At the end of the day I know how I CAN make this work but I wanted to reach out for advice on the best way to accomplish this.
I am building a modular matrix system of 7 panels wide and 4 panels tall with each panel having 49 pixels (7x7 vertical layout)
My question is what is the simplest, lowest time, to create the element organization, preview, and channel mapping. Also, this is spanning two controllers.
Here is the current way I think I can make this work but channel mapping will be highly time consuming.
The first image is the preview layout (vertical) and element organization (columns first), (49x28)
The second image is the starting channel per pixel, coloring is to keep panels and controllers straight.
Any help simplifying mapping and preview would be appreciated. Thanks
Those big video billboards and modern jumbotrons work almost exactly like this. They usually have more than 49 pixels though. Generally it's a multiple of 16x16 for ease of electronics. The difference is that the source doesn't route the signal, the electronics in each panel do it. Basically one uniform signal goes to all panels and contains the entire picture. And each panel is configured for its position in the matrix and then is able to extract only the data it cares about for its own use. The panels are connected serially (daisy chain) but since each panel receives and forwards the entire signal, it doesn't matter why order they're connected.
Vixen and most pixel controllers aren't designed for such an application. You can certainly get it done, but as you've identified, it's going to be a tedious patching process.
You can cut the grass with a pair of scissors, but a lawnmower is much easier.
--Jon Chuchla--
Sent from my iPhone using Tapatalk
The falcon team has designed a new controller for driving p10 panels. Beagle bone black $45, (BBB) plus I octoscroller board ($5) and running latest FPP software(free). It can drive 64 panels per BBB , each panel is 16x32 rob nodes. Each panel can be purchased from China for $11 shipped. Thus a 8x8 matrix would have 98,304 channels or 32,768 rub nodes. Your sequencer renders show into a fseq file, send it to the FPP, when show runs , data from sd card in BBB goes to panels via 40 parallel pins. No data is on network. Read about all of this on www.falconchristmmas.com. Dave Pitts and capt Murdoch made the controller and software.
My only input to them is this is causing shows around 100k channels and greater. We have added a 256x128 matrix to our regression suite
Sent from my iPhone using Tapatalk
I should have been more clear, I am running 2 SanDevice e682 to make this happen. I was looking for vixen tips to make this faster but thank you for the responses and this could be handy in the future.
Sent from my SAMSUNG-SM-G900A using Tapatalk
So are using something like a 8x8 or 8x16 ws2812. I have a smaller matrix like that. I'll let the vixen folk explain how to set it up, thanks
Sent from my iPhone using Tapatalk
If you're going to do it in vixen, It sounded like you got the concept right in your first post.
In vixen a matrix is a logical group of pixels in a certain arrangement. You set up the elements and preview in this order. The matrix will be one group and will start in the lower left corner. It will contain subgroups which then consist of the pixels that make up the rows, bottom to top.
Then if your actual channel mapping doesn't match, you need to set up the patching to match. In your case, based on the decisions you made in the hardware and wiring design, this is going to mean you'll need to patch in many steps, 7 pixels at a time.
...to be continued
--Jon Chuchla--
Sent from my iPhone using Tapatalk
...continued
I wanted to switch to my computer for the rest of the response.
I assume that each 7x7 panel is connected to one controller output.
Your second spreadsheet indicates that you're zigzagging the pixels up and then back down. It's easier to handle this in the controller rather than in Vixen's patching. The controller is much more efficient in this task. If you set your outputs for a zigzag of 7, then you can send the data from vixen straight thru without going back and forth. This would reduce 7 separate patching operations down to one and you won't need to alternate the reverse checkbox for every operation.
To patch the first panel, you'll select the first 7 pixels in the first 7 rows. You can multi-select by holding the control key and clicking the channels (in order) then select the first 147 controller channels. Then hit patch. Then move on to the first 7 pixels in rows 8-14 and select outputs 148-294. The key to the order is to do it in sACN universe/channel order. If you choose not to use zigzag on the controller, you'll have to do it only one column at a time and check the "reverse elements" button for every row that's going downward, and uncheck it for every row that's going up.
You can shortcut the process a bit and not select the individual output channels. You can instead click the controller itself (instead of the specific channels) and make sure the "unused channels only" option is checked. This will fill in the unused channels sequentially as you patch. So long as you're going in order, it will work out the same.
One thing I'm not 100% sure on is how the 682 handles zigzag across strings in a cluster. I don't know if it will return each output to start at the bottom, or if it handles all strings in the cluster as one unit. I'm hoping it starts each string anew. Since you have an odd number of zigs, the last and first column on each string are both going upward. I hope that's not an issue on the 682. changing to an 8x8 panel array would make things simpler. with binary counting that's used in software and electronics, increments of 2 always work better, and 8 is often even better.



