conference session: I didn't know ACN could do that!

[gafftaper]

I didn’t know ACN could do that!

Conference session with: Alan Martello, Daniel W. Antonuk, Kevin Loewen, Ulrich Kunkel, Gary Douglass

Recent History
AMX in the late 70’s 192 channels
DMX 1986, 1990, 1996, 2004
RDM-Takes DMX and makes it bi-directional communication.

Why Go further? Why go to Ethernet distribution when we have DMX and RDM?
Network infrastructure exists and is easy to use.
Equipment is cheap and easily available.
DMX gear is expensive and difficult to get (impossible on a Saturday night at 7pm).
Art-Net, gave us a network system but it is too proprietary, didn’t scale well, doesn’t play well with others

Solution: ACN

Goals:
-Works on existing networks
-Scalable
-Plug and Play

ANSI Standard E1.17 (ACN)
It was optimized for single sources communicating with many devices. It’s a good network citizen. It’s based on a property/value model… Each device has multiple properties and each property has a value

Full ACN had problems Problems
Complicated compared to DMX
Requires too much processing power
Requires too much memory
So they created a stripped down version called Streaming ACN ANSI Standard E1.31

BUT streaming ACN was one directional which fails to meet a huge part of the ACN goals. It was a step back from RDM. So now we have E1.33 Which is Streaming ACN plus RDM.

The Current Status
The new ETC consoles all use FULL ACN for configuration, management, and status purposes.

ETC, Pathport, Strand, Barco/HES use Streaming ACN for actual control purposes.

No one is yet using the RDM over ACN standard because it is not complete yet.

Why isn’t there more E1.17 Equipment?
ACN consoles and devices are more difficult to make. More difficult means higher cost and more time to develop. Chicken and Egg problem with the consoles and devices… both types of manufacturers don’t feel the need to spend the money until there is a need from the other. In the end the biggest problem is that it’s new and users aren’t demanding it yet.
The Adoption rate of E1.31 Streaming ACN is quite high.
It’s just DMX over Ethernet so it’s easy for manufacturers to adopt. It’s cheaper to adopt, It’s easy for users to put into effect with their old system. The adoption rate of E1.33 (RDM over ACN) will probably be fast once it is fully completed. The adoption rate of Full ACN will probably continue to be slow.

What followed was a fascinating description of how ACN was used to operate a very complex exhibit in the German pavilion for EXPO 2010 in Shanghai. This was a project that simply would be impossible with DMX. There is a 9 meter diameter, 1.3 ton sphere with over 300,000 LED’s on it giving near HD quality. The video is run via ACN. The motion control system that flies the ball, controlled by ACN. Moving lights talk to the sphere control via ACN so that they know where the sphere is (so they can move around the room without hitting the giant sphere). Latency problems and different speeds of moving lights create problems for the lights to track so the sphere control projects 2 seconds ahead of where the sphere will be. Audio can be triggered by the position of the sphere AND the media servers can be can be triggered by audio system input (Control can be either way). The audio system had a set of microphones located around the room that can track the loudest sound in the room and tell the motion control system to move the sphere to that sound. There are skateboarders moving around the room with 36 moving lights tracking them automatically… all communicate together with ACN.

There is amazing video of this system in use on Youtube. I’ll post a link later.

 

[JohnHuntington]

Thanks for the excellent write up! I was supposed to be on this panel (and was looking forward to it) but went to the Ithaca College reunion instead.

I found a couple videos, if you find more that would be great:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

YouTube - The Pendulum for the German Pavillion "Balancity" at the EXPO 2010 in Shanghai

Here's a news report:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

YouTube - Shanghai World Expo 2010 - German Pavilion unveils "Balancity"

How many people were in attendance? We need to get people moving forward on ACN to do exactly cool stuff like this.

Thanks!

John Huntington
- John's Blog Main Page

 

[gafftaper]

Sadly there were only about 30 people in attendance. The ACN and RDM sessions were just the sort of thing we need to educate the customers and begin creating a real demand for these products.

 

[otherwho]

This was a project that simply would be impossible with DMX. .... 300,000 LED’s on it giving near HD quality.

not impossible, just much harder(and expensive) . 576 universes of DMX would needed. The moving light not hitting the move ball is a simple 3D projection problem (most computer/xbox360 games do 8+ lights of compiled shadows real time, a simple thing like see if a light beam hit can be 300+ lights at 120Hz.) The central show system would likely would like the one that it used in the real place. The only real problem is equipment failure troubles. Most of the problem is same in DMX and ACN it just that DMX is starting to show it age, this makes DMX to costly to do the same effect

HD is 1280x720 that is 921,600 Pixels or at 3 light for color(RGB) 2,764,800 Leds need for HD 720p video. 1080p needs 2,073,600 Pixels (6,220,800 LEDS) So 300,000 LED about 11% of led need for HD 720p. That is not even close to HD.

 

[gafftaper]

not impossible, just much harder(and expensive) . 576 universes of DMX would needed. The moving light not hitting the move ball is a simple 3D projection problem (most computer/xbox360 games do 8+ lights of compiled shadows real time, a simple thing like see if a light beam hit can be 300+ lights at 120Hz.) The central show system would likely would like the one that it used in the real place. The only real problem is equipment failure troubles. Most of the problem is same in DMX and ACN it just that DMX is starting to show it age, this makes DMX to costly to do the same effect

Yes some aspects of this exhibit could have been done with DMX (and effectively were since they used ACN 1.31 Streaming for many of the command tasks). But, DMX is one way communication. All DMX can do is say "1 at 0-256, 2 at 0-256, 3 at 0-256," and so on (technically it isn't even that complex). DMX can't ask, "Where will the ball be in 2 seconds? Where will the actor be in 2 seconds? What movers do I have in that area that can hit the actor without hitting the ball?" Those other systems can send data to a DMX console all day long, it simply is not going to hear the answer coming back.

Let's say we sat up a computer to do the tracking and the calculations of which lights to use and where to move them. How is that data going to get into the GrandMA so it can tell the movers where to go? You can't do that with DMX.

HD is 1280x720 that is 921,600 Pixels or at 3 light for color(RGB) 2,764,800 Leds need for HD 720p video. 1080p needs 2,073,600 Pixels (6,220,800 LEDS) So 300,000 LED about 11% of led need for HD 720p. That is not even close to HD.

I'm not a video guy and these weren't my numbers, but I believe he was not talking about the strict definition of HD. Instead he was talking about the appearance to an audience member who is a distance away. It's the same idea as a person buying a 32" TV that he always watches from 15' away should buy a 720p because he will never see the detail of a 1080p at that distance. But that's my interpretation of what was said, I can't say exactly what was intended.

 

[otherwho]

All DMX can do is say "1 at 0-256, 2 at 0-256, 3 at 0-256," and so on (technically it isn't even that complex). DMX can't ask, "Where will the ball be in 2 seconds? Where will the actor be in 2 seconds? What movers do I have in that area that can hit the actor without hitting the ball?" Those other systems can send data to a DMX console all day long, it simply is not going to hear the answer coming back.

DMx is 0-255 or 8 bits of data per address and can send 512 address at 43Hz that 179,224 bits per second => ~=179Kb/s.(Kbits/s) That digital data can be used for data about where a actor/ball is at/velocity/acceleration/etc. Why does does in ball need to know about what the movers are lighting? Why do we need answer coming back? If the movers hit the spot that the DMX tells them everything is fine.

Let's say we sat up a computer to do the tracking and the calculations of which lights to use and where to move them. How is that data going to get into the GrandMA so it can tell the movers where to go? You can't do that with DMX.

The data leaving the computer is DMX data for the movers.

 

[gafftaper]

Oops yes DMX is 0-255... I keep telling myself I have to stop posting numbers late at night.

I see what you are thinking. Essentially this could all be setup as a long string of one way communication. One device is always simply telling the next device what to do and that does work for most tasks. However with this system you have multiple devices that need to know the position of the sphere, the location of two actors, the location of the loudest sound source in the room, the position of the movers, and what point in the show the media servers are on... that's just too many devices that need information going both ways. As soon as you say audio can trigger motion control and motion control can trigger audio, the usefullness of DMX is gone.

 

[JChenault]

Yes some aspects of this exhibit could have been done with DMX (and effectively were since they used ACN 1.31 Streaming for many of the command tasks). But, DMX is one way communication. All DMX can do is say "1 at 0-256, 2 at 0-256, 3 at 0-256," and so on (technically it isn't even that complex). DMX can't ask, "Where will the ball be in 2 seconds? Where will the actor be in 2 seconds? What movers do I have in that area that can hit the actor without hitting the ball?" Those other systems can send data to a DMX console all day long, it simply is not going to hear the answer coming back.

Let's say we sat up a computer to do the tracking and the calculations of which lights to use and where to move them. How is that data going to get into the GrandMA so it can tell the movers where to go? You can't do that with DMX.

First - I am a fan of ACN - but I think there may have been a bit of hype in the presentation. ( I was unable to get to it unfortunately).

As I read the spec, ACN could tell you the current position of the ball, and tell you where it has been told to go - but for your use case, something else is going to have to understand where the ball is in the room, where the movers are in the room, and where the movers should point to avoid the ball. ACN cannot predict the future, nor does it have a three dimensional model of the room magically built in.

ACN is a messaging protocol. It is not smart enough to understand three dimensional space. Now if the device it is talking to understands 3d space, then it could transmit that devices information - but that's about it.

The original vision of ACN (IMHO) tried to do too much and be too general. It suffered from the second system syndrome ( See Brooks The Mythical Man Month ). I personally don't expect that the full version will be successful, although I hope that some of the reduced versions will get out there and become a new standard.