EAI-708 support #1807
Comments
|
@cdrolle is looking at 708 support. Could you comment on the above? Thanks! |
|
Any news? 🐔 |
|
🍯 |
|
Any comments about your implementation plans? I have the 708 parser implemented, I would gladly contribute if our code changes can be aligned. (As my V1 parsing is not compatible with the current state of V2) |
|
We have some 708 code internally that hasn't been pushed to GitHub yet. @cdrolle - Can you make it happen? Thanks. |
Issue: #1807 ------------- Created by MOE: https://github.com/google/moe MOE_MIGRATED_REVID=144726542
|
The change above adds what we've done so far, but it's not yet a complete solution. Some further work is needed on the MediaSource/Extraction side to expose the CEA-708 tracks so that they can be selected for rendering. |
|
What's the current state of getting the github version up to date with the 708 changes? |
|
Hi. Has there been any further movement on this 708 support? |
|
Last I have looked 708 decoder is found: https://github.com/google/ExoPlayer/blob/release-v2/library/core/src/main/java/com/google/android/exoplayer2/text/cea/Cea608Decoder.java |
|
@roticv That is for 608, not 708. I think there are a few pull requests pending for quite improved 608 and 708 support. Unfortunately, it seems they wont be merged in this form as reviewing them is quite difficult.
Reworking the pull requests into smaller digestible changes will take some time. |
|
@ojw28 is there any plan to add full support for CEA-708. We have requirement to enable CEA-708. It would be helpful if this gets some priority. @zsmatyas we have tried Sarnoff streams with #4595 changes. It works much better with that. But still there are many outstanding issues. We are happy to help in #4595 merge (and further improvement). Please let us know. Thanks. |
|
We don't have capacity to work on this any time soon. We would welcome high quality external contributions. In terms of what that means, I think we'd be looking for:
|
In this commit - Handling the sequence number discontinuity in caption channel packet header. Handle the multiple Service Blocks within a single caption channel packet. An outer while loop is added to function processCurrentPacket() to achieve that. The processCurrentPacket returns if the packet length does not match with the currentIndex. That assumption is wrong. As per spec the the packet can end on reception of next cc_type = 0x3. Added a logic to handle multi byte commands. These commands can be within the service block or spread across service blocks. Processing of service block needs to wait until all the command bytes arrive. An ArrayList is used to accumulate the bytes belonging to same service block. In this change the current flow is not altered, i.e. the parsing of service block is not started until the complete caption channel packet arrives. These changes are verified through Sarnoff Test streams. Many tests are still failing after this change. Subsequent changes are planned to address that.
Fixed the verticalAnchorType and horizontalAnchorType calculation The anchorID 0, 1 and 2 should correspond to verticalAnchorType=ANCHOR_TYPE_START, anchorID 3, 4, 5 is ANCHOR_TYPE_MIDDLE, anchorID 6, 7 and 8 is ANCHOR_TYPE_END The anchorID 0, 3 and 6 should correspond to horizzonatlAnchor=ANCHOR_TYPE_START, anchorID 1, 4, 7 is ANCHOR_TYPE_MIDDLE, anchorID 2, 5 and 8 is ANCHOR_TYPE_END
|
@ojw28 I have sent the test stream links over developer email. Please have a look at them. |
In this commit - Handling the sequence number discontinuity in caption channel packet header. Handle the multiple Service Blocks within a single caption channel packet. An outer while loop is added to function processCurrentPacket() to achieve that. The processCurrentPacket returns if the packet length does not match with the currentIndex. That assumption is wrong. As per spec the the packet can end on reception of next cc_type = 0x3. Added a logic to handle multi byte commands. These commands can be within the service block or spread across service blocks. Processing of service block needs to wait until all the command bytes arrive. An ArrayList is used to accumulate the bytes belonging to same service block. In this change the current flow is not altered, i.e. the parsing of service block is not started until the complete caption channel packet arrives. These changes are verified through Sarnoff Test streams. Many tests are still failing after this change. Subsequent changes are planned to address that. This is a cherry-pick of one commit from a pending pull request: google#8356 And the commit was: google@93829f9 The other two commits in the pull request are already included, cherry-picked from Google's 1.15.1 The conflicts resolved are: * use previousSequenceNumber over lastSequenceNumber * take Google's spacing choices * Google's code for the anchorId to Cue.ANCHOR_TYPE to match 2.15.1
