19 Oct
2015
19 Oct
'15
3:11 p.m.
Hi Carl.
To what degree (if any) is DCP-o-matic scriptable?
I've got to make a huge number of DCPs of short films (about 300) for a
festival, each one from a single video file. What I have in mind is:
* I put all the video files in a single folder
* I write a script which executes DCP-o-matic for each file passing the
path of the video file
* For each video, DCP-o-matic makes a DCP-o-matic project file (with
default settings) and save it to another folder
* (Ideally) DCP-o-matic also runs 'Analyse audio' on each video
Why do I want to do this? I'm using batch processing to make the actual
DCPs once I have DCP-o-matic project files, which works great. But with
this many films, I spend a lot of time creating a DCP-o-matic project
file for each file, waiting for it to analyse, then waiting for it to
analyse audio. Each of these processes runs pretty fast, but with this
volume of films, the time spent watching the progress bars is pretty
significant (and boring!)
If possible, it'd be great to be able to script that process so that for
each film I just need to open the project file that's already created
and adjust any settings that I need to. The save the project and move it
into another folder for batch processing to make the actual DCP.
I'm using the v1 branch as stability is key at this point.
Is the CLI interface documented anywhere? And if this isn't possible at
present, could you point me to the part of the codebase which deals with
these processes? Maybe I could hack around and see if I can add this
feature (though C++ is not my strong suit).
Jim
19 Oct
19 Oct
3:29 p.m.
Have you considered cinemaslides?
I see the interest in analyzing audio if you correct it in every
project before doing the batch, is that your workflow?
Not sure about the video format "guessing abilities" or the analyze
speed of v1 vs. v2, but v2 is working flawlessly.
dcpomatic_cli --help
Carl, what's the JSON server there for?
Can the CLI be used for just project creation?
Manuel AC
On Mon, Oct 19, 2015 at 11:11 AM, Jim Dummett via DCPomatic
<dcpomatic(a)carlh.net> wrote:
Hi Carl.
To what degree (if any) is DCP-o-matic scriptable?
I've got to make a huge number of DCPs of short films (about 300) for a
festival, each one from a single video file. What I have in mind is:
* I put all the video files in a single folder
* I write a script which executes DCP-o-matic for each file passing the path
of the video file
* For each video, DCP-o-matic makes a DCP-o-matic project file (with default
settings) and save it to another folder
* (Ideally) DCP-o-matic also runs 'Analyse audio' on each video
Why do I want to do this? I'm using batch processing to make the actual DCPs
once I have DCP-o-matic project files, which works great. But with this many
films, I spend a lot of time creating a DCP-o-matic project file for each
file, waiting for it to analyse, then waiting for it to analyse audio. Each
of these processes runs pretty fast, but with this volume of films, the time
spent watching the progress bars is pretty significant (and boring!)
If possible, it'd be great to be able to script that process so that for
each film I just need to open the project file that's already created and
adjust any settings that I need to. The save the project and move it into
another folder for batch processing to make the actual DCP.
I'm using the v1 branch as stability is key at this point.
Is the CLI interface documented anywhere? And if this isn't possible at
present, could you point me to the part of the codebase which deals with
these processes? Maybe I could hack around and see if I can add this feature
(though C++ is not my strong suit).
Jim
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
6:43 p.m.
On Mon, 19 Oct 2015 11:29:47 -0400
Manuel AC via DCPomatic <dcpomatic(a)carlh.net> wrote:
Carl, what's the JSON server there for?
It's so you can get the status of a dcpomatic_cli run (making a DCP) if
you want to (for a web service that makes DCPs, say).
Can the CLI be used for just project creation?
Yes, see my other mail (dcpomatic_create).
Regards,
Carl
Manuel AC
On Mon, Oct 19, 2015 at 11:11 AM, Jim Dummett via DCPomatic
<dcpomatic(a)carlh.net> wrote:
Hi Carl.
To what degree (if any) is DCP-o-matic scriptable?
I've got to make a huge number of DCPs of short films (about 300) for a
festival, each one from a single video file. What I have in mind is:
* I put all the video files in a single folder
* I write a script which executes DCP-o-matic for each file passing the path
of the video file
* For each video, DCP-o-matic makes a DCP-o-matic project file (with default
settings) and save it to another folder
* (Ideally) DCP-o-matic also runs 'Analyse audio' on each video
Why do I want to do this? I'm using batch processing to make the actual DCPs
once I have DCP-o-matic project files, which works great. But with this many
films, I spend a lot of time creating a DCP-o-matic project file for each
file, waiting for it to analyse, then waiting for it to analyse audio. Each
of these processes runs pretty fast, but with this volume of films, the time
spent watching the progress bars is pretty significant (and boring!)
If possible, it'd be great to be able to script that process so that for
each film I just need to open the project file that's already created and
adjust any settings that I need to. The save the project and move it into
another folder for batch processing to make the actual DCP.
I'm using the v1 branch as stability is key at this point.
Is the CLI interface documented anywhere? And if this isn't possible at
present, could you point me to the part of the codebase which deals with
these processes? Maybe I could hack around and see if I can add this feature
(though C++ is not my strong suit).
Jim
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
6:41 p.m.
Hi Jim,
On Mon, 19 Oct 2015 16:11:11 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
Hi Carl.
To what degree (if any) is DCP-o-matic scriptable?
"To some extent" is probably the answer...
I've got to make a huge number of DCPs of short
films (about 300) for a
festival, each one from a single video file. What I have in mind is:
* I put all the video files in a single folder
* I write a script which executes DCP-o-matic for each file passing the
path of the video file
* For each video, DCP-o-matic makes a DCP-o-matic project file (with
default settings) and save it to another folder
There's a command line tool called dcpomatic_create which will do this; e.g.
dcpomatic_create --dcp-content-type FTR --container-ratio 185 --content-ratio 185 -o
my_project frobozz.mp4
* (Ideally) DCP-o-matic also runs 'Analyse
audio' on each video
This can't be scripted as it stands: what is the purpose of doing this?
Also, what platform are you doing this on?
Why do I want to do this? I'm using batch
processing to make the actual
DCPs once I have DCP-o-matic project files, which works great. But with
this many films, I spend a lot of time creating a DCP-o-matic project
file for each file, waiting for it to analyse, then waiting for it to
analyse audio. Each of these processes runs pretty fast, but with this
volume of films, the time spent watching the progress bars is pretty
significant (and boring!)
Apart from the analysis part that should all be possible.
If possible, it'd be great to be able to script
that process so that for
each film I just need to open the project file that's already created
and adjust any settings that I need to. The save the project and move it
into another folder for batch processing to make the actual DCP.
I'm using the v1 branch as stability is key at this point.
Is the CLI interface documented anywhere? And if this isn't possible at
present, could you point me to the part of the codebase which deals with
these processes? Maybe I could hack around and see if I can add this
feature (though C++ is not my strong suit).
There isn't really any documentation I'm afraid. The CLI parts are
dcpomatic_create: creates a DCP-o-matic project from some content files
dcpomatic_cli: creates a DCP from a set-up DCP-o-matic project
dcpomatic_kdm_cli: creates KDMs
They all have help if you run them with --help
The only thing is I'm not sure if they are distributed with the
Windows / OSX versions of the 1.x branch. Let me know what platform
you are using and I can check.
Regards,
Carl
11:26 p.m.
Hi Carl + Manuel.
Thanks very much for your swift replies.
I'm on running on OS X (a mix of 10.10 Yosemite and 10.6 Snow Leopard
machines).
Amazing! dcpomatic_create sounds like exactly what I'm looking for. Except:
1. I can't see a dcpomatic_create command in
/Applications/DCP-o-matic/Contents/MacOS (with either v1.79.0 or
v2.4.0). Sounds like its Linux only. Would it be hard for me to compile
it from source on OS X?
2. Ideal would be to not pass any `--content-ratio` etc arguments and
leave DCP-o-matic to decide itself the way it does in the GUI. Will it
provide defaults based on the format of the source file based on these
arguments?
The reason for running "analyse audio" is that for each programme of
short films, the audio levels in the source files from film-makers will
vary enormously. So I like to adjust them all to peak at the same level
(-10dB). This is a rather rough measure - it still needs adjustment in
the cinema - but it at least gets all the films' sound levels in the
same ballpark. But running "analyse audio" in the GUI is the slowest
part of the process (apart from the DCP render itself, but I'm batching
that).
Is it ffmpeg doing the audio analysis under the hood? If so, maybe I
could run that command directly and then format the results and insert
them into the DCP-o-matic project folder to "fool" DCP-o-matic into
thinking it's done the audio analysis already.
By the way, Carl, you mention the JSON server is intended for
applications like "for a web service that makes DCPs, say". Do you know
of anyone who's used DCP-o-matic for something like this?
Sorry for the barrage of questions.
Jim
On 19/10/2015 19:41, Carl Hetherington via DCPomatic wrote:
Hi Jim,
On Mon, 19 Oct 2015 16:11:11 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
Hi Carl.
To what degree (if any) is DCP-o-matic scriptable?
"To some extent" is
probably the answer...
I've got to make a huge number of DCPs of
short films (about 300) for a
festival, each one from a single video file. What I have in mind is:
* I put all the video files in a single folder
* I write a script which executes DCP-o-matic for each file passing the
path of the video file
* For each video, DCP-o-matic makes a DCP-o-matic project file (with
default settings) and save it to another folder
There's a command line tool
called dcpomatic_create which will do this; e.g.
dcpomatic_create --dcp-content-type FTR --container-ratio 185 --content-ratio 185 -o
my_project frobozz.mp4
* (Ideally) DCP-o-matic also runs 'Analyse
audio' on each video
This can't be scripted as it stands: what is the
purpose of doing this?
