• Register
  • Help
Page 1 of 3 123 LastLast
Results 1 to 10 of 22

Topic: OT: most efficient way to render GS3 to Host?

Share/Bookmark
  1. #1

    OT: most efficient way to render GS3 to Host?

    Hi there,

    Just putting the finishes touches on a film project and I need some advice from our forum members.

    I am tracking the 'giga parts' written (rendering them to audio track within host - via ADAT).

    Here's my set up

    PC 1 - Host running SX

    PC 2 - giga PC

    PC 3 - giga PC

    PC 4 - giga PC

    Am I being too picky rendering each instrument to its own audio file (I do this to not limit my mixing later on - usually balance issues)? Problem - it takes forever

    Should I render these many many parts ONLY by orch section? Of course - my 'midi mix' would have to be closer than my usual attitude of 'levels will be fine tuned in the final mix'.

    Given my PC set up, what would others suggest / are doing?

    Many thanks for any advice offered. Just looking to be faster at this stage of the project.

    Rob
    Rob Elliott Music
    www.robelliottmusic.com

  2. #2
    Senior Member Bruce A. Richardson's Avatar
    Join Date
    Sep 1999
    Location
    Dallas, Texas
    Posts
    5,755

    Re: OT: most efficient way to render GS3 to Host?

    Hi Rob,

    You mention that you have the parts coming in via ADAT. Does that mean you're bringing in multiple ADAT lines at 8 channels each?

    You can generate an output file for each hard output. So, I would enable every possible output on your Giga machines, and split the work out that way...assigning parts to logical output groups, and then splitting those groups out to every available output. You can either input this all into the sequencer, or render out and drag in across the network. I like this option, because I can just point all the groups to a common output in the composing stage, then break them out just for rendering. In this fashion, you're just making a few rendering passes rather than one for every instrument, since you can write multiple stereo output files simultaneously for as many outputs as exist.

    I have offered up the idea that they should implement a group output mode as well, so that one could literally do every bit of rendering in one pass. This shouldn't be a big deal, since the rendering engine can lay back quite a bit and just talk to the disk at its leisure (unlike the input side). I always render out to a non-sample disk, and have never had any issues with dropouts, etc., doing that. I can imagine that attempting to render to the disks containing sample data might cause some internal consternation as the app juggles its appetite for incoming versus the need to spit out the record buffers to disk in a timely fashion. However, all of this stuff is going on in low latency in real time to the playback buffers, so there is probably not too much overhead difference in printing these files to the GigaStudio machine versus piping it in to the DAW over ADAT lines.

    I use the "MIDI Start" feature. My personal methodology is that I use the sequencer to build the basics, render the files out to disk as they are done, and drag those files into Vegas's desktop across the network. I never have problems with throughput over gigabit ethernet, so this is extremely convenient. If I subsequently step on those files with later versions, the later versions just come up in Vegas, and nothing changes about the mix, except that the newest version of the track is the one being mixed.

    So I guess what I am functionally doing is a constant "track freeze," across three applications (sequence-SONAR, sample output-GIGA, mix engine-Vegas).

    That process was panned as "ridiculously convoluted for no reason at all" in another thread. I beg to differ. I am using the full brute force of three machines to strike a balance between the power and sound quality I want to work with in real time, versus the desire to keep the rig simple and processes redundant. I can dedicate the sequencer to a single operation--locking my musical ideas to picture. I dedicate the two synth/sampler machines to using their full CPU potential to deliver lots of recordable tracks in real time. Finally, I dedicate the mix to an engine which is arguably the best audio/video mix environment that exists in the world.

    All combined, I am not taking any more time to use this system than someone working around various sections of a single sequencer to accomplish the same task. But what I get in return is a far more robust and efficient work experience for my business needs. Closing a sequence does not alter the sound production end of the deal. So, it is extremely quick to move from piece to piece within the same basic soundset (which can be extensive using two full computers). That is a huge timesaver over the course of adding up the minutes you'd spend over a week of loading and unloading. Little slices of time DO add up.

    Also, I am using the exquisite power of GigaStudio's QuickSound search capabilities combined with the Distributed Wave audition in order to keep every scrap of audio data at my fingertips. So, this adds a very big amenity to both the sequencer and the final audio/video mix app. I can use Giga to "find" things, and just drag them right across the network onto Vegas. For instance, the entire Hollywood Edge library is QuickSound-friendly. QuickSound will hit the title and keyword blocks of any standard WAV file, so you get this monstrously fast way to find and audition things...literally seconds to load up every instance of "footstep" or "subway" and play through them on the keyboard.

    I use a lot of musique concrete techniques, even in my most orchestral work. I like pitching the whine of traffic as a pad, or the call of an animal strangely assembled as melody. I can knock this stuff out in seconds, because I can audition in QuickSound until I find the exact thing I love, then drag it right across the network to Vegas, as I mentioned. Having auditioned against the track, I know exactly where the pitch needs to go, so I just start playback and "Ctrl-DRag" the end handle in Vegas and release when the pitch is where I want it. Done, just like that. There is no DAW app on the planet which would make processes like that faster. We are talking literally instantaneous.

    The redundancy factor is huge. Having data distributed over several machines reduces the chance that you will ever experience a catastrophic loss, especially with minimal backup procedures in place. Using an NDE like Vegas as the final mix engine means that the tracks and stems will always be totally virginal and readable by any standard application. So, even if your mix got totally destroyed somehow, you would be able to load up the tracks and go. Save off EDL's of various formats as a matter of course, and the level of redundancy goes even deeper. Now, you could rebuild it on any system you like. You don't get these levels of redundancy in single app scenarios--a single app "owns" too much data in native forms, so a catastrophic loss on the DAW machine can very likely wipe out both the work and the sources. With distributed sources, you are hedging your bets that very little can happen to wipe out existing work. It would take a concerted effort, or a catastrophic event like a fire to sytematically destroy all the ways of recovering the work.

    One tip I would offer is that I try to be constantly doing this work while I'm in the composing process--getting tracks into the audio domain as soon as they're cleaned up and ready, on a track by track basis. If I want to make a change, I just punch in and replace. Guess I'm a bit old school in that way, but for me, it seems to keep the process chugging along.

    That is WAY more than the original question asked, I realize, but I was thinking about this last night, and just wanted to capture my thoughts...

  3. #3
    Senior Member Bruce A. Richardson's Avatar
    Join Date
    Sep 1999
    Location
    Dallas, Texas
    Posts
    5,755

    Re: OT: most efficient way to render GS3 to Host?

    Hey Rob, in rereading your post, you could just substitute dragging the files back into SX and muting the MIDI for what I do in Vegas, of course. It would be functionally similar. Cubase always made its own copies of audio when I used it, but I'm sure by now it has a nondestructive editing mode that would just track the file rather than load it up. Even if it wants to make redundant tracks, all the better. You can always erase 'em later. Better to have too many copies of something laying around than not enough (i.e., crying in beer time)!!!

  4. #4

    Re: OT: most efficient way to render GS3 to Host?

    Wow, Bruce - quite a tutorial. Much of which will require my study

    I don' t have Vegas, but it sounds like something eventually I'll need to consider.

    Right now, I may be using GS3 in a limited fashion. As I have the cue / sequence finished, I simply mute all other midi tracks - except the one in question - route the GS digital out to the SX's card ADAT in and record the midi part - one at a time.

    My RME digi 52/96 has all three GS PC's locked up (all in all 24 input channels). Sounds like I could at least simultaneously route multiple channels to the SX host PC via ADAT.

    Not sure what you mean about GS3's ability to 'render' - need to get into that more (now see it as a sample playback tool )

    Thanks again for the reply.

    Rob
    Rob Elliott Music
    www.robelliottmusic.com

  5. #5

    Re: OT: most efficient way to render GS3 to Host?

    Rob:

    GS3 has a capture to WAV ability which works best, as Bruce mentioned, if you set it to start capturing upon receiving a midi start message... which enables you to import the wav file into your sequencer in perfect sync to the midi track that played it. Assuming you can configure SX to emit midi start messages. Another possibility is to play the midi with your sequencer while recording the audio output(s) of GS3 to audio tracks in SX. Don't know if SX can do that but it might be worth a try... might let you capture a bunch of instruments to a bunch of tracks simultaneously.

    Howard

  6. #6

    Re: OT: most efficient way to render GS3 to Host?

    Quote Originally Posted by howardv
    Rob:

    GS3 has a capture to WAV ability which works best, as Bruce mentioned, if you set it to start capturing upon receiving a midi start message... which enables you to import the wav file into your sequencer in perfect sync to the midi track that played it. Assuming you can configure SX to emit midi start messages. Another possibility is to play the midi with your sequencer while recording the audio output(s) of GS3 to audio tracks in SX. Don't know if SX can do that but it might be worth a try... might let you capture a bunch of instruments to a bunch of tracks simultaneously.

    Howard
    Howard - thanks for the reply. Not sure if my pea brain is wrapped around this right but right now I run the sequence with the ONE midi track solo'd and record in the host what is coming from the one or couple of gs3 tracks (via ADAT). Am I missing something from what you explained. How would rendering in GS be better.

    Many thanks.

    Rob
    Rob Elliott Music
    www.robelliottmusic.com

  7. #7
    Senior Member
    Join Date
    Apr 2004
    Location
    Decatur Illinois
    Posts
    901

    Re: OT: most efficient way to render GS3 to Host?

    What about what Bruce mentioned about assigning as many GS instruments or groups to as many different outputs as possible and then recording them as seperate inputs into SX. That is, if like he asked, you have multiple channels of lightpipe going from GS machines to host soundcard. This is essentially what you are already doing but with multiple tracks at a time and ending with the same result. That being an individual audio track of each GS midi track.

  8. #8

    Re: OT: most efficient way to render GS3 to Host?

    So long as you don't get unacceptable delays or pops, simultaneous play/record is a pretty efficient way to render. A while back as a test I did it both ways and found there was a fixed delay using record instead of capture. Which I think relates to the recording buffer size. But it was easy enough to compensate for. Which might not even be worth worrying about, especially if all tracks start out GS3 midi and get the same treatment.

    Howard

  9. #9

    Re: OT: most efficient way to render GS3 to Host?

    Quote Originally Posted by EricWatkins
    What about what Bruce mentioned about assigning as many GS instruments or groups to as many different outputs as possible and then recording them as seperate inputs into SX. That is, if like he asked, you have multiple channels of lightpipe going from GS machines to host soundcard. This is essentially what you are already doing but with multiple tracks at a time and ending with the same result. That being an individual audio track of each GS midi track.

    Yes - that is exactly what I am going to do (Thanks Bruce and Eric). Just as soon as this project is in the can - I'll test this out on one of the cues.

    Had the director over last night for final mix (part of the contract ) - went great, but I would have been killed if I didn't have tracks seperated - at least by type and style of instruments (i.e - he would say can I have a little more of that trombone splat, less horn their, etc.)

    Thanks again guys.

    Rob
    Rob Elliott Music
    www.robelliottmusic.com

  10. #10

    Re: OT: most efficient way to render GS3 to Host?

    Quote Originally Posted by howardv
    So long as you don't get unacceptable delays or pops, simultaneous play/record is a pretty efficient way to render. A while back as a test I did it both ways and found there was a fixed delay using record instead of capture. Which I think relates to the recording buffer size. But it was easy enough to compensate for. Which might not even be worth worrying about, especially if all tracks start out GS3 midi and get the same treatment.

    Howard
    Hey Gentlemen,

    For me - this is the best option. SX will record multiple tracks/channels at the same time.

    Here's my challenge though. Out of GS - since I am running two GP instances in each machine (mid/far or close/mid, etc.) I run my signal through to the 'group bus' then 'Aux bus' (to pick up the two GP).

    Since I only have two instances of GP (limited by CPU usuage at around 50-60% with polyphony) - I only have really twopaired outputs 'at a time' when running the host's midi, GS's playback out.

    Hopefully I am making myself clear. I see it as a routing issue now. How can I take advantage of the four digital output pairs coming out of GS. As explained - doing it this way I only have two digital outputs available (because I want to run things through GP).

    Any ideas? If not - 2 pairs out at a time is better than one

    Rob
    Rob Elliott Music
    www.robelliottmusic.com

Go Back to forum

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •