Author Topic: Colorlight board with BeagleBone possibility?  (Read 4961 times)

Offline brichi

  • Hero Member
  • *****
  • Join Date: Dec 2017
  • Location:
  • Posts: 528
  • Kudos: 1
Re: Colorlight board with BeagleBone possibility?
« Reply #75 on: February 14, 2018, 05:24:25 PM »
see, its what i do, LOL


glad to test out when you need

Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 9,856
  • Kudos: 214
Re: Colorlight board with BeagleBone possibility?
« Reply #76 on: February 14, 2018, 05:27:15 PM »

LOL... you found a bug... The code on master doesn't add the *input* configuration to the backup.  :)     Captain?   

I will get that updated.  It needs to backup everything in the config directory to catch the other new JSON config flies as well.
-
Chris

Offline Cjaqua

  • Jr. Member
  • **
  • Join Date: Nov 2015
  • Location:
  • Posts: 82
  • Kudos: 2
Re: Colorlight board with BeagleBone possibility?
« Reply #77 on: February 14, 2018, 07:24:05 PM »
I ran a 72 universe p5 matrix on the latest master-alpha BBB version with a colorlight card and everything looked good. Must be something that rears its head when you get to 84 universes.

Offline brichi

  • Hero Member
  • *****
  • Join Date: Dec 2017
  • Location:
  • Posts: 528
  • Kudos: 1
Re: Colorlight board with BeagleBone possibility?
« Reply #78 on: February 14, 2018, 07:57:56 PM »
especially in bridge mode, when i run in standalone withe an uploaded sequence like a 30 second "butterfly" effect sequence it looks perfect, once i run from xlights to bridge then it stutters a lot. Ill wait till the next release and do more testing and take videos of whats going on

Offline brichi

  • Hero Member
  • *****
  • Join Date: Dec 2017
  • Location:
  • Posts: 528
  • Kudos: 1
Re: Colorlight board with BeagleBone possibility?
« Reply #79 on: February 15, 2018, 07:56:34 AM »

LOL... you found a bug... The code on master doesn't add the *input* configuration to the backup.  :)     Captain?   

I will get that updated.  It needs to backup everything in the config directory to catch the other new JSON config flies as well.


just out of curiosity was that uploaded yesterday? I see a 2018-02-14 21:37  796M upload but not sure if thats the fix 

Offline dkulp

  • Moderator
  • *****
  • Join Date: Sep 2013
  • Location: Framingham, MA
  • Posts: 1,586
  • Kudos: 92
    • KulpLights.com
Re: Colorlight board with BeagleBone possibility?
« Reply #80 on: February 15, 2018, 08:40:56 AM »

just out of curiosity was that uploaded yesterday? I see a 2018-02-14 21:37  796M upload but not sure if thats the fix 


No... that just flips from Apache to NGINX to match the latest Pi images.

Offline brichi

  • Hero Member
  • *****
  • Join Date: Dec 2017
  • Location:
  • Posts: 528
  • Kudos: 1
Re: Colorlight board with BeagleBone possibility?
« Reply #81 on: February 15, 2018, 08:44:00 AM »
ok, ill wait to hear from you then to test any changes in a future update. Thank you!!

Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 9,856
  • Kudos: 214
Re: Colorlight board with BeagleBone possibility?
« Reply #82 on: February 15, 2018, 09:13:34 AM »

LOL... you found a bug... The code on master doesn't add the *input* configuration to the backup.  :)     Captain?   

