So I understand the number of timing marks is important...

Discussion in 'LightShow Pro (LSP)' started by nutz4lights, Dec 13, 2014.

  1. nutz4lights

    nutz4lights Full Time Elf

    Joined:
    Dec 12, 2012
    Messages:
    305
    Likes Received:
    0
    Location:
    Melbourne, Florida
    Hey all,

    I have heard Eddy reference this many times (most recently in one of my posts where I was battling poor performance out of the scheduler). The suggestion is that, as you create macros, even if you start with ZERO timing marks when you create the sequence, you are always adding timing marks with the macro creation. Eddy's recent suggestion to me was, when creating "whole RGB display" effects, which for my display is around 30,000 channels, should be done with a frames per second setting of 8-10 instead of 15. Basically cutting the number of timing marks created in the grid below in half.

    As I was thinking about this a little bit more, something hit me and it is confusing the @#$% out of me right now.

    Due to challenges with the sequencing I have been trying to do, I have so far limited myself to creating 1-2 minute sequences wherein all the effects are "whole display" macros. So, if I start with a blank sequence, no timing marks, and I go through and create, let's say, nine x 10 second macro effects in a 90 second sequence, using 10FPS as the macro setting, I should wind up with 90 x 10 timing marks = 900 timing marks. What I have seen happen, is, I don't get perfect alignment between one macro to the next, so where the macro ends and a new one begins, I end up with two timing marks instead of one on top of each other. That's what got me thinking...

    ... If I was sequencing the way that most everybody sequences (hence my question to y'all), I would probably end up wanting to do something like the following:

    From time interval 0:35 to 0:45 (10 seconds):

    macro #1 applied to the megatree
    macro #2 applied to the minitrees
    macro #3 applied to the twig trees
    macro #4 applied to the tall palm trees
    macro #5 applied to the roof (if it wasn't dead because of the @#$% Technicolor pixels!!)
    macro #6 applied to ornaments
    macro #7 applied to the wreaths
    macro #8 applied to bushes, short palm trees, and oaks
    macro #9 applied to Merry Christmas sign

    You might see where I'm going with this... when I did the "whole display" macro, within the 0:35 to 0:45 time window, there was no way I was going to wind up with more than 10 x 10 = 100 timing marks using 10FPS. I can almost guarantee that it is impossible to get perfect alignment of my 9 macros in the example above, which could wind up giving me, worse case scenario, 9 x 10 x 10 = 900 timing marks in the same 10 second window instead of 100.... let's say that I get lucky and half of the macros line up perfectly... I'd still wind up with 500 timing marks instead of 100...

    So, if timing marks are so important to display performance with high channel counts, how do y'all do your sequencing? I have seen plenty of videos and it appears to me that most folks don't apply "whole display" macros throughout their songs... only in places... is it necessary to meticulously line up all of the macros so the timing marks line up?

    Thanks for reading, sorry for the length,

    -Louie
     
  2. OP
    OP
    nutz4lights

    nutz4lights Full Time Elf

    Joined:
    Dec 12, 2012
    Messages:
    305
    Likes Received:
    0
    Location:
    Melbourne, Florida
    The season is over. I had another year without music due to challenges in using LSP with my high channel count display (ca. 30,000). I have sworn that I won't let that happen again (basically ran out of time at the end of the year), so I am back again with the same questions and starting early! You can read what I typed above. Basically, last year I ended up having to settle with five animation sequences, 2 minutes long each, running macros applied to the entire display. Not only that, but to get those "whole display" macros to work, I had to set the macro setting to 10FPS (Thanks Eddy!)... the default 15FPS caused massive amounts of stuttering and slowing in the light output. When I switched to 10FPS it just worked.

    As I typed above, 99% of my (and I assume others) sequencing is not done "whole display" when music is involved. There are always groups of elements or individual elements that one would want lighting to the music. What I see in LSP when I try to sequence multiple different elements or element groups in LSP is that, you wind up with multiple sets of timing marks, slightly shifted from each other. If you sequence five different elements/groups in the same time window, you could wind up with 5X the timing marks compared to a "whole display" macro applied to the same timing window.

    I mentioned above how sensitive my light output was between 10FPS and 15FPS... that is only 50% more, so imagine what would happen with 5X the timing marks! Well, I did, and it scared the @#% out of me, so I tried it... and sure enough, once again, the output fell on its face. Same stuttering. This makes the software not usable to me.

    Before I give up, however, I thought I would post here (and the LSP forum) and see what y'all do... and maybe make a suggestion... Those of you with high channel counts... do you go through and make sure all of your timing marks line up perfectly? Again, 10FPS worked for me, so obviously, in a 2 minute sequence, 1200 timing marks works... and 1800 doesn't.

    So why is there no an option to, after creating a macro in a given time window for a single element/group, to snap the next and subsequent macros to those SAME timing marks that are already there... and for that matter... if I know 1200 timing marks works in a 2 minute sequence (10FPS), why can't I just PUT those timing marks there and snap ALL macros/effects to those existing timing marks?

    Am I missing something? I looked in the macro generation window and I see two options when it comes to timing marks... create linear timing marks and use audio timing marks... my sequences so far have been animation sequences, not music based, so the audio timing marks seems to cause the program to hang.

    Any thoughts?

    Thanks,

    -Louie
     
  3. fasteddy

    fasteddy I have C.L.A.P Global Moderator Generous Elf

    Joined:
    Apr 26, 2010
    Messages:
    6,471
    Likes Received:
    16
    Location:
    Albion Park NSW
    Find Me On:
    The timing marks are one of the reasons LSP is getting a complete overhaul because in a very short few years we have gone from talking channels to now talking universes so the amount of data needed to be process has grown massively and running on old core software is not keeping up, thats why we are seeing most software developers now upgrading their software to cope with the explosion in channels.

    Lets do the maths
    1200 timings x 30000 channels = 36000000 data points
    1600 timings x 30000 channels = 48000000 data points

    so approx a 25% decrease in the amount of data being processed or a reduction of 12 million data points by reducing the timing

    So timings become critical the more channels you have

    Unfortunatly with anything there are limitations and in most cases the software is trying to catch up with the hardware capabilaties, so when we decide we want 30K channels then we need to consider how we will manage those channels, everything has a limit. The new software will definetly improve that limit

    One final question were you using the scheduler or the sequencer to run your show
     
  4. mararunr

    mararunr Here to learn! (and share)

    Joined:
    Dec 31, 2013
    Messages:
    238
    Likes Received:
    0
    Location:
    3505 NW Edgewood Ave, Bentonville Arkansas, USA
    Find Me On:
  5. OP
    OP
    nutz4lights

    nutz4lights Full Time Elf

    Joined:
    Dec 12, 2012
    Messages:
    305
    Likes Received:
    0
    Location:
    Melbourne, Florida
    Eddy,

    First, this is running in Scheduler, still running Unicast (want to experiment with Multicast). Both of the following are running on the same computer in the same mode, #1 works completely smoothly, #2 doesn't.

    1. 2 minute sequence, individual 20 second segments comprised of a macro applied to all 30,000 channels using the 10FPS setting in the macro.

    2. 1 minute sequence, individual 20 second segments comprised one macro each applied ten groups of channels using the 10FPS setting in the macro (although, the individual macros don't line up so in any given one second interval, there could be upwards of 80 individual timing marks/frames

    I understand that the amount of data by the increased timing marks is a problem... I am trying to figure out how to alleviate that.

    What I want to know is, once I put one macro into a given 20 second segment applied to one set of elements, creating 200 (20 x 10fps) linear timing marks, why can't the program simply utilize THOSE timing marks when I create the next and subsequent macros in the SAME 20 second time segment??

    If the program could do that (can it?), then I would not be having the problem I'm having right now with #2 above. If the program could do it like this, I would simply have the 2 x 60 x 10 = 1200 timing marks, but with a bunch of different macros linking to the SAME timing marks within a given window of time...

