How do I do things ...
 
Notifications
Clear all

How do I do things with matrices

20 Posts
6 Users
0 Reactions
3,160 Views
Posts: 6728
Topic starter
(@p-short)
Illustrious Member
Joined: 20 years ago
[#58655]

I've designed a 2720-pixel, 4-Port ESP8266-based controller using a μSD card as the data source, and I want to do some testing of it with a 25x32 (HxW) matrix using seed pixels (don't laugh, Ernie) with a 2" spacing (aspirational). It'll be installed on my 4'x5' (HxW) front picture window, and I intend to build it using pixel strings hanging from a curtain rod across the top of the window.

The first few questions beg the issue of pixel spacing:
1) How do I implement a horizontal single-pixel chase effect when the strings are oriented vertically?
2) How do I implement a single-pixel chase effect at an arbitrary angle?

In the vixen 'Setup Display' window I've defined the controller as a 8160-channel generic serial controller. I've selected the generic serial controller as a place-holder, as I've written a Windows program that takes exported csv files and writes the data to files for copying to μSD cards, and I've chosen 8160 as the size because 2720 is the maximum number of pixels on the controller. The grid, on the other hand, has 800 pixels. Right now I patch the first 200 pixels to channels 1-600, the next 200 pixels to channels 2041-2640, etc, which leaves large gaps in the list of controller channels. When exporting the csv file I get a warning message about needing all of the controller channels fully patched for the export to produce correct results. What is likely to be incorrect about the csv export file?

Lastly - my pixel strings have 4" spacing, but I need/want 2" spacing in the window prop. One way to achieve this is to put loops in the wires between each pixel (either with cable ties or thingies from Boscoyo, etc). Another way is to 'fold' each pixel string at the 48" point (12/13 pixels) and bring it back up with the pixels on the rising leg interleaved with the pixels on the falling leg, so there are pixels every 2". This would place the even-numbered pixels on one leg, and the odd-number pixels on the other leg. Is there any feature in Vixen that would make this feasible without the need for extremely pixel-by-pixel patching?


19 Replies
ErnieHorning
Posts: 4434
(@erniehorning)
Famed Member
Joined: 19 years ago

I'm surprised nobody's made a single comment all day. I'm planning on two 45" by 47" matrices' this year. Both will be 1-inch spaced Seeds (5-volt even), 705 pixels on 3 ports using a ESP32-Wroom controller, 2115 total pixels. I ordered 1-inch spacing specifically so I wouldn't have to double back. Every 47th pixel, the vertical string will be reversed. In xLights this was pretty easy but I thought I'd seen a post or two referencing how to do it in Vexen 3 also, though I don't believe it was as straight forward.

So far, I've only set it up in xLights and the layout screen shows a representation of the matrix. The graphics fully function on screen and the controller is supposedly doing it's thing. I've received the 1-inch Seeds and will probably start working on the matrices' soon, hoping for the best.

I would expect that patching the 25" by 32" would be a seperate prop since you'd want it in the farthest corner from the start so it'll represent the pixels that will mostly likely show issues, if the controller can't keep up.


Reply
MartinMueller2003
Posts: 7615
(@martinmueller2003)
Illustrious Member
Joined: 14 years ago

I think Phil is looking for a less intensive way of achieving rows and columns than doing it in the setup by manually copying matrix pixels into a set of special models. I was hoping one of the Vixen devs had implemented some new mechanism that would allow us to skip that rather time consuming operation. I have seen other very helpful things appear in Vixen so I know my knowledge is way behind the times.


Reply
jeffu231
Posts: 2458
(@jeffu231)
Famed Member
Joined: 15 years ago

The Wipe effect is very good for doing horizontal type chases when items are not structured for the conventional Chase effect. If you think that you would be heavy on horizontal type effects over vertical ones, then you could model the Matrix in Vixen as a horizontal one and select the options to patch it in zigzags to match the physical layout. You should be able to do that with the Grid setup wizard to create it vs just manually making the elements.

I have been working on some solutions for challenges like these that would make it easier to build what you would typically create as additional groups for creative purposes. In essence model the prop as it physically is, and then make creative groups on the fly at either the global level or at the sequence level. It is still in the early stages, so I don't have a delivery target yet.