I will get that updated.  It needs to backup everything in the config directory to catch the other new JSON config flies as well.


just out of curiosity was that uploaded yesterday? I see a 2018-02-14 21:37  796M upload but not sure if thats the fix 

No, yesterday was V day, so no coding last night.  The fix for this will be in the github repo, so you will just need to run a manual update once I push a change.  I opened a ticket to track this,   https://github.com/FalconChristmas/fpp/issues/342

Offline brichi

  • Hero Member
  • *****
  • Join Date: Dec 2017
  • Location:
  • Posts: 528
  • Kudos: 1
Re: Colorlight board with BeagleBone possibility?
« Reply #83 on: February 15, 2018, 09:19:12 AM »
yeah, my wife was getting upset because i was sitting on the Zoom meeting last night with everyone, LOL

Offline brichi

  • Hero Member
  • *****
  • Join Date: Dec 2017
  • Location:
  • Posts: 528
  • Kudos: 1
Re: Colorlight board with BeagleBone possibility?
« Reply #84 on: February 17, 2018, 02:52:17 PM »



I had to give up on the 5A, I cant do 9 rows of panels with it unless you guys can see if im missing anything,

in the LED Vision software the most I can choose is 128 for height but with 9 panels its 144

Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 9,856
  • Kudos: 214
Re: Colorlight board with BeagleBone possibility?
« Reply #85 on: February 17, 2018, 05:06:43 PM »



I had to give up on the 5A, I cant do 9 rows of panels with it unless you guys can see if im missing anything,

in the LED Vision software the most I can choose is 128 for height but with 9 panels its 144

The layout in LED Vision and the layout in FPP do not have to be the same, FPP can do several things that the ColorLight doesn't allow by itself.  In LED Vision, you just need to connect up the panels and configure them so that you get lights on each panel.  Then, in FPP, you will go in and tell FPP about the orientation of each panel and it's location in the overall matrix, which output each panel is on, and which panel number the panel is on that output.  FPP does all the calculations to put each pixel's data in the correct location in the packets sent to the ColorLight receiver.   My P10 matrix is made up of 14 panels.  For testing, I switch between having it connected as 2 chains of 7 panels or one chain of 14 panels.  The actual layout of the matrix is 112x64.  I have some panels in their normal orientation, some upside down, and some turned on their side to fit the 46" wide window that the matrix sits in.  I couldn't use 4 panels horizontal because that would be too wide for the window and 3 panels wide would have been too skinny, so I used an odd configuration with some horizontal and some vertical.  I could have turned them all on their side, but I didn't want to do that because of how I built the frame for the matrix.

Offline brichi

  • Hero Member
  • *****
  • Join Date: Dec 2017
  • Location:
  • Posts: 528
  • Kudos: 1
Re: Colorlight board with BeagleBone possibility?
« Reply #86 on: February 17, 2018, 05:13:59 PM »



I had to give up on the 5A, I cant do 9 rows of panels with it unless you guys can see if im missing anything,

in the LED Vision software the most I can choose is 128 for height but with 9 panels its 144

The layout in LED Vision and the layout in FPP do not have to be the same, FPP can do several things that the ColorLight doesn't allow by itself.  In LED Vision, you just need to connect up the panels and configure them so that you get lights on each panel.  Then, in FPP, you will go in and tell FPP about the orientation of each panel and it's location in the overall matrix, which output each panel is on, and which panel number the panel is on that output.  FPP does all the calculations to put each pixel's data in the correct location in the packets sent to the ColorLight receiver.   My P10 matrix is made up of 14 panels.  For testing, I switch between having it connected as 2 chains of 7 panels or one chain of 14 panels.  The actual layout of the matrix is 112x64.  I have some panels in their normal orientation, some upside down, and some turned on their side to fit the 46" wide window that the matrix sits in.  I couldn't use 4 panels horizontal because that would be too wide for the window and 3 panels wide would have been too skinny, so I used an odd configuration with some horizontal and some vertical.  I could have turned them all on their side, but I didn't want to do that because of how I built the frame for the matrix.


right, I set it all up like that but the 9th row of 6 do not work, the just mirror almost whats on row 8, so even show I have output 8 set to 1-12, its actually putting out 1-6 and then the bottom row is showing the same like panel 8 and 2 are duplicating , 3 and 9 and so on.


