Hey folks,
  I've been using Sonar for quite some time and officially working on my first 5.1 surround project for a short film.  I've barely had time to scratch the surface and really research standard industry practices for sound recording -> mixing -> mastering/Authoring workflows to be able to practically approach this project, so I've been just learning as I go and I've noticed some peculiar behavior in Sonar X2 Producer becoming apparent.

1. Rendered audiosync loss: Towards the end of the rough sound design I printed a render of the sound mix so far.  Upon bringing the printed bounce into the NLE for critical viewing with the director it was immediately apparent things weren't in sync.  At first, some elements were ahead, and within a couple of minutes elements began to sync up, and then quickly lose sync and fall behind by the time we arrive to about 4 minutes into the picture.  7-8 minutes into the picture everything drifts back closer to sync, and by 12-13 minutes in to the picture it's more than 6-10 seconds behind!
   1. After several searches, and several bounce attempts, I spent some time really trying to pin point where the sync issue was happening.  Back in the NLE I checked all the audio clips (externally recorded and synced up manually IN the NLE - then exported the OMF of the entire picture to import into Sonar).  Everything played in sync there, so I re-rendered the entire picture to a new video file to see if something had changed in the picture.  I imported the new video clip into Sonar - everything still plays in-sync inside of sonar.
   2. After seeing nothing had changed in the timestamps for the audio clips that were imported into Sonar from the timestamps within the NLE, I figured it was worth looking into the container formats to make sure the video formatting was done correctly.  The film was shot with a Red One MX, and all the native .r3d video files imported directly into Premiere Pro, which is the NLE we're using for the project.  After scouring the web for a couple hours and finding other people's posts about Sonar and audio drift issues, especially in Premiere, there are plenty of people with the issue however all of them blame variable frame-rate containers like MP4 (a main feature of the h264 being temporal compression).  People were citing this issue most common when using mp4 video shot on consumer devices like mobile phone cameras etc.  NONE had issues with the R3d format video.  After running into a dead end there I simply went back through the video project and double checked every step of the project was in the proper framerate format, and that none of the clips or sequence data had strayed from the 24 fps.
   3. I went back into Sonar and double checked that not only was I viewing the grid in SMPTE (hours:minutes:seconds:frames) but also that the video I had imported was at 24 fps, and the grid measure was also at 24 fps.  Everything lines up.  So I'm still left with the resolution that something is happening between the render from Sonar and the imported final audio asset into the NLE.  While some people claimed there were issues with Premiere Pro not being able to handle long format audio files this isn't true, I've had multiple projects I've done in the past in stereo that have never lost sync, and those were shot on DSLR and Samsung Infuse phone cameras!
   4. So finally, I am looking through my Sonar project, and I decide to just re-arrange my plugin processing to a per track basis, and freeze each track, anything like the mastering grade processors that need lookahead, or tools like Waves Vocal Rider obviously create buffers,  so when I'm trying to listen back within the DAW my buffer size in my recording interface is quite high (2048 samples to comfortably play back, 1024 with occasional minor clicks and pops).  After I froze some 100+ tracks I bounced out a new render and dropped it into the NLE, BAM, sync achieved!
2. Working To Picture = stability issues
   1. My only hitch with Sonar has been trying to work to picture.  While the options allow you to import AVI, MOV, and WMV, the only codec I can get to work is WMV.  The longer the video and the further you get into the timeline, the longer Sonar takes to respond to play and pause.  It seems like it has a hard time indexing the video.
   2. I've come to terms that working to HD picture for critical viewing within Sonar is not the ideal workflow, instead working with as small a picture as possible to keep the video filesize to a minimum it handles fairly well. 

I own Pro Tools 10 (11 is out now, however, which boasts video editing abilities) and I know it's the de-facto DAW of choice when it comes to audio for video post.  I'm just curious if anyone else out there has had any success stories of managing a workflow inside of Sonar editing and mixing to picture.  Has anyone else come across these problems and what have been your solutions? how have you managed to work around them?  I'd love to get more into detail eventually about specific mixing procedures and practices but let's save that for another topic, if you've had experience working across any of the material mentioned above i'd love for you to share!

This is very interesting. I'm wanting to get into film scores myself with Sonar. I'll be keeping my eyes on this thread.

Like you, I learned the hard way that I have to freeze tracks that were going to require critical sync.  Also, like you, I don't bother with any other codec in Sonar other than wmv.  Wish I had more to offer, but honestly, you really summed it up nicely.  Those who think they are going to import 4k footage into Sonar and run 128 tracks of audio with convolution reverb are deluding themselves.

sometimes I will simply render out submix stems from Sonar and import them into my NLE.  Since the LA2A works as a VST, I can use that on the master bus in my NLE.
 

One would think a WMV with mainline 4.1 compression at 720p would be fine for viewing, considering my average file size is 581 mb for the 17 minute 720p @ 24fps WMV, the small (320x240 @24 fps) WMV weighs in at only 78.1 mb.  Much easier for Sonar to parse and index. 

I really love that Sonar has the surround bridge feature to automatically make stereo plugins available on 5.1 busses.  Unlinking the bridged instances to get more control over the discreet audio channels is proving very fruitful, however I'm concerned moving back to processing on surround bussing will create the sync issue again.

I've found that for sound design and editing, whatever processing I can do on the per-track basis, I'll take advantage of the take lanes.

• Original imported audio gets the 1st top lane, I'll hold Shift+ctrl and drag the clip into a new lane, and mute the originating lane, apply the plugins diretly to the clip (make sure to extend the clip out well beyond what you want in the audio to account for any tail or editing afterward that you might need back in the mix later).  For example, my first pass would be denoising a track (the camera on set sometimes would be quite audible especially in small interior spaces due to the Red One MX being a heavy metal chassis - the internal fans on the camera were extremely loud while it tried to stay cool when not shooting, and while it does go into "silent mode" as soon as you hit record, the fans are still perceptible to the microphones especially as gain gets boosted in post production to bring dialog and other featured sounds up.
  
• One particular favorite of mine WAS to simply pack up the plugin chain I had created on one track to easily apply the chain to all the tracks, I tried to apply a plugin chain directly onto a clip and the chain container on the audio clip ends up empty. I have to unpack the FX chain and copy drag each plugin individually in the proper order onto the clip.
  
  
• Much easier to use the take lanes to iterate and quickly bounce versions of audio clips to keep resources to a minimum.  If you're doing heavy processing with multiple plugins (Just the 1 FX chain for "fight outside" there was 6 plugins used), I would highly suggest packing up your FX chain and labeling them the version OF the take lane you're in so you know which FX chain was used to create which bounced version of that audio clip.  ie: Fight Outside 1.0, Fight Outside 1.1 etc. or 1, 2, ABC etc... In some cases you may be swapping out entire plugins or using completely new ones each time.  Keep each plugin chain on your channel in the proper order and just left disabled.
• Another practice I've started using is manually saving out preset files native to each plugin based on what I'm doing to build up my private Preset library.

Turns out there's a simple, but rather silly solution for dropping FX Chains onto clip bins.  When you drag and drop the FX chain over a clip, the clip's FX bin will pop up, you have to make sure you let go of left click while over this pop-up bin window, if you let go just while over the clip, the FX chain will be empty.  This works regardless of whether clip(s) are highlighted/selected.

Just went through final print master.  Several renders later I decided to just leave "fast bounce" on and audio drift problems disappeared.  During playback in Sonar I would have to frequently pause it, and let it get back in sync and hit play again.  I was afraid I'd have to turn all of the plugins off, bounce 1 group bus at a time, apply the plugins on a separate project for just that stem mix, bounce that, bounce another stem mix unprocessed, apply the processing to that stem on it's own and render both out together, and slowly compile the whole master print the hard, long way!  I'd just really like to know what causes the program to lose the ability to keep track of clip timings... 
The good news is I didn't have to resort to that, bad news is i'm 12 hours behind schedule now :P

Curious to see if this behavior changes in X3 in any way?

Yes very curious I'm eager to see what updates they've done under the hood besides all the feature list articles they've been posting up everywhere.  So, here's another update on this project.  Trying to bounce with even just 1 plugin that has any sort of latency like Ozone 5 maximizer throws everything out of whack. Just 1 simple plugin on the master bus causes havoc on the sync.
 
After numerous failed attempts to bounce, I disabled all FX and just tried to bounce at 48k/16 bit multichannel wav (I'm assuming Sonar automatically chooses signed 16-bit sub) and left Fast Bounce on. While this did the trick it left me confused as to what exactly is causing sync loss and audio drift.  My project is in 96k/24 bit and with the # of plugins active only on the surround busses I'm able to play back the entire project in-sync with my audio driver's buffer size around 1024 (otherwise clicks pops and scratching does start to occur as opposed to just playing out of sync). 

1. My next logical choice was to try and figure out a way to bounce stem mixes without any bus processing that is essentially bringing the gain and mix together JUST to see if Sonar could indeed render out a print that does line up to the video.  Viola, as long as no plugins OR pro-channel modules are enabled everything is in sync.  There were only a few tracks (not buses) that had pro-channel modules being used and thus I had to freeze those tracks in order to get the processing chain in proper order.
2. I decided to make another bounce of the same bus solo'd with the plugins active on them, but the master chain still disabled, effectively printed a bounce of the mix phase the way it's intended.  Dropped that in the video NLE and that syncs up too!  I quickly bounced all the stems (each of these were my surround buses: Dialog Master, SFX Master, Ambience Master, Music Master) and started a separate sonar project and appended the title to say "Mux" purely for organization.  I brought all the multichannel waves into the new project and checked playback to the preview WMV.  Everything plays in sync.
3. From here I applied the master FX chain I had saved out from the original project file.  At this point all of the stems already have their mix processing baked in, so naturally all that's left to do is master.  I bounce out the master and the whole thing yet again comes out of sync EVEN though I'm using the same render setttings mentioned above - 48k/16 bit FAST bounce.  The "mux" project is also in the same format.  So now i'm confused if the clip timings came out of sync in the baked stem mixes or if it was just the fact that adding a plugin to the master chain can cause audio drift.
4. My final option i'm trying right now is to purely go through the mux without processing just to bring the audio together into a single 5.1 mix, and find a way to apply the mastering chain in Sound Forge Pro and see if that does the trick.  I just have to be weary before bouncing there are no peaks in the stems that cause the master to pass "digital zero" and go into the red.

My thoughts are that Surround Bridge when handling plugins with latency - due to it creating multiple instances, somehow causes this audio drift.  while the feature of being able to bridge stereo plugins for surround use can be really awesome (if I have more time to explore it for creative purposes), clearly it's implementation is something VERY specific, or just plain far too complex for someone to just "get into" and start creating surround mixes right away.  This whole thing has put me several days behind schedule already I just hope I can close it out tonight, the VFX artist is coming by tomorrow to finalize some assets on the remaining shots and then final Print Master rendering will take about 24-28 hours..
 
I'm curious if X3 has addressed anything under the hood that helps to keep plugins in sync across variable latencies each plugin might create.

This may or may not answer your questions...but here is a video I made about how I work with picture.   i do not mix my music or work with surround sound.
 
http://www.youtube.com/watch?v=zh9Ico1HgC0