Bmgv6-cfbWfgwn7l-z0McUUI1rMjaWEwrN_Q30jaWk4/edit?usp=sharing

[louaybassbouss]

@nlsdvl is database.json now up-to-date with newly generated HEVC content? If yes, we will generate the HEVC tests and provide them in staging branch for testing.

[nlsdvl]

The pull request with the latest changes to the HEVC content had been reviewed but was not merged. cta-wave/Test-Content#70 - now it is. @louaybassbouss please pull the latest changes from the master branch.

This database update also fixes issues/63 on AVC switching set. Only the mpd was modified, content wasn't re-encoded/re-packaged.

[jpiesing]

There was a comment from @louaybassbouss in a previous DPCTF meeting that there was some kind of disconnect between the names of the HEVC streams in the sparse matrix and in database.json.

[nlsdvl]

was some kind of disconnect between the names of the HEVC streams in the sparse matrix and in database.json

@louaybassbouss - the issue wasn't brought to my attention until now, can you provide details on how the db mismatches the sparse matrix ?

[nlsdvl]

@rbouqueau reported inconsistency between AVC and HEVC database keys for splicing, but that as far as I understand, that is not related to the sparse matrix.

[louaybassbouss]

was some kind of disconnect between the names of the HEVC streams in the sparse matrix and in database.json

@louaybassbouss - the issue wasn't brought to my attention until now, can you provide details on how the db mismatches the sparse matrix ?

maybe I miss something, can you point me in database.json for example to HEVC streams 50, 51 and 52 from the sparse matrix

[rbouqueau]

@louaybassbouss Did you search for "t50", "t51", and "t52" in https://github.com/cta-wave/Test-Content/blob/master/database.json ? Which method did you use?

[louaybassbouss]

@rbouqueau yes I searched for the streams in database.json and I found results created in January for example clg1_sets/12.5_25_50/t50/2025-01-15/ this is why I was not sure if this is the right entry or not, I just need your confirmation if it is the case.

[FritzHeiden]

While generating the tests I noticed that chunked tests (e.g. playback of chunked content) use t2, t22, t42, t62 content. In the database.json there is also HEVC content labelled as "chunked" (instead of t2 etc.). What is the difference there?

@rbouqueau yes I searched for the streams in database.json and I found results created in January for example clg1_sets/12.5_25_50/t50/2025-01-15/ this is why I was not sure if this is the right entry or not, I just need your confirmation if it is the case.

This is resolved by the "CMAF media profile" row in the sparse matrix, which helps matching the content in the database.json.

[rbouqueau]

It is the same content, also generated as "chunked" (per the spec). Same applied in the AVC sparse matrix. Does that make sense?

Maybe we could improve the naming?

[FritzHeiden]

It is the same content, also generated as "chunked" (per the spec). Same applied in the AVC sparse matrix. Does that make sense?

Maybe we could improve the naming?

Thanks for clarifying. I think this is fine.

[FritzHeiden]

The new HEVC tests are now available in the staging branch.

[nlsdvl]

I searched for the streams in database.json and I found results created in January for example clg1_sets/12.5_25_50/t50/2025-01-15/ this is why I was not sure if this is the right entry or not, I just need your confirmation if it is the case.

@louaybassbouss indeed, t50 was generated in January. Some HEVC test vectors have been updated after that, just not this one. I tried to follow the naming scheme used by existing content, so the number from the sparse matrix are prefixed with a t.

[louaybassbouss]

Thanks @nlsdvl for clarification now everything is clear and the HEVC tests are generated and ready to be tested in the dpctf-deploy@staging branch. We will some tests during next days since we are going to showcase this at IBC (@jpiesing FYI)