heres a few pics of my panel layout and the results on the screen


https://drive.google.com/file/d/1OMD-G8yFSjxzoYLTfpfa2PBwiIn3EzEY/view?usp=sharing
https://drive.google.com/file/d/1w1X0iqroCxJiF_ZHcU6V47bF1uZxnP2j/view?usp=sharing
https://drive.google.com/file/d/1Nz_mpzu9jnVGMpy5WqPc0vPKnt3rbQCm/view?usp=sharing
https://drive.google.com/file/d/1N8Pe1yhJCrOdasXl9FnUuxjKJUt82xT9/view?usp=sharing
« Last Edit: February 17, 2018, 05:26:46 PM by brichi »

Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 9,856
  • Kudos: 214
Re: Colorlight board with BeagleBone possibility?
« Reply #87 on: February 17, 2018, 05:49:54 PM »
right, I set it all up like that but the 9th row of 6 do not work, the just mirror almost whats on row 8, so even show I have output 8 set to 1-12, its actually putting out 1-6 and then the bottom row is showing the same like panel 8 and 2 are duplicating , 3 and 9 and so on.

In LED Vision, did you configure the display as 384x128 so that LED Vision would recognize that you have 12 panels on the 8th output?  That layout is far from optimal when using a ColorLight or Linsn card since we have to send data for every row for the whole width of the longest row which is 384 in your case.  So for most rows, we are sending twice as much data as needed.  It is the same number of packets, they are just larger.    Since FPP supports the panels in any orientation and layout regardless of which actual output they are on on the receiver, octoscroller, Pi matrix hat, etc., you may want to consider recabling your panels to make the chain lengths closer to equal.  Even if you don't want to turn panels upside down, you could still get creative in the layout, possibly even just having 6 chains of 9 panels with each column on its own output.

Offline brichi

  • Hero Member
  • *****
  • Join Date: Dec 2017
  • Location:
  • Posts: 528
  • Kudos: 1
Re: Colorlight board with BeagleBone possibility?
« Reply #88 on: February 17, 2018, 06:01:09 PM »
right, I set it all up like that but the 9th row of 6 do not work, the just mirror almost whats on row 8, so even show I have output 8 set to 1-12, its actually putting out 1-6 and then the bottom row is showing the same like panel 8 and 2 are duplicating , 3 and 9 and so on.

In LED Vision, did you configure the display as 384x128 so that LED Vision would recognize that you have 12 panels on the 8th output?  That layout is far from optimal when using a ColorLight or Linsn card since we have to send data for every row for the whole width of the longest row which is 384 in your case.  So for most rows, we are sending twice as much data as needed.  It is the same number of packets, they are just larger.    Since FPP supports the panels in any orientation and layout regardless of which actual output they are on on the receiver, octoscroller, Pi matrix hat, etc., you may want to consider recabling your panels to make the chain lengths closer to equal.  Even if you don't want to turn panels upside down, you could still get creative in the layout, possibly even just having 6 chains of 9 panels with each column on its own output.


you are a genius my friend! the 384 was what fixed it, I had it on 192 being that was "real" amount of pixels from left to right but I see now that I needed the pixel count of the entire panel and then FPP takes care of it all, THANK YOU!!


and I also want to add that the latest Pi and BBB updates seemed to have cleared up my screen a lot! I still get the errors on the status page for live packets at U84+ but its definitely more clean on the image side

Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 9,856
  • Kudos: 214
Re: Colorlight board with BeagleBone possibility?
« Reply #89 on: February 17, 2018, 06:07:42 PM »
the 384 was what fixed it, I had it on 192 being that was "real" amount of pixels from left to right but I see now that I needed the pixel count of the entire panel and then FPP takes care of it all, THANK YOU!!

Great.  One thing to keep in mind is that this is the equivalent of sending 128 universes of E1.31 since the ColorLight protocol uses one packet per LED row.  This is why I need to optimize the ColorLight code to use the same sendmmsg() call that Dan added to the E1.31 Channel Output when he optimized the E1.31 code.  This is where the Linsn receiver actually can be more optimal since it packs the data into large packets, so it can send fewer packets than rows.

 

Back to top