Fix requires changes to the merged pull request google#8415 that are still part of the unmerged pull request google#8356 In this commit - Handling the sequence number discontinuity in caption channel packet header. Handle the multiple Service Blocks within a single caption channel packet. An outer while loop is added to function processCurrentPacket() to achieve that. The processCurrentPacket returns if the packet length does not match with the currentIndex. That assumption is wrong. As per spec the the packet can end on reception of next cc_type = 0x3. Added a logic to handle multi byte commands. These commands can be within the service block or spread across service blocks. Processing of service block needs to wait until all the command bytes arrive. An ArrayList is used to accumulate the bytes belonging to same service block. In this change the current flow is not altered, i.e. the parsing of service block is not started until the complete caption channel packet arrives. These changes are verified through Sarnoff Test streams. Many tests are still failing after this change. Subsequent changes are planned to address that. # Please enter the commit message for your changes. Lines starting # with '#' will be ignored, and an empty message aborts the commit. # # Author: sneelavara <sneelavara@gmail.com> # Date: Wed Dec 16 13:03:37 2020 -0800 # # On branch t-fix-cea708Merge # Your branch and 'tivo-pvt/t-fix-cea708Merge' have diverged, # and have 1 and 2 different commits each, respectively. # (use "git pull" to merge the remote branch into yours) # # Changes to be committed: # modified: library/core/src/main/java/com/google/android/exoplayer2/text/cea/Cea708Decoder.java # # Untracked files: # CCURStream_3000kbps720p29_97fps-1_1661286386_init.cmfv?ccur_keyrot_t=1666096234 # CCURStream_3000kbps720p29_97fps-1_T1666098015585589~D6006000.cmfv # MERGE_COMMIT_MESSAGE # VelocixVTP.mp4 # X # test_url # tivo-docs/DASH Encoding Guidelines.md # tivo-docs/Encoding HLS Renditions.pdf # tivo-docs/HANDLING_2.15.1_MERGE_CONFLICTS.pdf # tivo-docs/Logging Updates For Audio Position.md # tivo-docs/Logging Updates For Audio Position.pdf # tivo-docs/TiVoExoPlayerReleaseMapping.md # tivo-docs/TiVoExoPlayerReleaseMapping.pdf # tivo-docs/Understanding ExoPlayer Position Tracking.md # tivo-docs/Understanding ExoPlayer Position Tracking.pdf # widevine.cmfv #
Previous version of the v1 and v2-dev branches had the entire SEI data put into a buffer and then read by the Eia608Parser. The parser had a check that the type and validity bits are set to channel 1 of the 608 content, and everything else was dropped. See: V1
if (ccType != 0) { seiBuffer.skipBits(16); continue; }Here the ccType values mean:
As only value 0 was handled, and no additional checks were used to separate channel 1 and 2: a single 608 channel was supported. I added my additional channel and EIA-708 support by updating the Eia608Parser to collect the data instead of dropping it, so channel 3-4 and EIA-708 bits were collected and parsed appropriately.
In the latest v2 code, all these additional data are dropped at SeiReader level:
V2
like:
if (ccValidityAndType != 0x04) { seiBuffer.skipBytes(2); }The change was: Link
SeiReader seems to be the correct location to filter 608/708 and channel information, but needs serious changes in the architecture I had. Are you guys planning to ever support EIA-708 (as it is an FCC requirement from 2017)? Did you consider supporting the other channels of EIA-608? Any similar changes like the one you made in the v2-dev branch requires lots of changes to support all subtitles correctly, that is why I am interested in the plans about supported formats.
Are you willing to accept pull requests related to these features or are you maybe already working on these issues?
Big Thanks!
The text was updated successfully, but these errors were encountered: