HEVC test streams

[jpiesing]

This issue is to track the HEVC test stream generation.

We have a sparse matrix from @nicholas-fr at https://docs.google.com/spreadsheets/d/1Bmgv6-cfbWfgwn7l-z0McUUI1rMjaWEwrN_Q30jaWk4/edit?usp=sharing.

[jpiesing]

As far as I know, the next step is for @nicholas-fr to update the sparse matrix so it correctly reflects the conclusion in the DPCTF meeting on February 14th - "Mr. Frame will remove the debug profiles for everything except and remove all the non-square pixels."

[rbouqueau]

@nicholas-fr Please keep track of the modifications since the first round of HEVC content is based on the same sparse matrix document (cf README). Thanks!

[jpiesing]

@nicholas-fr Please keep track of the modifications since the first round of HEVC content is based on the same sparse matrix document (cf README). Thanks!

The modifications are as follows ...

Question 1: What to do about the 20 debug streams (i.e. 5 streams for each of the 4 media profiles)? Should we keep them? Drop them or just keep them for one media profile?

Conclusion: Keep them for chh1 but drop them for the other 3 media profiles. i.e. keep columns H to L inclusive but drop X to AB, AN to AR and BE to BI.

Question 2: What about 1440x1080? i.e. columns P, AF, AW and BN

Conclusion: delete all 4 columns listed.

[nicholas-fr]

Thanks for the summary of the conclusions, I've applied the updates to the HEVC sparse matrix. The only change that impacts the initial HEVC content is the removal of column P (non-square pixel aspect ratio for chh1 profile).

[nlsdvl]

Hello @nicholas-fr

I've been generating the chh1 and chd1 content related to this issue based on the available HEVC matrix.

I picked up the following issues while working on it:

CHH1

• t1_splice_ad / t1_splice_main : the HEVC matrix lists A1 and AD-A1 as mezzanine labels. The resolution of these files doesn't match, I used B1 and AD-B1 instead.

CHD1

• I couldn't find any information about what mezzanine content should be used for 'non_square_par' test vector generation. As I understand, the above comment, the non_square_par variant has been dropped, so that would be expected.
• There is currently no mezzanine content for the 14.985/29.97/59.94 frame rates. Are we looking for one ?

CUD1

• t21, t22 : the matrix refers to 30s duration, however mezzanine content is only available for 60s. I was expecting the mezzanine to match the target duration as found in previous work, so I did not yet process it.
• t31, t32 : the matrix refers to 0.5 framerate factor (eg. 30fps, 29.97fps, or 25fps), however mezzanine content is only available for 60fps, 59.94fps, 50fps. Again, my understanding is based on analysis of previous work, I was expecting the mezzanine to be supplied in the target framerate.
• The requirements document I have make mention of bt2020 colorspace, however no such mezzanine content could be found.

I will process the missing items as soon as I can sort these issues out.

Uploaded content:

Here's the list of content I uploaded already:

CHH1

chh1_sets/14.985_29.97_59.94/t1/2024-10-25/t1.zip
chh1_sets/12.5_25_50/t1/2024-10-25/t1.zip
chh1_sets/15_30_60/t1/2024-10-25/t1.zip
chh1_sets/14.985_29.97_59.94/t1-cenc/2024-10-25/t1-cenc.zip
chh1_sets/12.5_25_50/t1-cenc/2024-10-25/t1-cenc.zip
chh1_sets/15_30_60/t1-cenc/2024-10-25/t1-cenc.zip
chh1_sets/14.985_29.97_59.94/t2/2024-10-25/t2.zip
chh1_sets/12.5_25_50/t2/2024-10-25/t2.zip
chh1_sets/15_30_60/t2/2024-10-25/t2.zip
chh1_sets/14.985_29.97_59.94/t3/2024-10-25/t3.zip
chh1_sets/12.5_25_50/t3/2024-10-25/t3.zip
chh1_sets/15_30_60/t3/2024-10-25/t3.zip
chh1_sets/14.985_29.97_59.94/t4/2024-10-25/t4.zip
chh1_sets/12.5_25_50/t4/2024-10-25/t4.zip
chh1_sets/15_30_60/t4/2024-10-25/t4.zip
chh1_sets/14.985_29.97_59.94/t5/2024-10-25/t5.zip
chh1_sets/12.5_25_50/t5/2024-10-25/t5.zip
chh1_sets/15_30_60/t5/2024-10-25/t5.zip
chh1_sets/14.985_29.97_59.94/t6/2024-10-25/t6.zip
chh1_sets/12.5_25_50/t6/2024-10-25/t6.zip
chh1_sets/15_30_60/t6/2024-10-25/t6.zip
chh1_sets/14.985_29.97_59.94/t7/2024-10-25/t7.zip
chh1_sets/12.5_25_50/t7/2024-10-25/t7.zip
chh1_sets/15_30_60/t7/2024-10-25/t7.zip
chh1_sets/14.985_29.97_59.94/t10/2024-10-25/t10.zip
chh1_sets/12.5_25_50/t10/2024-10-25/t10.zip
chh1_sets/15_30_60/t10/2024-10-25/t10.zip
chh1_sets/14.985_29.97_59.94/t11/2024-10-25/t11.zip
chh1_sets/12.5_25_50/t11/2024-10-25/t11.zip
chh1_sets/15_30_60/t11/2024-10-25/t11.zip
chh1_sets/14.985_29.97_59.94/t12/2024-10-25/t12.zip
chh1_sets/12.5_25_50/t12/2024-10-25/t12.zip
chh1_sets/15_30_60/t12/2024-10-25/t12.zip
chh1_sets/14.985_29.97_59.94/splice_ad/2024-10-25/splice_ad.zip
chh1_sets/12.5_25_50/splice_ad/2024-10-25/splice_ad.zip
chh1_sets/15_30_60/splice_ad/2024-10-25/splice_ad.zip
chh1_sets/14.985_29.97_59.94/splice_ad-cenc/2024-10-25/splice_ad-cenc.zip
chh1_sets/12.5_25_50/splice_ad-cenc/2024-10-25/splice_ad-cenc.zip
chh1_sets/15_30_60/splice_ad-cenc/2024-10-25/splice_ad-cenc.zip
chh1_sets/14.985_29.97_59.94/splice_main/2024-10-25/splice_main.zip
chh1_sets/12.5_25_50/splice_main/2024-10-25/splice_main.zip
chh1_sets/15_30_60/splice_main/2024-10-25/splice_main.zip
chh1_sets/14.985_29.97_59.94/splice_main-cenc/2024-10-25/splice_main-cenc.zip
chh1_sets/12.5_25_50/splice_main-cenc/2024-10-25/splice_main-cenc.zip
chh1_sets/15_30_60/splice_main-cenc/2024-10-25/splice_main-cenc.zip
chh1_sets/14.985_29.97_59.94/non_square_par/2024-10-25/non_square_par.zip
chh1_sets/12.5_25_50/non_square_par/2024-10-25/non_square_par.zip
chh1_sets/15_30_60/non_square_par/2024-10-25/non_square_par.zip

CHD1

chd1_sets/12.5_25_50/t61/2024-10-25/t61.zip
chd1_sets/15_30_60/t61/2024-10-25/t61.zip
chd1_sets/12.5_25_50/t62/2024-10-25/t62.zip
chd1_sets/15_30_60/t62/2024-10-25/t62.zip
chd1_sets/12.5_25_50/t70/2024-10-25/t70.zip
chd1_sets/15_30_60/t70/2024-10-25/t70.zip
chd1_sets/12.5_25_50/t71/2024-10-25/t71.zip
chd1_sets/15_30_60/t71/2024-10-25/t71.zip
chd1_sets/12.5_25_50/t72/2024-10-25/t72.zip
chd1_sets/15_30_60/t72/2024-10-25/t72.zip

[nicholas-fr]

Hello @nlsdvl

Thanks for your feedback. My apologies for the wait, please find my responss below. In summary though, I agree there are multiple missing mezzanine streams that I'll try to generate and upload by the end of the week.

CHH1

• t1_splice_ad / t1_splice_main : the HEVC matrix lists A1 and AD-A1 as mezzanine labels. The resolution of these files doesn't match, I used B1 and AD-B1 instead.

This makes sense. I updated the labels to B1/AD-B1 in the HEVC spreadsheet. I see splice mezzanine missing for 3840x2160 resolutions, I've updated the label for those to C1/AD-C1. I will create the missing streams.

CHD1

• I couldn't find any information about what mezzanine content should be used for 'non_square_par' test vector generation. As I understand, the above comment, the non_square_par variant has been dropped, so that would be expected.

Yes exactly, it was agreed not to generate non square pixel aspect ratio test content.

• There is currently no mezzanine content for the 14.985/29.97/59.94 frame rates. Are we looking for one ?

Referring to the Sol Levante HDR10 mezzanine content here, indeed fractional versions of the same mezzanine videos are to be created. I will generate them.

CUD1

• t21, t22 : the matrix refers to 30s duration, however mezzanine content is only available for 60s. I was expecting the mezzanine to match the target duration as found in previous work, so I did not yet process it.

Thanks, 30s variants are indeed missing from the UHD SDR in mezzanine release v4. I will generate them.

• t31, t32 : the matrix refers to 0.5 framerate factor (eg. 30fps, 29.97fps, or 25fps), however mezzanine content is only available for 60fps, 59.94fps, 50fps. Again, my understanding is based on analysis of previous work, I was expecting the mezzanine to be supplied in the target framerate.

Thanks, 0.5 variants are indeed missing from the UHD SDR in mezzanine release v4. I will generate them.

• The requirements document I have make mention of bt2020 colorspace, however no such mezzanine content could be found.

I don't believe we decided to excplicitly use BT.2020 content. If we do, then I can propose/generate new BT.2020 SDR mezzanine for the CUD1 tests.

[nicholas-fr]

As just discussed in the DPCTF test coordination call today, it was agreed to use BT.2020 content for CUD1 test vectors. I will also create new mezzanine for that.