VSE Linked external scene with audio with input Sequencer only works if the scene is NOT also linked with input Camera #69444

New Issue

gabriel montagné · 2019-09-03T13:22:16+02:00

gabriel montagné commented

2019-09-03 13:22:16 +02:00

System Information
Operating system: Linux-4.9.0-9-amd64-x86_64-with-debian-9.9 64 Bits
Graphics card: Mesa DRI Intel(R) HD Graphics 520 (Skylake GT2) Intel Open Source Technology Center 4.5 (Core Profile) Mesa 13.0.6

Blender Version
Broken: version: 2.81 (sub 3), branch: master, commit date: 2019-08-20 22:35, hash: daba3e871f
Worked: Haven't yet tried other versions.

Short description of error
VSE: Linking external scene with audio with input Sequencer only works if the scene is NOT also linked with input Camera.

Exact steps for others to reproduce the error

Create file A with Scene-A and add audio to it on the VSE.
Create file B
From the VSE in file B, link the Scene-A from file A. Set it's input to "Sequencer". On file B, playing the strip will correctly play the sound from Scene-A.
Duplicate (or re-instance) Scene-A on the VSE sequencer. If the Input is still "Sequencer", both clips will play the audio, as expected.
Switch one of the Scene-A instances to use the Camera as input.

Expected behaviour:

We should be able to both see the Camera strip and hear the Sequence strip for Scene-A

Actual behaviour:

For all the strips of Scene-A that are set to Input "Sequencer", the audio will stop playing.

Note: I've determined that it is not that the audio crashes: other clips with sound on the sequencer keep playing normally.

Note: If I delete the strip with Input Camera and leave only the input Sequencer strips, the problem will not go away, unless I save the file like this and "revert" to it.

I cannot find a suitable work-around. In the video that I'm producing I need to edit local sound effects on the shot sequence, and incorporate it on the larger edit.
I cannot, therefore, both see and hear my sequence on the host sequencer.

If a workaround is possible (that is not reconstruct and re-import the sounds on the original scene) I'd be a happy man :-)
Thanks!

**System Information** Operating system: Linux-4.9.0-9-amd64-x86_64-with-debian-9.9 64 Bits Graphics card: Mesa DRI Intel(R) HD Graphics 520 (Skylake GT2) Intel Open Source Technology Center 4.5 (Core Profile) Mesa 13.0.6 **Blender Version** Broken: version: 2.81 (sub 3), branch: master, commit date: 2019-08-20 22:35, hash: `daba3e871f` Worked: Haven't yet tried other versions. **Short description of error** VSE: Linking external scene with audio with input Sequencer only works if the scene is NOT also linked with input Camera. **Exact steps for others to reproduce the error** 1. Create file A with Scene-A and add audio to it on the VSE. 2. Create file B 3. From the VSE in file B, link the Scene-A from file A. Set it's input to "Sequencer". On file B, playing the strip will correctly play the sound from Scene-A. 3. Duplicate (or re-instance) Scene-A on the VSE sequencer. If the Input is still "Sequencer", both clips will play the audio, as expected. 4. Switch one of the Scene-A instances to use the Camera as input. Expected behaviour: We should be able to both see the Camera strip and hear the Sequence strip for Scene-A Actual behaviour: For all the strips of Scene-A that are set to Input "Sequencer", the audio will stop playing. Note: I've determined that it is not that the audio crashes: other clips with sound on the sequencer keep playing normally. Note: If I delete the strip with Input Camera and leave only the input Sequencer strips, the problem will not go away, unless I save the file like this and "revert" to it. I cannot find a suitable work-around. In the video that I'm producing I need to edit local sound effects on the shot sequence, and incorporate it on the larger edit. I cannot, therefore, both see and hear my sequence on the host sequencer. If a workaround is possible (that is not reconstruct and re-import the sounds on the original scene) I'd be a happy man :-) Thanks!

gabriel montagné commented

2019-09-03 13:22:16 +02:00

Added subscriber: @GabrielMontagne

blender-admin commented

2019-09-03 13:22:16 +02:00

#88319 was marked as duplicate of this issue

blender-admin commented

2019-09-03 13:22:16 +02:00

#66117 was marked as duplicate of this issue

Richard Antalik was assigned by Sybren A. Stüvel

2020-01-16 16:53:58 +01:00

Brecht Van Lommel commented

2020-02-03 10:27:55 +01:00

Added subscriber: @brecht

Brecht Van Lommel commented

2020-02-03 10:27:55 +01:00

I'm removing the 2.82 tag from this bug, since I don't think it's a showstopper and it was already tagged for 2.81 and not fixed.

Richard Antalik commented

2020-02-04 12:08:13 +01:00

Added subscribers: @tintwotin, @iss, @jaker-3, @Sergey, @neXyon

gabriel montagné commented

2020-03-03 18:12:41 +01:00

Hello dear devs, how can one see where in the priority stack does this ticket float?
If there was a mechanism for "for tickets with the same depth on the back-log, click here -> paypal -> dev fund to give it a bit more weight" I'd happily use it for this one. : )
Thanks!

Hello dear devs, how can one see where in the priority stack does this ticket float? If there was a mechanism for "for tickets with the same depth on the back-log, click here -> paypal -> dev fund to give it a bit more weight" I'd happily use it for this one. : ) Thanks!

Richard Antalik commented

2020-03-04 02:56:30 +01:00

@GabrielMontagne I haven't got time to work on this in past months, now I do and there are only 4 high priority bugs, so this should be fixed in 2.83

There is quick hack attached so you can build it if you can, I can make only windows builds. or if you need help you can visit https://blender.chat vse room and we can work out some temporary solution.

@GabrielMontagne I haven't got time to work on this in past months, now I do and there are only 4 high priority bugs, so this should be fixed in 2.83 There is quick hack attached so you can build it if you can, I can make only windows builds. or if you need help you can visit https://blender.chat vse room and we can work out some temporary solution.

gabriel montagné commented

2020-03-04 11:01:31 +01:00

In #69444#884676, @iss wrote:
[...] so this should be fixed in 2.83

Awesome sauce : ) that's great news!

In #69444#884676, @iss wrote:
[...] There is quick hack attached so you can build it if you can [...]

True! I'll see if I can build it from the hack, that'd be great.

Thanks again for the update, and thanks for the help. Cheers.

> In #69444#884676, @iss wrote: > [...] so this should be fixed in 2.83 Awesome sauce : ) that's great news! > In #69444#884676, @iss wrote: > [...] There is quick hack attached so you can build it if you can [...] True! I'll see if I can build it from the hack, that'd be great. Thanks again for the update, and thanks for the help. Cheers.

gabriel montagné commented

2020-03-05 10:27:42 +01:00

Thanks :-) I built with your patch and it worked really nice for my setup. I left a note on the patch conversation as well. Thanks.

Richard Antalik commented

2020-04-20 13:31:30 +02:00

I will lower priority here, because this requires some design consideration and ideally testing time. So it is not really good idea to work on this in BCON3 stage.

I guess I could keep high priority as this issue was introduced relatively recently, but I think that original design wasn't really thought through to begin with, so there isn't really any known good state to try to achieve.

I will lower priority here, because this requires some design consideration and ideally testing time. So it is not really good idea to work on this in BCON3 stage. I guess I could keep high priority as this issue was introduced relatively recently, but I think that original design wasn't really thought through to begin with, so there isn't really any known good state to try to achieve.

gabriel montagné commented

2020-04-20 18:34:06 +02:00

Could an interim solution be just letting the sound through for both Camera and Sequencer? At least we could use it until the redesign of how these should be linked is fleshed out?
I was very happy to see the bcon3 tag on this one, even if just for those glorious 3 days : )

Could an interim solution be just letting the sound through for both Camera and Sequencer? At least we could use it until the redesign of how these should be linked is fleshed out? I was very happy to see the bcon3 tag on this one, even if just for those glorious 3 days : )

Richard Antalik commented

2020-04-20 19:32:30 +02:00

Well it was on bcon3 workboard for last 2 releases.

I keep this report in mind very much and will try to look in solutions as soon as I can. I just can't promise it to correctly design implement and test within a month. which is time limit for bcon3.

Well it was on bcon3 workboard for last 2 releases. I keep this report in mind very much and will try to look in solutions as soon as I can. I just can't promise it to correctly design implement and test within a month. which is time limit for bcon3.

gabriel montagné commented

2020-04-20 20:34:02 +02:00

Of course, it's just me being annoying : ) Please do what you need to do! I have that workaround thing you'd prepared and it totally works for me. No problem Richard, I know you guys are crazy busy.

Richard Antalik commented

2021-05-25 05:16:13 +02:00

Added subscriber: @joshm-2

Michael commented

2022-01-31 03:07:47 +01:00

Added subscriber: @michaelf

gilad commented

2022-05-22 21:45:48 +02:00

Added subscriber: @Giladfrenkel

gilad commented

2022-05-22 21:49:04 +02:00

Hi, is this bug going to be fixed anytime soon? is there any workaround for the meantime?

Richard Antalik commented

2022-05-24 01:27:04 +02:00

In #69444#1362516, @Giladfrenkel wrote:
Hi, is this bug going to be fixed anytime soon? is there any workaround for the meantime?

Anytime soon I can't tell. I am not sure if there exists infrastructure to handle this issue. There are developers that seem to be able to implemet features in audaspace though so hopefully. For workaround, I would think that duplicating scene and using one solely for sound and another for image would work?

> In #69444#1362516, @Giladfrenkel wrote: > Hi, is this bug going to be fixed anytime soon? is there any workaround for the meantime? Anytime soon I can't tell. I am not sure if there exists infrastructure to handle this issue. There are developers that seem to be able to implemet features in audaspace though so hopefully. For workaround, I would think that duplicating scene and using one solely for sound and another for image would work?

gabriel montagné commented

2022-05-24 12:08:56 +02:00

In #69444#1363251, @iss wrote:

In #69444#1362516, @Giladfrenkel wrote:
Hi, is this bug going to be fixed anytime soon? is there any workaround for the meantime?

Anytime soon I can't tell. I am not sure if there exists infrastructure to handle this issue. There are developers that seem to be able to implemet features in audaspace though so hopefully. For workaround, I would think that duplicating scene and using one solely for sound and another for image would work?

Unfortunately, that won't work. The bug itself is that it's not possible to do this.

@Giladfrenkel , the workaround that I was using is not great: I made my own Blender build where the check to see if the scene has sound was ignored. I personally think that would be a good enough fix (at least for now).

> In #69444#1363251, @iss wrote: >> In #69444#1362516, @Giladfrenkel wrote: >> Hi, is this bug going to be fixed anytime soon? is there any workaround for the meantime? > > Anytime soon I can't tell. I am not sure if there exists infrastructure to handle this issue. There are developers that seem to be able to implemet features in audaspace though so hopefully. For workaround, I would think that duplicating scene and using one solely for sound and another for image would work? Unfortunately, that won't work. The bug itself is that it's not possible to do this. @Giladfrenkel , the workaround that I was using is not great: I made my own Blender build where the check to see if the scene has sound was ignored. I personally think that would be a good enough fix (at least for now).

Joerg Mueller commented

2022-05-24 21:48:33 +02:00

In #69444#1363251, @iss wrote:
There are developers that seem to be able to implemet features in audaspace though so hopefully.

How is this an issue in audaspace? For me it sounds this issue has to do with the blender side of things? I wonder if this is one of the issues introduced with the depsgraph?

> In #69444#1363251, @iss wrote: > There are developers that seem to be able to implemet features in audaspace though so hopefully. How is this an issue in audaspace? For me it sounds this issue has to do with the blender side of things? I wonder if this is one of the issues introduced with the depsgraph?

Richard Antalik commented

2022-05-24 23:28:09 +02:00

In #69444#1363761, @neXyon wrote:

In #69444#1363251, @iss wrote:
There are developers that seem to be able to implemet features in audaspace though so hopefully.

How is this an issue in audaspace? For me it sounds this issue has to do with the blender side of things? I wonder if this is one of the issues introduced with the depsgraph?

As I said, I am not sure if there is infrastructure for proper fix, which to me is having "mixing node" that would combine sounds of scene/meta strip into one unique sound per strip instance. I did not meant to suggest that this is missing in audaspace in particular, just meant to say "I don't know this stuff very well, but saw some development, so hopefully this could be tackled".