Also, what platform are you doing this on?
Why do I want to do this? I'm using batch
processing to make the actual
DCPs once I have DCP-o-matic project files, which works great. But with
this many films, I spend a lot of time creating a DCP-o-matic project
file for each file, waiting for it to analyse, then waiting for it to
analyse audio. Each of these processes runs pretty fast, but with this
volume of films, the time spent watching the progress bars is pretty
significant (and boring!)
Apart from the analysis part that should all be
possible.
If possible, it'd be great to be able to
script that process so that for
each film I just need to open the project file that's already created
and adjust any settings that I need to. The save the project and move it
into another folder for batch processing to make the actual DCP.
I'm using the v1 branch as stability is key at this point.
Is the CLI interface documented anywhere? And if this isn't possible at
present, could you point me to the part of the codebase which deals with
these processes? Maybe I could hack around and see if I can add this
feature (though C++ is not my strong suit).
There isn't really any
documentation I'm afraid. The CLI parts are
dcpomatic_create: creates a DCP-o-matic project from some content files
dcpomatic_cli: creates a DCP from a set-up DCP-o-matic project
dcpomatic_kdm_cli: creates KDMs
They all have help if you run them with --help
The only thing is I'm not sure if they are distributed with the
Windows / OSX versions of the 1.x branch. Let me know what platform
you are using and I can check.
Regards,
Carl
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic 
11:36 p.m.
Am 20.10.2015 um 01:26 schrieb Jim Dummett via DCPomatic:
Hi Carl + Manuel.
Thanks very much for your swift replies.
I'm on running on OS X (a mix of 10.10 Yosemite and 10.6 Snow Leopard machines).
Amazing! dcpomatic_create sounds like exactly what I'm looking for. Except:
1. I can't see a dcpomatic_create command in /Applications/DCP-o-matic/Contents/MacOS
(with either v1.79.0 or v2.4.0). Sounds like its Linux only. Would it be hard for me to
compile it from source on OS X?
Oops - Carl will comment, but I think dcpomatic_create is not available on the Mac
platform. I think that so far there is only the CLI version of the encoding server aside
from the GUI main application.
- Carsten
20 Oct
20 Oct
12:24 a.m.
On Tue, 20 Oct 2015 00:26:22 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
Hi Carl + Manuel.
Thanks very much for your swift replies.
I'm on running on OS X (a mix of 10.10 Yosemite and 10.6 Snow Leopard
machines).
Amazing! dcpomatic_create sounds like exactly what I'm looking for. Except:
1. I can't see a dcpomatic_create command in
/Applications/DCP-o-matic/Contents/MacOS (with either v1.79.0 or
v2.4.0). Sounds like its Linux only. Would it be hard for me to compile
it from source on OS X?
This is just a packaging error (the executable is built but not
included in the .dmg, and I didn't notice). This could easily be
corrected for 2.x and moderately easily corrected for 1.x.
Building the OS X version from source is something of a Sisyphean ordeal
that I wouldn't wish on you ;)
2. Ideal would be to not pass any `--content-ratio`
etc arguments and
leave DCP-o-matic to decide itself the way it does in the GUI. Will it
provide defaults based on the format of the source file based on these
arguments?
Yes I'm pretty sure it will (with the same guesswork as the normal
front-end).
The reason for running "analyse audio" is
that for each programme of
short films, the audio levels in the source files from film-makers will
vary enormously. So I like to adjust them all to peak at the same level
(-10dB). This is a rather rough measure - it still needs adjustment in
the cinema - but it at least gets all the films' sound levels in the
same ballpark. But running "analyse audio" in the GUI is the slowest
part of the process (apart from the DCP render itself, but I'm batching
that).
Is it ffmpeg doing the audio analysis under the hood? If so, maybe I
could run that command directly and then format the results and insert
them into the DCP-o-matic project folder to "fool" DCP-o-matic into
thinking it's done the audio analysis already.
Sadly not, it's DCP-o-matic code. That wouldn't be too hard to add to
dcpomatic_create though.
By the way, Carl, you mention the JSON server is
intended for
applications like "for a web service that makes DCPs, say". Do you know
of anyone who's used DCP-o-matic for something like this?
I built a rather half-baked django-based site (cine-matic.com) which is
still up but probably no longer working. There wasn't much interest.
The data transfer is a bit of a problem, and it's quite clumsy
providing detailed control of the output with a web UI. I'm sure a
more experienced developer could do a better job mind you.
I have vaguely thought about a menu option in DCP-o-matic to "make dcp
in the cloud (TM)" but the amount of data transfer puts me off.
Cheers,
Carl
Sorry for the barrage of questions.
Jim
On 19/10/2015 19:41, Carl Hetherington via DCPomatic wrote:
Hi Jim,
On Mon, 19 Oct 2015 16:11:11 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic Hi Carl.
To what degree (if any) is DCP-o-matic scriptable?
"To some extent" is
probably the answer...
I've got to make a huge number of DCPs of
short films (about 300) for a
festival, each one from a single video file. What I have in mind is:
* I put all the video files in a single folder
* I write a script which executes DCP-o-matic for each file passing the
path of the video file
* For each video, DCP-o-matic makes a DCP-o-matic project file (with
default settings) and save it to another folder
There's a command line tool
called dcpomatic_create which will do this; e.g.
dcpomatic_create --dcp-content-type FTR --container-ratio 185 --content-ratio 185 -o
my_project frobozz.mp4
* (Ideally) DCP-o-matic also runs 'Analyse
audio' on each video
This can't be scripted as it stands: what is the
purpose of doing this?
Also, what platform are you doing this on?
Why do I want to do this? I'm using batch
processing to make the actual
DCPs once I have DCP-o-matic project files, which works great. But with
this many films, I spend a lot of time creating a DCP-o-matic project
file for each file, waiting for it to analyse, then waiting for it to
analyse audio. Each of these processes runs pretty fast, but with this
volume of films, the time spent watching the progress bars is pretty
significant (and boring!)
Apart from the analysis part that should all be
possible.
If possible, it'd be great to be able to
script that process so that for
each film I just need to open the project file that's already created
and adjust any settings that I need to. The save the project and move it
into another folder for batch processing to make the actual DCP.
I'm using the v1 branch as stability is key at this point.
Is the CLI interface documented anywhere? And if this isn't possible at
present, could you point me to the part of the codebase which deals with
these processes? Maybe I could hack around and see if I can add this
feature (though C++ is not my strong suit).
There isn't really any
documentation I'm afraid. The CLI parts are
dcpomatic_create: creates a DCP-o-matic project from some content files
dcpomatic_cli: creates a DCP from a set-up DCP-o-matic project
dcpomatic_kdm_cli: creates KDMs
They all have help if you run them with --help
The only thing is I'm not sure if they are distributed with the
Windows / OSX versions of the 1.x branch. Let me know what platform
you are using and I can check.
Regards,
Carl
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
21 Oct
21 Oct
3:44 a.m.
Hi Carl.
Well it would be great if the dcpomatic_create executable could be added
to the OS X package in a future release. But for now I'm running it
inside a Ubuntu VM and it's working well.
The --content-ratio argument is compulsory but seems to affect the
container size rather than affecting how the content is interpreted.
Can you possibly point me to the part of the DCP-o-matic code base which
analyses the audio? If it's not too complex, I could maybe port it and
get it running stand-alone.
Thanks loads for all your help!
Jim
On 20/10/2015 01:24, Carl Hetherington via DCPomatic wrote:
On Tue, 20 Oct 2015 00:26:22 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
Hi Carl + Manuel.
Thanks very much for your swift replies.
I'm on running on OS X (a mix of 10.10 Yosemite and 10.6 Snow Leopard
machines).
Amazing! dcpomatic_create sounds like exactly what I'm looking for. Except:
1. I can't see a dcpomatic_create command in
/Applications/DCP-o-matic/Contents/MacOS (with either v1.79.0 or
v2.4.0). Sounds like its Linux only. Would it be hard for me to compile
it from source on OS X?
This is just a packaging error (the executable is built
but not
included in the .dmg, and I didn't notice). This could easily be
corrected for 2.x and moderately easily corrected for 1.x.
Building the OS X version from source is something of a Sisyphean ordeal
that I wouldn't wish on you ;)
2. Ideal would be to not pass any
`--content-ratio` etc arguments and
leave DCP-o-matic to decide itself the way it does in the GUI. Will it
provide defaults based on the format of the source file based on these
arguments?
Yes I'm pretty sure it will (with the same guesswork as the normal
front-end).
The reason for running "analyse audio"
is that for each programme of
short films, the audio levels in the source files from film-makers will
vary enormously. So I like to adjust them all to peak at the same level
(-10dB). This is a rather rough measure - it still needs adjustment in
the cinema - but it at least gets all the films' sound levels in the
same ballpark. But running "analyse audio" in the GUI is the slowest
part of the process (apart from the DCP render itself, but I'm batching
that).
Is it ffmpeg doing the audio analysis under the hood? If so, maybe I
could run that command directly and then format the results and insert
them into the DCP-o-matic project folder to "fool" DCP-o-matic into
thinking it's done the audio analysis already.
Sadly not, it's DCP-o-matic
code. That wouldn't be too hard to add to
dcpomatic_create though.
By the way, Carl, you mention the JSON server is
intended for
applications like "for a web service that makes DCPs, say". Do you know
of anyone who's used DCP-o-matic for something like this?
I built a rather
half-baked django-based site (cine-matic.com) which is
still up but probably no longer working. There wasn't much interest.
The data transfer is a bit of a problem, and it's quite clumsy
providing detailed control of the output with a web UI. I'm sure a
more experienced developer could do a better job mind you.
I have vaguely thought about a menu option in DCP-o-matic to "make dcp
in the cloud (TM)" but the amount of data transfer puts me off.
Cheers,
Carl
Sorry for the barrage of questions.
Jim
On 19/10/2015 19:41, Carl Hetherington via DCPomatic wrote:
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic Hi Jim,
On Mon, 19 Oct 2015 16:11:11 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic Hi Carl.
To what degree (if any) is DCP-o-matic scriptable?
"To some extent" is
probably the answer...
I've got to make a huge number of DCPs of
short films (about 300) for a
festival, each one from a single video file. What I have in mind is:
* I put all the video files in a single folder
* I write a script which executes DCP-o-matic for each file passing the
path of the video file
* For each video, DCP-o-matic makes a DCP-o-matic project file (with
default settings) and save it to another folder
There's a command line tool
called dcpomatic_create which will do this; e.g.
dcpomatic_create --dcp-content-type FTR --container-ratio 185 --content-ratio 185 -o
my_project frobozz.mp4
* (Ideally) DCP-o-matic also runs 'Analyse
audio' on each video
This can't be scripted as it stands: what is the
purpose of doing this?
Also, what platform are you doing this on?
Why do I want to do this? I'm using batch
processing to make the actual
DCPs once I have DCP-o-matic project files, which works great. But with
this many films, I spend a lot of time creating a DCP-o-matic project
file for each file, waiting for it to analyse, then waiting for it to
analyse audio. Each of these processes runs pretty fast, but with this
volume of films, the time spent watching the progress bars is pretty
significant (and boring!)
Apart from the analysis part that should all be
possible.
If possible, it'd be great to be able to
script that process so that for
each film I just need to open the project file that's already created
and adjust any settings that I need to. The save the project and move it
into another folder for batch processing to make the actual DCP.
I'm using the v1 branch as stability is key at this point.
Is the CLI interface documented anywhere? And if this isn't possible at
present, could you point me to the part of the codebase which deals with
these processes? Maybe I could hack around and see if I can add this
feature (though C++ is not my strong suit).
There isn't really any
documentation I'm afraid. The CLI parts are
dcpomatic_create: creates a DCP-o-matic project from some content files
dcpomatic_cli: creates a DCP from a set-up DCP-o-matic project
dcpomatic_kdm_cli: creates KDMs
They all have help if you run them with --help
The only thing is I'm not sure if they are distributed with the
Windows / OSX versions of the 1.x branch. Let me know what platform
you are using and I can check.
Regards,
Carl
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
8:18 a.m.
Hi Jim,
On Wed, 21 Oct 2015 04:44:24 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
Hi Carl.
Well it would be great if the dcpomatic_create executable could be added
to the OS X package in a future release. But for now I'm running it
inside a Ubuntu VM and it's working well.
The --content-ratio argument is compulsory but seems to affect the
container size rather than affecting how the content is interpreted.
It looks like that is left over from before DCP-o-matic did better
guessing of ratio from the input size. It can probably be made
optional. Are you using a 1.x version?
Can you possibly point me to the part of the
DCP-o-matic code base which
analyses the audio? If it's not too complex, I could maybe port it and
get it running stand-alone.
Mostly it is in src/lib/analyse_audio_job.cc. If you want to do the
same thing standalone you might find it easier to do something like
ffmpeg -i audio.mxf -af "volumedetect" -f null /dev/null
Regards,
Carl
Thanks loads for all your help!
Jim
On 20/10/2015 01:24, Carl Hetherington via DCPomatic wrote:
On Tue, 20 Oct 2015 00:26:22 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic Hi Carl + Manuel.
Thanks very much for your swift replies.
I'm on running on OS X (a mix of 10.10 Yosemite and 10.6 Snow Leopard
machines).
Amazing! dcpomatic_create sounds like exactly what I'm looking for. Except:
1. I can't see a dcpomatic_create command in
/Applications/DCP-o-matic/Contents/MacOS (with either v1.79.0 or
v2.4.0). Sounds like its Linux only. Would it be hard for me to compile
it from source on OS X?
This is just a packaging error (the executable is built
but not
included in the .dmg, and I didn't notice). This could easily be
corrected for 2.x and moderately easily corrected for 1.x.
Building the OS X version from source is something of a Sisyphean ordeal
that I wouldn't wish on you ;)
2. Ideal would be to not pass any
`--content-ratio` etc arguments and
leave DCP-o-matic to decide itself the way it does in the GUI. Will it
provide defaults based on the format of the source file based on these
arguments?
Yes I'm pretty sure it will (with the same guesswork as the normal
front-end).
The reason for running "analyse audio"
is that for each programme of
short films, the audio levels in the source files from film-makers will
vary enormously. So I like to adjust them all to peak at the same level
(-10dB). This is a rather rough measure - it still needs adjustment in
the cinema - but it at least gets all the films' sound levels in the
same ballpark. But running "analyse audio" in the GUI is the slowest
part of the process (apart from the DCP render itself, but I'm batching
that).
Is it ffmpeg doing the audio analysis under the hood? If so, maybe I
could run that command directly and then format the results and insert
them into the DCP-o-matic project folder to "fool" DCP-o-matic into
thinking it's done the audio analysis already.
Sadly not, it's DCP-o-matic
code. That wouldn't be too hard to add to
dcpomatic_create though.
By the way, Carl, you mention the JSON server is
intended for
applications like "for a web service that makes DCPs, say". Do you know
of anyone who's used DCP-o-matic for something like this?
I built a rather
half-baked django-based site (cine-matic.com) which is
still up but probably no longer working. There wasn't much interest.
The data transfer is a bit of a problem, and it's quite clumsy
providing detailed control of the output with a web UI. I'm sure a
more experienced developer could do a better job mind you.
I have vaguely thought about a menu option in DCP-o-matic to "make dcp
in the cloud (TM)" but the amount of data transfer puts me off.
Cheers,
Carl
Sorry for the barrage of questions.
Jim
On 19/10/2015 19:41, Carl Hetherington via DCPomatic wrote:
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic Hi Jim,
On Mon, 19 Oct 2015 16:11:11 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
> Hi Carl.
>
> To what degree (if any) is DCP-o-matic scriptable?
"To some extent" is probably the answer...
> I've got to make a huge number of DCPs of short films (about 300) for a
> festival, each one from a single video file. What I have in mind is:
>
> * I put all the video files in a single folder
> * I write a script which executes DCP-o-matic for each file passing the
> path of the video file
> * For each video, DCP-o-matic makes a DCP-o-matic project file (with
> default settings) and save it to another folder
There's a command line tool called dcpomatic_create which will do this; e.g.
dcpomatic_create --dcp-content-type FTR --container-ratio 185 --content-ratio 185 -o
my_project frobozz.mp4
> * (Ideally) DCP-o-matic also runs 'Analyse audio' on each video
This can't be scripted as it stands: what is the purpose of doing this?
Also, what platform are you doing this on?
> Why do I want to do this? I'm using batch processing to make the actual
> DCPs once I have DCP-o-matic project files, which works great. But with
> this many films, I spend a lot of time creating a DCP-o-matic project
> file for each file, waiting for it to analyse, then waiting for it to
> analyse audio. Each of these processes runs pretty fast, but with this
> volume of films, the time spent watching the progress bars is pretty
> significant (and boring!)
Apart from the analysis part that should all be possible.
> If possible, it'd be great to be able to script that process so that for
> each film I just need to open the project file that's already created
> and adjust any settings that I need to. The save the project and move it
> into another folder for batch processing to make the actual DCP.
>
> I'm using the v1 branch as stability is key at this point.
>
> Is the CLI interface documented anywhere? And if this isn't possible at
> present, could you point me to the part of the codebase which deals with
> these processes? Maybe I could hack around and see if I can add this
> feature (though C++ is not my strong suit).
There isn't really any documentation I'm afraid. The CLI parts are
dcpomatic_create: creates a DCP-o-matic project from some content files
dcpomatic_cli: creates a DCP from a set-up DCP-o-matic project
dcpomatic_kdm_cli: creates KDMs
They all have help if you run them with --help
The only thing is I'm not sure if they are distributed with the
Windows / OSX versions of the 1.x branch. Let me know what platform
you are using and I can check.
Regards,
Carl
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
6:27 p.m.
Hi Carl.
Brilliant. I'll have a look.
By the way, cine-matic.com is an interesting concept. But as you say,
the problem is the volume of data that has to go up to, and down again
from the cloud. I'm kind of working on something similar - a web app
that controls DCP creation, but my approach has been that the web app in
the cloud only acts as command-and-control - the actual DCP-encoding
happens on worker machines down here in the "real world", with big
drives attached. So the video data never needs to go via the cloud.
I'm still working on it and road-testing it with this big batch of
films, but I'll let you all know when it's in workable shape.
Jim
It looks like that is left over from before
DCP-o-matic did better
guessing of ratio from the input size. It can probably be made
optional. Are you using a 1.x version
Yes, I'm on v1.83.0. --content-ratio does have an effect, just not the
effect I expected! It dictates the container (scope, flat etc). The
--container-ratio argument isn't compulsory and I didn't try it.
Can you
possibly point me to the part of the DCP-o-matic code base which
analyses the audio? If it's not too complex, I could maybe port it and
get it running stand-alone.
Mostly it is in src/lib/analyse_audio_job.cc. If you
want to do the
same thing standalone you might find it easier to do something like
ffmpeg -i audio.mxf -af "volumedetect" -f null /dev/null
22 Oct
22 Oct
8:52 a.m.
On Wed, 21 Oct 2015 19:27:41 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
Hi Carl.
It might make more sense if you did use --container-ratio as well; if
you don't specify it dcpomatic_create will set container_ratio =
content_ratio. I'm not quite sure why it does that ... sounds like
there is some tidying up needed.
[snip]
It looks like that is left over from before
DCP-o-matic did better
guessing of ratio from the input size. It can probably be made
optional. Are you using a 1.x version
Yes, I'm on v1.83.0. --content-ratio does have an effect, just not the
effect I expected! It dictates the container (scope, flat etc). The
--container-ratio argument isn't compulsory and I didn't try it. By the way, cine-matic.com is an interesting concept.
But as you say,
the problem is the volume of data that has to go up to, and down again
from the cloud. I'm kind of working on something similar - a web app
that controls DCP creation, but my approach has been that the web app in
the cloud only acts as command-and-control - the actual DCP-encoding
happens on worker machines down here in the "real world", with big
drives attached. So the video data never needs to go via the cloud.
I'm afraid your cine-matic job won't run at the moment; the backend
script is now out of date with the API of the cloud servers that I was
using and I haven't got around to fixing it.
Your creation app sounds interesting... what makes you want a web app
rather than a local app? Just so you can initiate DCP creation from
anywhere?
Cheers,
Carl
2:35 p.m.
Hi Jim
The thought of a Web based creation app is imho really interesting, Say at
a company with users from different departments, but all on the same fiber
network.
I'm thinking users could:
1. Upload the material for conversion (still/video/separate audio).
2. Create project and Input metadata (dcp name, kind, bitrate, etc)
3. Be able to choose Audio Analyse and get either just the peak or maybe
even a graphical representation (like in DOM).
4. Choose to put the DCP in queue (that doesn't start until you tell it to)
or encode it right away. (transfer DCP to FTP/SFTP)
Encoding would happen on local machines on the LAN or on just one machine.
It would basically be a web frontend for DOM.
Is that somewhere in the ballpark of what you have in mind?
Best regards,
Jonathan
On Oct 22, 2015 10:52 AM, "Carl Hetherington via DCPomatic" <
dcpomatic(a)carlh.net> wrote:
On Wed, 21 Oct 2015 19:27:41 +0100
Jim Dummett via DCPomatic <dcpomatic(a)carlh.net> wrote:
Hi Carl.
It might make more sense if you did use --container-ratio as well; if
you don't specify it dcpomatic_create will set container_ratio =
content_ratio. I'm not quite sure why it does that ... sounds like
there is some tidying up needed.
[snip]
It looks like that is left over from before
DCP-o-matic did better
guessing of ratio from the input size. It can probably be made
optional. Are you using a 1.x version
Yes, I'm on v1.83.0. --content-ratio does have an effect, just not the
effect I expected! It dictates the container (scope, flat etc). The
--container-ratio argument isn't compulsory and I didn't try it. By the way, cine-matic.com is an interesting
concept. But as you say,
the problem is the volume of data that has to go up to, and down again
from the cloud. I'm kind of working on something similar - a web app
that controls DCP creation, but my approach has been that the web app in
the cloud only acts as command-and-control - the actual DCP-encoding
happens on worker machines down here in the "real world", with big
drives attached. So the video data never needs to go via the cloud.
I'm afraid your cine-matic job won't run at the moment; the backend
script is now out of date with the API of the cloud servers that I was
using and I haven't got around to fixing it.
Your creation app sounds interesting... what makes you want a web app
rather than a local app? Just so you can initiate DCP creation from
anywhere?
Cheers,
Carl
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
10:38 p.m.
Am 22.10.2015 um 16:35 schrieb Jonathan K. F. Jensen via DCPomatic:
1. Upload the material for conversion
(still/video/separate audio).
2. Create project and Input metadata (dcp name, kind, bitrate, etc)
3. Be able to choose Audio Analyse and get either just the peak or maybe even a graphical
representation (like in DOM).
4. Choose to put the DCP in queue (that doesn't start until you tell it to) or encode
it right away. (transfer DCP to FTP/SFTP)
Encoding would happen on local machines on the LAN or on just one machine. It would
basically be a web frontend for DOM
But they can do all this already now by running their local copy of DCP-o-matic? Yes, you
would say that with a web-frontend, no local installation would be necessary. But who
would think of a work scenario where multiple people would routinely do DCP
creation/conversion, but would not be allowed a local installation of a DCP conversion
software (and one that is more or less free of side-effects like DOM is)?
Dynamic render node allocation would be an interesting task, but I think we are getting
into a VERY professional scenario here. I can hardly imagine why so many people would
routinely do so many DCPs?
Carl, usually we talk about one DOM Master GUI talking to many render nodes. What if two
DOM Master GUIs try to talk to the same render node?
Regarding Jims approach of unattended batch conversion - sorry, but we DO already have the
issue of so many festival films being in weird formats suffering from unprofessional
mastering. How could an automated approach for a DCP conversion improve that situation?
No, if one is converting all festival films into DCP for ease of presentation, it should
naturally go with a quality check before the conversion that makes sure every major aspect
of the film and its conversion is controlled by a trained person. Yes, 300 is a lot, but
that is just what festivals are about. A festival should pay respect to the artwork it is
aiming to present, and should not deal with it like canned beans for reasons of
efficiency.
Settting these projects up for conversion is not such a big deal and HAS to be done
manually, the J2k coding is still the major time consuming part of this job.
Preparing 300 films for presentation simply needs time. Shooting them took it's time
as well. That's just the way it goes - after all, an all-DCP presentation scheme does
save a lot of time and hassle already, that can be invested in proper preparation.
Create a festival database, feed it, setup DOM for every film, check format, color
conversion, audio channels, levels, etc. send it to a batch conversion queue.
- Carsten
26 Oct
26 Oct
10:15 a.m.
On Fri, 23 Oct 2015, Carsten Kurz via DCPomatic wrote:
Am 22.10.2015 um 16:35 schrieb Jonathan K. F. Jensen via DCPomatic:
1. Upload the material for conversion (still/video/separate audio).
2. Create project and Input metadata (dcp name, kind, bitrate, etc)
3. Be able to choose Audio Analyse and get either just the peak or maybe even a
graphical representation (like in DOM).
4. Choose to put the DCP in queue (that doesn't start until you tell it to) or
encode it right away. (transfer DCP to FTP/SFTP)
Encoding would happen on local machines on the LAN or on just one machine. It would
basically be a web frontend for DOM
[snip]
But they can do all this already now by running their
local copy of
DCP-o-matic? Yes, you would say that with a web-frontend, no local
installation would be necessary. But who would think of a work scenario
where multiple people would routinely do DCP creation/conversion, but
would not be allowed a local installation of a DCP conversion software
(and one that is more or less free of side-effects like DOM is)?
[snip]
Carl, usually we talk about one DOM Master GUI talking
to many render
nodes. What if two DOM Master GUIs try to talk to the same render node?
It should work fine. The only thing that I think will happen with the
current code is each render node will offer all its cores to anyone who
asks, so it may end up taking on too much work. But you could just run
some render nodes and then let anyone in an organisation connect to them.
Settting these projects up for conversion is not such
a big deal and HAS
to be done manually, the J2k coding is still the major time consuming
part of this job. Preparing 300 films for presentation simply needs
time. Shooting them took it's time as well. That's just the way it goes
- after all, an all-DCP presentation scheme does save a lot of time and
hassle already, that can be invested in proper preparation.
Maybe there is scope to improve the workflow of DCP-o-matic for this kind
of application.
Regards,
Carl
27 Oct
27 Oct
10:49 a.m.
Maybe there is scope to improve the workflow of
DCP-o-matic for this kind
of application.
Maybe, instead of/in addition to creating the full DCP automatically, there could also be
an option to create project files/folders with most parameters being set automatically,
e.g. source and destination, name, etc. Then the projects would just need to be opened
again for final adjustment and then be sent to the batch processor? That would still allow
for all manual overriding. If CPU power permits, one could create all DCPs well in
advance, watch them for QC on 'real' Equipment, then go back to individual project
files for adjustment.
With +300 files, there certainly also is a need for special naming, if you want to track a
special program. Like adding a number for clear reference from a database.
Always wondered wether there could be a way to open project files directly by
double-clicking on them, or by using 'Sent-To/OpenWith->DOM'?
- Carsten
3 Nov
3 Nov
9:29 a.m.
On Tue, 27 Oct 2015 11:49:26 +0100
Carsten Kurz via DCPomatic <dcpomatic(a)carlh.net> wrote:
This could be done... I'd go along with the ideas of people who do this
kind of volume of conversions.
Maybe there is scope to improve the workflow of
DCP-o-matic for
this kind of application.
Maybe, instead of/in addition to creating the full DCP automatically,
there could also be an option to create project files/folders with
most parameters being set automatically, e.g. source and destination,
name, etc. Then the projects would just need to be opened again for
final adjustment and then be sent to the batch processor? That would
still allow for all manual overriding. If CPU power permits, one
could create all DCPs well in advance, watch them for QC on 'real'
Equipment, then go back to individual project files for adjustment. With +300 files, there certainly also is a need for
special naming,
if you want to track a special program. Like adding a number for
clear reference from a database.
Always wondered wether there could be a way to open project files
directly by double-clicking on them, or by using
'Sent-To/OpenWith->DOM'?
You mean the .xml file? I'm sure that's possible...
Regards,
Carl
3464
days inactive
3479
days old
15 comments
5 participants
participants (5)
-
Carl Hetherington -
Carsten Kurz -
Jim Dummett -
Jonathan K. F. Jensen -
Manuel AC