    That is what I want to know ;)

    Someone over at the LSP forum replied to the thread I created there and stated that, supposedly, LSP can PASTE and snap to existing timing marks... so why the heck can't it do the same when creating a macro where there are existing timing marks? The same person recommended having two sequences open, one that you create the macro in sort of a test mode and the second sequence is the actual sequence that would get used... you would create the macro in the "test/work" area, then cut and paste it into the actual sequence of interest, snapping to the nearest timing marks as part of the operation.

    Thanks for the reply, I really need to figure this out before moving forward.

    -Louie
     
  6. fasteddy

    fasteddy I have C.L.A.P Global Moderator Generous Elf

    Joined:
    Apr 26, 2010
    Messages:
    6,471
    Likes Received:
    16
    Location:
    Albion Park NSW
    Find Me On:
    There is also one other thing to suggest trying with the new scheduler and that is to break down the display into a couple of zones, from what ive been told this can help with larger displays because the data is split up and processed in parralel instead of all through the one bottleneck
     
  7. OP
    OP
    nutz4lights

    nutz4lights Full Time Elf

    Joined:
    Dec 12, 2012
    Messages:
    305
    Likes Received:
    0
    Location:
    Melbourne, Florida
    That has to be done in the Sequencer, no? Isn't the zone controlled in the individual element/controller dialogue box?

    -Louie
     
  8. fasteddy

    fasteddy I have C.L.A.P Global Moderator Generous Elf

    Joined:
    Apr 26, 2010
    Messages:
    6,471
    Likes Received:
    16
    Location:
    Albion Park NSW
    Find Me On:
    This is done first in the sequencer when creating a controller/element.
    then you must set up the output config in the scheduler to suit
     
  9. arw01

    arw01 Full Time Elf

    Joined:
    Dec 30, 2013
    Messages:
    384
    Likes Received:
    1
    Location:
    Eastern Washington
    Louie that's me over the LSP forum helping you think outside the box a bit.


    Eddy meant you set it up first in the sequencer with the zones and then in the scheduler with the same exact zones etc.


    You typed you were working with Unicast and wanted to try multi-cast, it's generally best to run in unicast as it's less traffic on the network.


    I had not heard of trying to use different zones to break up the large channel counts, should not be hard to do to test it out. Maybe try your matrix on another zone first?


    Have you tried Light Elf and moved the stuttering sequence over to xlights and played them there in the scheduler to see if you have the same issue?


    could you possibly have some bottlenecks in your home network, some routers and some switches don't handle the volume high channels can put on it, I've seen some wierd fixes from different cables to putting a dedicated to the lighting network controller cards into PC's.
     
  10. fasteddy

    fasteddy I have C.L.A.P Global Moderator Generous Elf

    Joined:
    Apr 26, 2010
    Messages:
    6,471
    Likes Received:
    16
    Location:
    Albion Park NSW
    Find Me On:
    you are correct I doubled up on the sequencer

    I was told to use multiple zones in passing by the developer a while back but havent seen issues that warranted doing it this way, so well worth a go to see if this helps because the scheduler should work fine with 30k channels and a lot more.
     
  11. OP
    OP
    nutz4lights

    nutz4lights Full Time Elf

    Joined:
    Dec 12, 2012
    Messages:
    305
    Likes Received:
    0
    Location:
    Melbourne, Florida
    I know it was you, but wasn't sure if folks around here knew you by name ;)

    There have been several threads around here suggesting that Multicast is what should work best with e1.31... I promised myself that I would at least try this since I have one of the gigabit switches with IGMP snooping capability...

    I will definitely entertain the idea of trying the multiple zones... I was planning on taking down the lights this weekend, but that might have to get delayed ;)

    I tried Light Elf, xLights, playback on the Pi and that is a whole 'nother thread of challenges... I don't even feel like starting into it here much, but all I'll say is that, I was trying that route back in November December before Eddy had me switch my macros to 10FPS and those other setups were capable of smooth playback with 15FPS macros... whereas LSP Scheduler was not.... I couldn't use the other setups though as the channels were out of whack in some of the sequences... and I couldn't figure out if it was an export / import issue with LSP, Light Elf, xLights/NC....

    Regarding changing hardware around, I have done that... I switched to a dedicated PCI Express Intel gigabit NIC (away from the onboard NIC), disabled all wireless and other networks to that machine... even upgraded my switch to a 16 port gigabit switch that cost around $100... different cables, etc.

    I really want to try your suggested method of cutting pasting snapping to get the effects to line up to existing timing marks... I was hoping there was an easier way to reduce the number of timing marks generated that are duplicative...

    -Louie
     
  12. arw01

    arw01 Full Time Elf

    Joined:
    Dec 30, 2013
    Messages:
    384
    Likes Received:
    1
    Location:
    Eastern Washington
    Certainly ran yourself through the ringer Louie. On your higher count sequences did you make a copy of the sequence and cut the controllers in half and replay it to see if it was better or not?


    I have some sequences from last year that i was messing with that I managed to break in some way, they would not play for 44 seconds and then would be fine. Created some from scratch without importing the controllers from the broken sequence and no problem. Was it all your big sequences you had issues with?


    Alan
     
  13. OP
    OP
    nutz4lights

    nutz4lights Full Time Elf

    Joined:
    Dec 12, 2012
    Messages:
    305
    Likes Received:
    0
    Location:
    Melbourne, Florida
    I'd say so... I am crazy persistent when something is not working that I want to work...

    I don't really know how to answer your second set of questions, though. I mean, if I cut half of my controllers away, I would be down to 15,000 channels and I have little doubt that a 15,000 channel sequence would work. There are plenty of folks with 10-15,000 channels working just fine with a comparable setup. There is even another local guy with over 10,000 that has zero problems.

    The way I see it, I have approached sequencing from two different paths... both involve all the controllers being used because I want all of the lights lit up. The first path involved going in and trying to assign effects to an element or common group of elements (i.e. megatree or group of mini/twig trees, group of wreaths, group of wire frame ornaments, etc) through the use of a macro into a set time interval (usually 5-15 seconds). That fell on its face, at which point, I started down the second path. The second path involved assigning effects to all elements in the sequence through the use of macros into a set time interval (again 5-15 seconds). That second path worked ONLY once Eddy suggested I decrease the macro FPS setting to 10FPS from 15FPS, which worked amazingly. I want that same playback quality with Path #1, since, I can't sequence to music the way I want applying effects to all of the controllers at once.

    Here is the thread where I posted my <incredibly boring!> videos:

    http://auschristmaslighting.com/forums/index.php?topic=6913.0

    -Louie
     
  14. fasteddy

    fasteddy I have C.L.A.P Global Moderator Generous Elf

    Joined:
    Apr 26, 2010
    Messages:
    6,471
    Likes Received:
    16
    Location:
    Albion Park NSW
    Find Me On:
    If you have your lights still set up then give the splitting into different zones a go and see if that allows you to sequence the way you want with 30K channels

    just remember that the scheduler and the sequencer are different so the sequencer still uses the old output code whilst the scheduler uses the new code which will work better with splitting up into zones.

    I ran 10K channel through LSP for the City of Sydneys Martin Place Christmas tree and in the month that it was running it never failed and worked great. But then 10K is far short of 30K
     
  15. OP
    OP
    nutz4lights

    nutz4lights Full Time Elf

    Joined:
    Dec 12, 2012
    Messages:
    305
    Likes Received:
    0
    Location:
    Melbourne, Florida
    Thanks Eddy. I plan on doing this tonight. The lights are still up. I plan on splitting off the megatree to another zone, since it is 7,200 channels on its own. Then, I will split my five tall (5-6m) palms on their own zone, since they are 4,800 channels combined... that would knock out 12,000 channels right there without much effort changing the elements... ten twig trees are another 6,000 and will go on another zone which will make zones of 7,200 4,800 6,000 & 12,000 which should be much more manageable...

    Will let you know how it goes.

    Louie
     

Share This Page