madshi bug tracker - eac3to
View Issue Details
0000512eac3tobugpublic2017-10-15 08:042017-11-19 04:46
msg7086 
madshi 
normalminoralways
assignedopen 
Windows 10
3.32
0000512: Exit silently when scanning capped HEVC 4k in TS files
The video in question is a commercial capped from a 4k TV station, broadcasting free of charge at that moment.

When scanning this file using eac3to 3.32, it exits silently after a fraction of second.

eac3to 3.31 was working partially since it does not try to parse HEVC stream, and at least the audio can be extracted correctly (without correct audio delay detected, of course).

The same issue applies to all similar video clips, HEVC+AAC inside TS files.
1. eac3to 4k-test.ts
Also expect eac3to to detect the correct audio delay so we don't have to manually sync video with audio.
No tags attached.
? 4k-test.ts (8,226,348) 2017-10-15 08:08
http://bugs.madshi.net/file_download.php?file_id=260&type=bug
? 4k-test2.ts (8,213,080) 2017-10-15 20:39
http://bugs.madshi.net/file_download.php?file_id=261&type=bug
Issue History
2017-10-15 08:04msg7086New Issue
2017-10-15 08:08msg7086File Added: 4k-test.ts
2017-10-15 20:33madshiNote Added: 0001850
2017-10-15 20:39msg7086File Added: 4k-test2.ts
2017-10-15 20:42msg7086Note Added: 0001851
2017-11-15 15:02madshiNote Added: 0001920
2017-11-15 20:46msg7086Note Added: 0001921
2017-11-15 20:46msg7086Note Edited: 0001921bug_revision_view_page.php?bugnote_id=1921#r477
2017-11-18 21:55madshiNote Added: 0001928
2017-11-18 21:55madshiAssigned To => madshi
2017-11-18 21:55madshiStatusnew => feedback
2017-11-19 04:11msg7086Note Added: 0001930
2017-11-19 04:11msg7086Statusfeedback => assigned
2017-11-19 04:46msg7086Note Added: 0001931

Notes
(0001850)
madshi   
2017-10-15 20:33   
With that TS file eac3to reports "The format of the source file could not be detected" on my PC. Maybe the file is too short?
(0001851)
msg7086   
2017-10-15 20:42   
My apologies, when cutting the file to get around the size limitation I seemed to cut in a weird position.

Double checked this time and uploaded a new sample.

Thank you for your time.
(0001920)
madshi   
2017-11-15 15:02   
Is it fixed in this build?

http://madshi.net/eac3toHevcTest.rar [^]
(0001921)
msg7086   
2017-11-15 20:46   
Testing 4k-test2.ts
----------
Testing 01.ts
-
Testing 02.ts

Testing 03.ts
-
Testing 04.ts

Testing 05.ts

Testing 06.ts

Testing 07.ts

Testing 08.ts
-

9 negatives. log.txt file shows nothing useful.

(0001928)
madshi   
2017-11-18 21:55   
Hopefully fixed in v3.33?

http://madshi.net/eac3to.zip [^]

The attached file was too short to properly detect the HEVC properties, but at least it doesn't exit silently, anymore.
(0001930)
msg7086   
2017-11-19 04:11   
Testing 4k-test2.ts
TS, 1 video track, 1 audio track, 0:00:03, 30p /1.001
1: h265/HEVC, unknown parameters
2: AAC, 2.0 channels, 185kbps, 48kHz, -740ms

Testing 01.ts
TS, 1 video track, 1 audio track, 0:30:21, 30p /1.001
1: h265/HEVC, unknown parameters
2: AAC, 2.0 channels, 185kbps, 48kHz, -1063ms

Testing 02.ts
TS, 1 video track, 1 audio track, 0:30:21, 30p /1.001
1: h265/HEVC, unknown parameters
2: AAC, 2.0 channels, 179kbps, 48kHz, -652ms

Testing 03.ts
TS, 1 video track, 1 audio track, 0:30:09, 30p /1.001
1: h265/HEVC, unknown parameters
2: AAC, 2.0 channels, 185kbps, 48kHz, -1040ms

Testing 04.ts
TS, 1 video track, 1 audio track, 0:30:27, 30p /1.001
1: h265/HEVC, unknown parameters
2: AAC, 2.0 channels, 191kbps, 48kHz, -601ms

Testing 05.ts
TS, 1 video track, 1 audio track, 0:30:37, 30p /1.001
1: h265/HEVC, unknown parameters
2: AAC, 2.0 channels, 191kbps, 48kHz, -713ms

Testing 06.ts
TS, 1 video track, 1 audio track, 0:30:23, 30p /1.001
1: h265/HEVC, unknown parameters
2: AAC, 2.0 channels, 192kbps, 48kHz, -613ms

Testing 07.ts
TS, 1 video track, 1 audio track, 0:30:15, 30p /1.001
1: h265/HEVC, unknown parameters
2: AAC, 2.0 channels, 186kbps, 48kHz, -570ms

Testing 08.ts
TS, 1 video track, 1 audio track, 0:30:17, 30p /1.001
1: h265/HEVC, unknown parameters
2: AAC, 2.0 channels, 191kbps, 48kHz, -882ms

I'll upload a longer example for further testing. Also I'll report if the audio delay is correct after comparing with my previous test.
(0001931)
msg7086   
2017-11-19 04:46   
https://drive.google.com/file/d/1vQUNV32Wu64jaqjNZYsAMLxmbRzje7q6/view?usp=sharing [^]

Should be long enough for HEVC properties.

Regarding the audio delay, here's what I ear-detected against videos encoded with DGNV avisynth source filter.

01.ts: eac3to reports -1063ms, ear detected -700ms.
02.ts: eac3to reports -652ms, ear detected -600ms.
03.ts: eac3to reports -1040ms, ear detected -600ms.
04.ts: eac3to reports -601ms, ear detected -650ms.
05.ts: eac3to reports -713ms, ear detected -750ms.
06.ts: eac3to reports -613ms, ear detected -600ms.
07.ts: eac3to reports -570ms, ear detected -450ms.
08.ts: eac3to reports -882ms, ear detected -850ms.

Since those video clips are just landscapes with music, no accurate sync point can be found. And we wouldn't know how DGNV determines its first usable frame either, so this is for reference only.