AFAIK currently all strip/scene sounds are iterated over and added to active scene. Currently the issue is that multiple instances of scene strip share same sound data and current code have problems dealing with this. The code for this is very simple, see SEQ_eval_sequences and seq_update_seq_cb. There are other issues that have roots in depsgraph design, but this one seems to be just lack of hierarchy that would reflect actual layout of strips/timeline.

That said, if such mixing is possible with existing API, I could try to implement that, but I will need bit of handholding probably first to point out how to structure data on audaspace side and on blender side I should be able to set this up.

> In #69444#1363761, @neXyon wrote: >> In #69444#1363251, @iss wrote: >> There are developers that seem to be able to implemet features in audaspace though so hopefully. > > How is this an issue in audaspace? For me it sounds this issue has to do with the blender side of things? I wonder if this is one of the issues introduced with the depsgraph? As I said, I am not sure if there is infrastructure for proper fix, which to me is having "mixing node" that would combine sounds of scene/meta strip into one unique sound per strip instance. I did not meant to suggest that this is missing in audaspace in particular, just meant to say "I don't know this stuff very well, but saw some development, so hopefully this could be tackled". AFAIK currently all strip/scene sounds are iterated over and added to active scene. Currently the issue is that multiple instances of scene strip share same sound data and current code have problems dealing with this. The code for this is very simple, see `SEQ_eval_sequences` and `seq_update_seq_cb`. There are other issues that have roots in depsgraph design, but this one seems to be just lack of hierarchy that would reflect actual layout of strips/timeline. That said, if such mixing is possible with existing API, I could try to implement that, but I will need bit of handholding probably first to point out how to structure data on audaspace side and on blender side I should be able to set this up.

Joerg Mueller commented

2022-05-25 20:27:43 +02:00

Well, theoretically you only have to call BKE_sound_add_scene_sound[_defaults] when the sound of a scene should be played and call BKE_sound_remove_scene_sound when you want to remove that sound again using the handle you got as return value from the first call. So all you should have to do is to make sure that for a scene strip with input Camera this is not called at all or removed if it is changed and vice versa for the input Sequencer.

Well, theoretically you only have to call `BKE_sound_add_scene_sound[_defaults]` when the sound of a scene should be played and call `BKE_sound_remove_scene_sound` when you want to remove that sound again using the handle you got as return value from the first call. So all you should have to do is to make sure that for a scene strip with input `Camera` this is not called at all or removed if it is changed and vice versa for the input `Sequencer`.

Joerg Mueller commented

2022-05-25 21:04:47 +02:00

I just investigated a bit and it seems like this bug was introduced by @Sergey in 89946834a1 with an improper fix for the issue of a Camera strip playing audio. I assume he never tested having two strips, one with Camera and one with Sequence. @Sergey can you fix this please?

I just investigated a bit and it seems like this bug was introduced by @Sergey in 89946834a1 with an improper fix for the issue of a `Camera` strip playing audio. I assume he never tested having two strips, one with `Camera` and one with `Sequence`. @Sergey can you fix this please?

Sergey Sharybin commented

2022-05-27 19:43:19 +02:00

The dependency graph is following the design within which there is a separation between data and its state. Basically, data is the .blend file and state is the result of the dependency graph evaluation. This enforces all areas adjacent to the dependency graph and scene evaluation to support such separation. This is something what might not be fully done yet due to various reasons.

I can not (for practical reasons) take ownership of code that does not follow the updated design of state evaluation. This is something that module/subproject developers are expected to contribute to.

The dependency graph is following the design within which there is a separation between data and its state. Basically, data is the .blend file and state is the result of the dependency graph evaluation. This enforces all areas adjacent to the dependency graph and scene evaluation to support such separation. This is something what might not be fully done yet due to various reasons. I can not (for practical reasons) take ownership of code that does not follow the updated design of state evaluation. This is something that module/subproject developers are expected to contribute to.

Philipp Oeser removed the

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

VSE Linked external scene with audio with input Sequencer only works if the scene is NOT also linked with input Camera #69444