-
Notifications
You must be signed in to change notification settings - Fork 0
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Ad Insertion content - Testcase 1 - Aligned content and period boundaries #1
Comments
TaC 2021/03/02:
|
The proposed content is 24Hz. I'd suggest to convert to 25Hz in order to get same frame rate as used for live including the nx1.92s segment duration. |
@jpiesing this is This topic still gives good details about the alignments, such as video(h264) an audio(aac) segments to have an equal perfect segment durations. This is something not easy to get with an integer 2,3,4,.. sec segments for aac codec. |
Here are some streams to start with:
Only the synthetic counter version plays well in dash.js at the moment. |
Is it a mistake to have Can you provide your ffmpeg+mp4box command lines for creating this content such as CMAF and 1.92s segmenting? |
Yes, thanks for reporting!
See here https://github.com/Dash-Industry-Forum/Test-Content/blob/e31b16f3fd188b00988de2906910f40c23fb98e2/ad-insertion-testcase1.sh. I've just created the PR. NB1: it uses one-liner commands using GPAC only (wherever possible). NB2: the refresh time of the server is quite long. So fixes may take time to apply. |
Comment from today's call: favour this client version: https://reference.dashif.org/dash.js/nightly/samples/dash-if-reference-player/index.html |
I did a check in dash.js and Shaka, please find the test results and observations here: |
We have done this multiperiod(main content, adverts, static mpd) a little over a year ago. We also test We are compiling a similar scenario for a dynamic(live) test content. |
@dsilhavy the @MediaPresentationDuration was not referenced in the document but is fixed. Still working on the other items. |
@dsilhavy I checked the duration of the periods and they seem ok. To dump the mpd as a file I used I'm going to generate a new batch of files soon. The other mandatory points should be fixed so I generated a batch3: |
@rbouqueau Thanks for the new vectors, I added a second tab to the Spreadsheet: https://docs.google.com/spreadsheets/d/1OLRzskFmO5yL_kJpF6cILrO_uy-USrTmvsWZHEgqgtw/edit#gid=498241230 Overall it looks good, I see a minor problem in dash.js having a small gap in the end for all testvectors. https://dash.akamaized.net/dashif/ad-insertion-testcase1/batch4/real/a/ad-insertion-testcase1.mpd has a major issue between period 2 and 3 in both dash.js and Shaka. Might be related to differences between PTO and tfdt->bmdt in the audio segments. |
I think I fixed the real/a content: https://dash.akamaized.net/dashif/ad-insertion-testcase1/batch5/real/a/ad-insertion-testcase1.mpd To summarize, since the AUs boundaries are exactly where we cut the Periods, working in the encoded or raw domain shouldn't make any difference. But it does so I switched to raw. I also applied the same patch in an attempt to fix the |
@rbouqueau Thanks for the feedback. Regarding gap jumping: Jumping to end of stream because of gap from 28.677133 to 28.8. Gap duration: 0.1228669999999994 The problem was that we did not call MSE.endOfStream if the last period was prebuffered once the previous period is still playing. This fixed here Dash-Industry-Forum/dash.js@a1297a2 . Changes should be visible in nightly. |
We can remove the following labels after comments in the CTA-WAVE DTCTF call today:
Once I get access to the appropriate information on the Interoperability WG I assume I'll have to regenerate the vectors. |
Test content is good - needs to be published and announced, please add to data base. |
Added to database: https://testassets.dashif.org/#testcase/details/61681dfa06d983046b7d4140 |
Test Call 2021/11/12:
|
This one is available - announce to DASH-IF IOP |
A short description of the use case for the new test content.
This test content is supposed to demonstrate the insertion of a midroll ad in a static MPD. In this testcase the segment- and period boundaries are perfectly aligned (as shown in the Figure below).
Test content requirements
SegmentTemplate
with$number$
SegmentTimeline
with$number$
or$time$
SegmentBase
Relation to CTA WAVE Test Vector database
The text was updated successfully, but these errors were encountered: