Using CasparCG for scene based playback in film


Hello all
I’ve just stumbled across this fascinating piece of software, but I’m so far unable to discern if it does what I require, perhaps someone out there can help. I require a software solution for cuing multiple servers or nodes to playback locally loaded video files on a scene based cue. Each file should have a predetermined behaviour such as :" loop" or “stop on last frame” or “continue to next file”. Determining which file each node plays should ideally be based on a list generated on the server/node, an xml file would be perfect. This list would be common to all the server/nodes so that the cues could easily be ordered on the client/controller end.
As yet I haven’t seen an interface for CasparCG that suggests this is possible, in fact it seems to be oriented to a more limited number of outputs. I require control over an unlimited number of nodes networked via cat5, each connected to it’s own screen.
What do you think folks, am I in the right forum or not?


Well it can but it needs a custom made client for that

The AMCP control protocol is really simple to understand soo with some
programming know how you can get there.



Well that’s a start at least, thanks. Is there an existing client close to my needs that I might use as a jumping in point?


Building on what Maurice said, CasparCG server would be used as each of the playout nodes, and can be controlled over amcp with a custom client to get the behaviour you need.

The 3 behaviour modes you described could each be easily achieved.
You can specify that a file should be looped when it is played.
By default any video will freeze on the last frame until cleared or something else is played
You can cue up the next file to be played as soon as the current file ends (ignoring the layers in caspar which you wont want to use, only one can be queued at a time, but that is sufficient to get seamless playback if you can keep feeding it commands)

I cant really comment on any existing clients, as I normally end up writing my own from scratch or using the standard client which wouldnt handle many servers in a friendly fashion.


That would be a nice project to build.

What would be a realistic amount of channels (screens)? Because „unlimited“ is a bit vague. There will be a point were too many packets would go over the network. But I guess up to 50 or 100 should be possible.


Yes a hundred is probably the upper limit of what I require, I can’t recall needing more. The overlay of graphics available in this software could also be very useful for generating “green screen” if content is missing or required from post production, it could also be a quick way of fixing content that seems overexposed to camera by putting a dark layer on top. Is there playback control for frame rate currently included?


Could you give some more informations about the project? Why do you need so much channels?


The requirement is for a playback control for screen playbacks. The content is essentially set decoration for film and TV sets. It could be a single screen on a desk in front of an actor or a huge video wall in an NSA/CIA style control room. I therefor need to be able to scale it up accordingly, a large degree of flexibility is essential as the plan seldom stays the same in this field. Additional screens may be added at a moments notice, content can also be required to change quickly. There are existing software solutions available, but they generally add $1000 to the cost of each node, this prohibits the option for the majority of tv productions.
Adjustments to playback speed and the ability to darken or lighten the content due to lighting conditions is also a frequent requirement, as are adjustable green screen options : colour, saturation, tracking marks.
The user needs to be able to rig a simple interface with scene numbered buttons that can instantly cue a switch in content.
Hope that helps, any ideas?


This is all possible with Caspar, a custom client and some video clips and/or templates can be used to do tracking marks etc. I’d also like to point out that you can use HTML templates with screen consumer in interactive mode, so it would be possible to do live interaction or hidden triggers.


Yup, it’s just like Hreinn said.
It will cost you a penny to get the initial setup working well (whether you’re paying with your own hours or someone else’s) but the node cost should be able to be cut drastically after that, whether you’ll be running multichannel machines or a single machine per channel.


Your´re definitly in the right forum :slight_smile:

Even unlimited nodes should be possible cause as didi already said correctly the bottleneck will be the network. But due to the fact we´re only sending metadata to control the servers and it´s not that much data - I would say that even 100-200 servers would be possible with gigabit ethernet. Maybe it needs some tweaking on the server sides to shut off all the response stuff and your control-client is only broadcasting in one direction. And if you need more: Well… go on 10/40G fiber network… it doesn´t cost that much and could be used for other productions as well :wink: - When copying new content to all nodes it will take some time without an fast network…

By the way: You can adjust brightness etc on the layers itself, no need for an extra “on top” layer to make things darker… there´s a whole palette of adjustments on the official client that could be used.

Please keep in mind that mixing of nodes will automatically end up in non-synchron scenarios. (or when using SDI on some nodes and screen consumers on others…)
(on synchron critical projects I personally would use Watchout for this - but that´s another type of budget :wink: )

May I ask where you´re from?


Hi Youth
Yes I’ve looked at the Dataton and Renewed Vision software options and they are very good, but as you say, rather expensive. I’m based a short distance west of London.


I am setting up a project with several nuc’s on the moment
which seams to work out well.
i guess as balte already stated the investment in time for the custom client can be
worth the money, Because deploying casparcg to the node’s is not expensive at all.
You can use a rsync like tool to sync the media folders i am now testing “syncthing” which
seams to do the job ok. Watchout is a great product but i don’t want to do a project with 100 destinations with it. !

Quick and crude example of sending AMCP commands over multicast