Intro to pixels hel...
 
Notifications
Clear all

Intro to pixels help

201 Posts
30 Users
0 Reactions
47.6 K Views
MartinMueller2003
Posts: 7613
(@martinmueller2003)
Illustrious Member
Joined: 14 years ago

First of all, the 'universe' concept has nothing to do with pixels or channels or anything else about simple lighting.

Short answer: A universe is a collection of up to 512 channels of lighting control.

Long Answer (with embellishments and simplifications to make it easier to understand):

The 'universe' concept is a hack that extends the original DMX lighting control specification. Here is a bit of history for you.

A long time ago, stage systems needed a way to control lights from a panel located away from the stage without having a huge number of wires running from the control panel to the dimmer packs. This was solved using a specification called DMX. It allowed you to put 512 channels of lighting control onto a single serial data line. That serial line ran from the lighting control board to the dimmer packs near the stage.

After a while, as shows became more and more complex, more serial lines were added and we again ended up with huge heavy bundles of wire running from the control panel to the stage. We needed a way to talk about the individual serial lines running everywhere. Eventually each serial line carrying 512 channels became a 'universe' (I cannot remember which update to the DMX spec introduced the term). NOTE: No mention of pixels or any such thing, just a need for an ever growing number of control channels.

At some point someone noticed this neat new communications infrastructure called 'Ethernet' that ran fast enough to handle a ton of data. This resulted in a specification known as the E1.31 (also sACN and a few other intermediate specifications) specification. Since the goal of the specification was to be easily understood by the existing stage electricians and to be easily transformed back into DMX serial data back at the stage end of the network, the terminology from DMX was simply mapped to the Ethernet packet. An Ethernet packet contains a DMX frame (the core of the original DMX serial data) and a universe wrapper plus some additional stuff.

So why do we keep talking about universes in these Christmas forums? Because it is easy to create an Ethernet network out in the yard and using an existing technology to move lighting control data across that network is simpler to do than to invent something new. A universe is simple an extension to the numbering that allows us to discuss / send / control more than 512 channels over Ethernet to a controller.

I would like to note that while some controllers extend the universe concept out to the ports to make it easy on the software to map things (this is where the 1-1, 1-2 nonsense comes from), some of the newer controllers are using it the way it was intended, as an extension to a numbering scheme that was initially limited to 1-512.

For example, the F16V2 maps universes and the channels within them into an internal contiguous set of channels. You then map those channels to pixels (or serial ports) and finally map the pixels to an output port, removing universes and the limitations they place on delivering data to controllers to mange lights.


Reply
Posts: 18
(@one787980)
Eminent Member
Joined: 11 years ago
MartinMueller2003 wrote:First of all, the 'universe' concept has nothing to do with pixels or channels or anything else about simple lighting.

Short answer: A universe is a collection of up to 512 channels of lighting control.

Long Answer (with embellishments and simplifications to make it easier to understand):

The 'universe' concept is a hack that extends the original DMX lighting control specification. Here is a bit of history for you.

A long time ago, stage systems needed a way to control lights from a panel located away from the stage without having a huge number of wires running from the control panel to the dimmer packs. This was solved using a specification called DMX. It allowed you to put 512 channels of lighting control onto a single serial data line. That serial line ran from the lighting control board to the dimmer packs near the stage.

After a while, as shows became more and more complex, more serial lines were added and we again ended up with huge heavy bundles of wire running from the control panel to the stage. We needed a way to talk about the individual serial lines running everywhere. Eventually each serial line carrying 512 channels became a 'universe' (I cannot remember which update to the DMX spec introduced the term). NOTE: No mention of pixels or any such thing, just a need for an ever growing number of control channels.

At some point someone noticed this neat new communications infrastructure called 'Ethernet' that ran fast enough to handle a ton of data. This resulted in a specification known as the E1.31 (also sACN and a few other intermediate specifications) specification. Since the goal of the specification was to be easily understood by the existing stage electricians and to be easily transformed back into DMX serial data back at the stage end of the network, the terminology from DMX was simply mapped to the Ethernet packet. An Ethernet packet contains a DMX frame (the core of the original DMX serial data) and a universe wrapper plus some additional stuff.

So why do we keep talking about universes in these Christmas forums? Because it is easy to create an Ethernet network out in the yard and using an existing technology to move lighting control data across that network is simpler to do than to invent something new. A universe is simple an extension to the numbering that allows us to discuss / send / control more than 512 channels over Ethernet to a controller.

I would like to note that while some controllers extend the universe concept out to the ports to make it easy on the software to map things (this is where the 1-1, 1-2 nonsense comes from), some of the newer controllers are using it the way it was intended, as an extension to a numbering scheme that was initially limited to 1-512.

For example, the F16V2 maps universes and the channels within them into an internal contiguous set of channels. You then map those channels to pixels (or serial ports) and finally map the pixels to an output port, removing universes and the limitations they place on delivering data to controllers to mange lights.

Thanks for the detail explanation. That leads to some more questions:
1 - So if the port can support 680 pixels, I hava a model that is only 500 pixels (matrix for example), what happens to the 180 left over pixels? Can I have to vertical strings of 60 pixels, controller separately from matrix?
2- what if the matrix is let's say 779 pixels, can I have another matrix start at the end of the first one to save unused ports on channel 2. And how would I wire the matrix it the pixel count is more that the max for the single port?


Reply
MartinMueller2003
Posts: 7613
(@martinmueller2003)
Illustrious Member
Joined: 14 years ago

Thanks for the detail explanation. That leads to some more questions:
1 - So if the port can support 680 pixels, I hava a model that is only 500 pixels (matrix for example), what happens to the 180 left over pixels? Can I have to vertical strings of 60 pixels, controller separately from matrix?
2- what if the matrix is let's say 779 pixels, can I have another matrix start at the end of the first one to save unused ports on channel 2. And how would I wire the matrix it the pixel count is more that the max for the single port?

Some of these questions are implementation dependent. Every display has a few common elements: Something that knows about a ton of channels and plays the sequence, a mechanism for delivering the sequence data to a controller connected to lights and the controller that is connected to the lights.

When designing a display element, you need to take a few factors into consideration, one of those is the number of lights that will be in a single string. This is actually a pretty important decision. For example: I have a set of arches that have 180 pixels on each arch. I set the arches up as a single string and ran a color change sequence on them. After about 250 pixels, there is a visibly noticeable lag between the first pixel and 250th pixel changing color. For this reason, I rarely use a string with greater than 200 pixels. FYI: I found that this effect is different between the various pixel controllers so a bit of testing using your equipment will be needed to determine the number of pixels you want on a single string.

In most cases, if a controllers output supports a large number of pixels and you send it data for a lower number of pixels, then the controller only sends as much data as you send to it. So a controller port that supports 640 pixels with a 50 pixel string attached to it, you only send 150 channels of data to the port and all is well. Same is true for sending 40 pixels worth of data to that same string. The last 10 picels will stay unlit (in reality there are noise issues that could cause those pixels to have random output).

Next, lets say I have a matrix with 1000 pixels. I have connected these pixels to 5 ports on my pixel controller with 200 pixels on each ports. Then (I will use Vixen 3 for this example) create a matrix in my sequencer, I patch the matrix pixels to a set of 3000 contiguous channel. I then set up the output controllers to each be sent 600 channels. The important concepts here are that the controllers connected to the pixels do not know nor care about each each other and that all coordination between all of these channels is done by the sequencer (the sequencer is the only place the concept of a matrix exists. To all other parts of the system there are only channels).

Hope this helps


Reply
Posts: 18
(@one787980)
Eminent Member
Joined: 11 years ago

So if there is a noticeable lag after 200 pixels, why do people build controllers that can support up to 680 pixels. Isn't this a waste of 480 pixels?
When I asked, how many controllers would take to light mega tree mentioned at the beginning of this post, MindTrick said that one Falcon can do it. So I am lost here. If we can use 200 pixels out of one port, so 200 *16= 3200 pixels that not even near that this mega tree is build of....


Reply
MartinMueller2003
Posts: 7613
(@martinmueller2003)
Illustrious Member
Joined: 14 years ago

There is noticeable lag on the controller I tested (early PX1). I have not tested other controllers so ymmv. I use the WS2811 pixels. The WS2811 specification allows for multiple microsecond gaps between sending data bits to the pixels. With software output loops, there are built in delays and other activities going on at the same time data is being sent to the pixels. This can result in overall slower update times over the entire string. Other designs use FPGAs (think of them as hardware accelerators) to generate the output bit stream to the pixels and will typically have less lag between the start and end of the string. I did mention that you would need to test your hardware to determine its limits. For my display design I simply went with a lowest common denominator model that I know works. I suspect the F16V2 and other FPGA based controllers will support far longer strings before the lag becomes noticeable to the human eye. This is the type of detail where the quality and capability / capacity of the controller becomes important.

Another thing to watch out for is the network you use to deliver the data to the controller. An Ethernet network should not be loaded above 60% of its rated capacity. At about that point, network delays become noticeable in your display elements as the same type of lag described above. Keep in mind that in a typical E1.31 implementation you can only get abut 170 pixels worth of data in a single Ethernet packet. This means that a 171 - 341 pixel string requires two Ethernet packets to be sent to the controller. Variation in the network delivery time may cause that second packet to arrive a bit slower than expected and cause the same sort of visible lag we have been discussing. As network utilization goes up, so to does the variation in the delivery times as packets wait for other packets to traverse the network.


Reply
Page 39 / 41
Share: