I'm not a programmer myself - I can typically decipher code and understand what the program is doing, but I could not write the same code of many of the talented people here. Hats off to those who can!! I relied on Frankr to assist me greatly in some of the scripting this year. If no changes are made to the script, I know I have a workflow that will work next year. I'm hoping that some of Frank's genius can be expanded so that some of his programs can be used for many folks looking to get into the video side of sequencing. On to your questions / concerns...
There is no native way to export certain segments of a show. To do this, here were my steps:
1.) Create my video in Vegas (or Premiere or whatever you're using) and export the video in segments - all NLE software will allow this at this point.
2.) Take all of those show segments and add them to LSP as custom transitions - I put all my transitions under the song name so I didn't have 100 video segments in one big pile of ... garbage in LSP.
3.) Create a new LSP media sequence and use custom length audio for the sections. For example, I broke my show up into 30 second segments because I knew I could export at this level without issue. If you were using 30 second segments, either copy / paste your original media file into exactly 30 second segments or (and this is easier if you're doing your entire show by video) generate exactly 30 seconds of silence in a wav file. You can use those custom length wav files over and over again. All you care about is a sequence that is exactly X number of seconds long (and that matches your exported video).
4.) You can REALLY speed up the render process by standardizing on a certain segment length by performing the following:
a.) Open LSP and add your 30 second media file. Add All your custom transitions that are 30 seconds in length to LSP.
b.) Drag your LSP transition to the appropriate layer. Make sure the transition is at the VERY beginning of the 30 second segment.
c.) Verify all your transition settings are correct - duration and frame rate need to match. At 30 seconds and 15 fps, you'll have 450 frames - the maximum number of frames you can specify for a transition is 500.
d.) Save the file as <song>Section#.lms. Then click on your transition and CHANGE THE Transition file, but leave ALL OTHER SETTINGS ALONE - UNCHECK RENDER NOW or GENERATE NOW. Save the file as the next section. Click the transition - change the file, Save as the next file section. You'll quickly have the entire song saved in 30 second segments.
e.) Open each transition in a seperate instance of LSP. Right-click the layer with the transition and "generate".
f.) Take all of these sections and export to LOR format.
g.) For the last song segment, you'll have a shorter movie / wav file duration. Just create a custom file for the last segment for the appropriate duration. Make CERTAIN you match up the frame rate / number of frames or you'll get unpredictable results with the transition effects generation. If your song is 10.5 seconds at 10 fps, you'll have 105 frames total. Specify that in your transition - 10.5 seconds, 105 frames.
2.) This is sort of a multi-part operation. You'll need to take the lor files and perform whatever operations are necessary for xlights. If using several etherdongles, you may not have to do much at all - I've not done a conversion yet so I just don't know. I think if your LSP file is setup correctly, nothing may need to occur. I will iron this out first of 2012.
Once your LOR files are processed, you'll need to convert each segment to individual xseq files. Then you'll need to use a script to append the xseq files. Frankr was a genius on this - I will check with him to make sure he's ok posting his script for all to use / abuse. Essentially his script allows you to specify a NEW (original) media file for the xseq format to reference, then attaches all of the xseq individual files together. This is what allows you to use a dummy 30 second clip of silence with LSP for the transition effects generation. During the combining process of the xseq, you point to the original file your video was built to. 4 x 30 second clips plus 1 x 10.5 second clip would be a 130.5 second clip which (hopefully) matches your original song.
I would highly recommend setting up a test lab to verify your lights perform as you need them to in relationship to brightness when doing the video transitions. Once you've rendered all the video to lights, run through a couple of external scripts, and played back the file in xlights, you're a LONG way from the raw data you started with!
One last word of caution... going to 50,000 channels is a MONSTER. I hope I'm not speaking out of turn, but I've seen a couple of comments asking if the smart string controllers in 2012 could allow a "group node" function. Essentially a 100 count strand could be controlled as 10 nodes if a grouping of 10 lights was used. There are not many places where I really need 100 bulbs controlled individually. Even if you can group by 2, you've cut your effective channel count in half! I think you're likely going to see some SERIOUS sequencing issues when you start playing with 50,000 channels.
Consider the following: I had 4,166 channels this year. At 15 fps, I was only able to export in 30 second segments. If we assume (and I DON'T know if this is the case - I've not played with the software to see) the relationship between LOR export and channel count / fps is linear in nature, your exported LOR files will be approximately 12.5 times larger than mine were. The 200 MB LOR file size seemed to be towards the upper limit of what I could export from LSP. If that's the case, to render a show with 50,000 lights, you're going to need to build that in 2.4 second segments for the same exported file size of 200 MB. A 3 minute song will be comprised of 75 segments. That is 15 GB of LOR data for 1 song that needs to be processed. My scripts took approximatley 3 minutes per 200 mb LOR file to update the network / channel numbers I needed. Converting to xseq (which must be performed 1 song at a time in the current xconvert implementation) took 1:25 per 200 mb file. It will take you 1:25 for every 2.4 second file. Perfectly optimized, you're looking at roughly 337 minutes of post effects generation processing. That's 5 hours. Per song. Shoot me now!
I would STRONGLY recommend looking for options to reduce channel count, whether by grouping nodes together, operating many strands in hybrid mode so you can control the entire strand with 3 channels, etc. I just foresee many, many problems when you're creating so many sequences, keep track of the sequence names, adding THAT MANY transitions to LSP. I would seriously consider installing LSP in a seperate folder for every song you were doing just to keep the load times down - or I could create a bat file to swap the transitions xml file based on the song you're loading. Silly things that you won't anticipate will arise, and I don't want you to be like me this year. The day before Thanksgiving, I had NO show. With 50,000 channels, eek!!!
Now, when you do succeed, and I think you will - there are simply too many good, helpful people on this board, I'm really excited to see what you come up with next year! I hope that helps a little, and perhaps sheds a little light on some of the issues you may encounter with running such a fantastically large display.