Reply
Posts: 6728
Topic starter
(@p-short)
Illustrious Member
Joined: 20 years ago

It would be nice if there was a section in the documentation to describe how to use matrices and to implement various effects on them. Or does that exist and I just haven't found them yet?


Reply
jeffu231
Posts: 2458
(@jeffu231)
Famed Member
Joined: 15 years ago

There is not anything in depth on Matrixes and using effects. They are similar to Mega Trees in some aspects. A Mega Tree is just a curved Matrix and look the same to the effect logic. Each effect has general documentation and the Layer mixing sections have a couple examples using a Mega Tree.

If you have some ideas about what you might find useful in this area in the documentation, we can add it as an item to enhance.


Reply
MartinMueller2003
Posts: 7615
(@martinmueller2003)
Illustrious Member
Joined: 14 years ago

From my point of view, the current mechanism gives us easy row OR column access. There is a need to be able to apply effects using either row or column in the same sequence. For example: I have a 48 x 96 pixel matrix. I wanted to map a simulated window outline on the grid that is 3 pixels wide and includes a bar through the middle. I had to create groups for each of the side + top + bottom + middle. Then I manually copied the pixels that made up these "lines" on the matrix. Now I apply effects to the window frame groups just like I do for my other windows. This was a very manually intensive endeavor. I think the OP is asking for a simpler way to partition a matrix into other groups so we can use it in innovative ways. An extension of this would be a request I had from a user for an idea for configuring a picture in picture effect. In essence we had to again copy pixels into groups that make up the various regions for the multiple effects.


Reply
dirknerkle
Posts: 10311
(@dirknerkle)
Illustrious Member
Joined: 18 years ago
jeffu231 wrote:
There is not anything in depth on Matrixes and using effects. They are similar to Mega Trees in some aspects. A Mega Tree is just a curved Matrix and look the same to the effect logic. Each effect has general documentation and the Layer mixing sections have a couple examples using a Mega Tree.

If you have some ideas about what you might find useful in this area in the documentation, we can add it as an item to enhance.

Many of us old fogies, (i.e. AC string people) who are having a bit of trouble going from a time/channel based sequencing screen to xLights or Vixen3 struggle with the new sequencers because of concepts like this. "Similar to MegaTrees" to a channel-based guy means "Oh good, 24 channels and I'm good to go." The channel-based guy thinks of a MegaTree sequencing as merely animated by illuminating each string in either a clockwise or counter-clockwise direction. This is an entire world different than what a pixel matrix megatree is because of the physical construction of the light strings, but our language gets in the way of understanding.

For many, it's easier to grasp a concept where a "matrix-megatree" (new term?) still enjoys a conical shape, modeled after an evergreen or spruce tree, with an X-Y matrix of pixel lights is simply bent around it. This may help the user envision a matrix-megatree as if it's just a flat matrix laying on the floor of the living room.


Reply
MartinMueller2003
Posts: 7615
(@martinmueller2003)
Illustrious Member
Joined: 14 years ago

Been thinking about this. I think an initial step would be to tak a model from the oled display libraries that support virtual screen / regions on the larger display. In this case we create a group and define the two corners to create a virtual region. Then Vixen makes copies of the correct pixels in the matrix and populates them into the group.


Reply
jeffu231
Posts: 2458
(@jeffu231)
Famed Member
Joined: 15 years ago
MartinMueller2003 wrote:
Been thinking about this. I think an initial step would be to tak a model from the oled display libraries that support virtual screen / regions on the larger display. In this case we create a group and define the two corners to create a virtual region. Then Vixen makes copies of the correct pixels in the matrix and populates them into the group.

I have this same issue. I have many creative groups that I use and they can take time to construct and there may be variants that I would only use in one sequence. This then makes my display structure very cluttered and hard to manage. So the task is to be able to construct those creative groups with some forms of automation that can make it easier, and then also let the user decide if they are just for the sequence at hand, or something they might use more globally. I had some POC version of logical target filtering before that were very helpful in creating them for someone very logically inclined. We thought at the time that it was just a different form of difficult to work with it for those that did not think in a very technical way. I am still looking for good user friendly ways to to approach this with some recent things I am working on. I have these same challenges in my display, so I am motivated to make improvements.


Reply
Page 1 / 2
Share: