[Matthew Keane]

As Aakash says, the key is to use a very wide, skinny shape emitter, for example a box emitter that’s 1x1x1000. You can then use one motion path for all of the particles and use Aux particles to get the trails.

[Matthew Keane]

Wait, I remembered incorrectly – The first, and any other, instance of the Element effect needs to be set to ‘Unified’ render mode. The ‘Render Mode’ of the last copy of the effect determines the render quality of the merged effects. Sorry about that – as I said, I try to use the ‘Aux Effect’ workaround to avoid the need for more than 5 groups whenever possible, so it’s been a while.

[Matthew Keane]

[Ishikawa Yu] “Is there an unlimited group valid button?”

Kind of. If you add two copies of the Element3D effect to a layer, you need to set the ‘Render Mode’ of the second copy to ‘Unified’. That will tell Element to merge the two effect layers, rather than rendering them separately.

I assume it works the same for multiple copies, but I’ve never used more than 2 – renders slow down enough with 10 groups! Also, I use the ‘Aux Channels’ quite a bit to control animation within one group, rather than splitting everything into different groups.

Matthew Keane

Freelancer based in Paris, France
– Motion Graphics, Video Editing & Effects, Media Server Programming & Live Operation.

[Matthew Keane]

Hi,

Not sure if you want to go down the plug-in route, but they will probably render faster than thousands of scale expressions.

If it were me, I’d use the Trapcode Form plug-in to do this – set up a grid of dots, tell it to use your gradient as a layer map to affect scale/opacity of your particles and then tweak until it looks the way you like.

On the other hand if you don’t have access to third-party plugins, maybe the built-in CC Ball effect can achieve something similar. I haven’t used it in a long time, but I seem to remember you can use a layer to affect the particles in a similar way to Form.

Matthew Keane

Freelancer based in Paris, France
– Motion Graphics, Video Editing & Effects, Media Server Programming & Live Operation.

[Matthew Keane]

Hi,

There are a few media-server systems I could imagine being used for concerts like this but, since it’s Elton John, it wasn’t to hard to find an article talking about the tech setup:
https://plsn.com/articles/production-profile/elton-john-farewell-yellow-brick-road-tour/

In this case, they used Disguise (previously called D3) servers, which are often used in high-end shows. Disguise does a lot more than just media playback – it allows 3D previsualisation of the show before you even get into the venue, has a robust system for automatic failover between player machines, and integrates with real-time graphic systems like Notch (which I thought might have been used for the fire FX but it’s not mentioned in the article).

There are other media servers out there – all with different strengths and features which will affect the choice of system for a show (in addition to reputation, local preferences, and availability of operators). As you suggest, unlike VJ software, media servers for this kind of performance can be integrated with, or act as, show-control systems – either as a master or slave (sending or receiving timecode, or network commands) to communicate with sound and lighting consoles or stage machinery.

Off the top of my head, other systems include:

– Dataton Watchout (one I use a lot). Not the most flexible system out there (any changes to the show or media have to be pushed from the master to player machines) but a very solid system which includes some compositing features useful for on-site changes to shows, and is easy to expand for multi-projector and large LED wall projects (just keep adding players until your budget runs out!). Supports 2D and 3D mapping.

– Modulo PI, created in France but gradually becoming more well-known abroad. it exists in 2 versions: Modulo Player – the simpler version designed for cue-list style media playback with an easy to use 2D warping system for video mapping. Modulo Kinetic is the more advanced version, probably closer to Disguise in features, with support for 3D mapping and more interactive features.

– Smode grow out of the demo scene and is designed for real-time media manipulation (like After Effects, but live). Exists in a ‘studio’ edition, designed for smaller projects, and the ‘station’ edition designed for larger multi-output shows.

There are also hardware systems like the AnalogWay PictureAll which supports 8K playback – but doesn’t – as far as I know – include any of the editing, FX or mapping features of media-server software. It’s more like a very expensive tape deck! But if your show is entirely prerecorded and just needs to be played back onto big LED walls, it could do the job.

There is probably a more exhaustive comparison of different systems out there, but that’s start.

Matthew Keane

Freelancer based in Paris, France
– Motion Graphics, Video Editing & Effects, Watchout Programming & Live Operation.

[Matthew Keane]

Perhaps not a complete replacement of Rift, but you might want to take a look at the Lazy script:
https://aescripts.com/lazy/

[Matthew Keane]

Forgot to say: The zip file contains an AE (CC 2017) project, which has the grid image imported. Also in the zip are an XML Display Builder project file, a CSV .txt export with all the crop values for the screen comps and a screen grab of the layout project.

Matthew Keane

Freelancer based in Paris, France
– Motion Graphics, Video Editing & Effects, Watchout Programming & Live Operation.

[Matthew Keane]

Like I said, I developed the tool to allow me to work with Watchout, so it doesn’t create AE comps for you. I used it to export a grid lineup image to check that my expression to position the master comp in individual screens was working correctly.

I added the expression to the anchor point of the main comp inside the 1280×720 screen comp. It looks at the name of the comp and uses that to calculate the position for the main comp so that the correct part is displayed in each screen. Now, in the Project window, hit ‘duplicate’ 79 times, and you end up with 80 comps showing the correct part of the main comp.

Example project (with just the first screen comp, ready to be duplicated) here:
12569_aeledwall01.zip

But – and it’s a big ‘but’ – if you make 80 copies of your main comp, you will be basically be rendering your huge master comp 80 times, and discarding pixels outside of the smaller screen comps – not very efficient. Which is why I suggested rendering the main comp out once, to an image sequence, and using that as the source in the screen comps – it will be much, much faster.

Matthew Keane

Freelancer based in Paris, France
– Motion Graphics, Video Editing & Effects, Watchout Programming & Live Operation.

[Matthew Keane]

Dave’s pencil-and-paper method is good advice but, for lazy people like me, there are some tools that can help automate the process…

If you are going to be doing a lot of large-scale LED wall projects, there is a script on AEscripts – https://aescripts.com/stagetool/ – that is designed to help layout comps for LED pixel mapping in Resolume, but it could perhaps be useful for setting up LED wall projects generally, so might be worth a look.

Personally though, any time I see a repetitive task like slicing up a giant comp, I’d start thinking about using expressions to speed up the process…

So, first of all, a shameless plug for one of my own tools (https://matkeane.com/project/watchout-display-builder-softedge-calculator), which I use for calculating the layout of grids of displays for Dataton’s Watchout software. Usually, I use it to calculate the softedge blending area between projectors, but with the softedge set to zero, it works for non-overlapping displays like LED walls too. I built it to be able to copy-paste layouts straight into Watchout, but one of the things on my to-do list is to create a direct export to AE. For the moment though, you can quickly export a list of the screen coordinates and crop values for a 16×5 wall of 1280×720 displays, and generate a lineup grid image while you’re at it.

I then imported the lineup image into AE and created a main comp at 20480x3600px with the grid image. Then I created a 1280×720 comp, dragged in the main comp, and then added an expression to the anchor point based on the comp name (which should be in the format ‘screen whatever 01’):


// Comp numbering should begin at 01.
grid_columns = 16; // Number of columns in grid.
grid_rows = 5; // Number of rows in grid.
screen_count_total = (grid_columns * grid_rows); // Total number of screens.
screen_number = thisComp.name.substr(-2,2); // Get screen number from last two characters of Comp name.
screen_id = parseInt(screen_number) - 1; // Parse chars to Int and change to zero-index value.
screen_row_id = Math.floor(screen_id / grid_columns); // Calculate row ID.
screen_column_id = (screen_id % grid_columns); // Calculate column ID.
screen_origin_x = (screen_column_id * thisComp.width) + (thisComp.width / 2); // Get screen horizontal pixel origin.
screen_origin_y = (screen_row_id * thisComp.height) + (thisComp.height / 2); // Get screen vertical pixel origin.
[screen_origin_x, screen_origin_y]


Duplicate the comp in the Project window until you have 80 comps, and each one will display the correct section of the main comp. Drag the all to the render queue and… go find something to do while AE renders your enormous main comp 80 times!

I think it would be faster to render out the master 20480x3600px comp as an image sequence, then re-import that into AE (or use it as a comp proxy) to render each screen.

Alternatively you could render only once, but add 80 output modules with different crop settings, but I don’t know how to automate setting that up. I wouldn’t be surprised if there is already a script or plug-in out there to do that…

Matthew Keane

Freelancer based in Paris, France
– Motion Graphics, Video Editing & Effects, Watchout Programming & Live Operation.

[Matthew Keane]

To be fair, the Zaxwerks Flag plugin can give good results for certains types of animations – try parenting a small banner to the motion-tracked movement of somebody waving their arms in the air, for example, and the material will flap back and forward in the air in a much more realistic way than other plugins I’ve played with.

But, as the original poster points out, it’s horribly unstable at high resolutions. I just about managed to coax some 4K renders out of it for a project by tweaking the render settings until it stopped crashing, but it was still very slow. For a very high-resolution render for a video wall, I would look at either one of the particle plug-ins, like Form or Stardust, for a stylised flag, or maybe Freeform Pro, if you need something more realistic-looking.

As Dave mentioned above, I would find out whether you actually need to deliver content at the final resolution, or whether a half-res, or even quarter-res render would be acceptable which would speed up the render times. If the display is a low-density pitch LED wall on the other side of a stadium, sometimes full-res renders are not necessary (or can be upscaled after rendering).

Matthew Keane

Freelancer based in Paris, France
– Motion Graphics, Video Editing & Effects, Watchout Programming & Live Operation.