Orchestral mixing with IRs - questions

[mderrico]

I'm just getting into impulse reverbs using orchestral samples and I have a few questions.

In order to give each orchestral section a different stage location/reverb, do you need to run a separate instance of the impulse reverb for each section? Or can you set the stage position of each section in one instance of the IR plug-in (which I doubt)? Should you use different impulses for different sections of the orchestra or can you use the same impulse and just lengthen/shorten the response tail?

I use a generic reverb while composing orchestral music. Then I render each MIDI track to an audio file without any reverb. Then I create a new project, load in all the rendered audio files, apply the reverb, EQ, etc and mix it down.

Is it better to render the MIDI tracks to audio files WITH the IR reverb or is it better to apply the IR reverb during mixdown (after all the MIDI tracks have been rendered to audio files without reverb)? I'm kinda concerned about this because I downloaded the Voxeno Pristine IR plug-in. I'e got an Athlon 2400 with 1GB RAM and as soon as I loaded a 4th instance of Pristine, Cubase became extremely sluggish (and that's without any tracks loadeda at all). I tried this several times and it just seemed too much for my computer. I was also using Peter Roos demo impulses (which sound great btw!)

I think I'll be buying the Pristine IR plug-in since SIR is only mono-to-stereo.

Any advice on how to mix orchestral tracks using the Pristine IR plug-in in SX would be greatly appreciated. I searched all over this messageboard and a few other places, but I could quite find the answer.

[Aleksey Vaneev]

To place each performer on the virtual stage you will need to run each performer track through an impulse processor. Indeed, this usually requires A LOT of CPU resources. From what I know, people on these forums make use of distributed computing - running FX Teleport plug-in on several networked PCs. This usually makes convolution processing much less problematic.

Alternatively, if you have a good spatialized IR set you may bounce each audio track run through a convolver and then work with processed audio tracks - this won't require any additional hardware. This approach is identical to real-time calculation because impulse responses recreate static spatial images - you won't even able to adjust the wet/dry balance with such stage impulses.

As for the Pristine Space reverb of which I'm a developer, it is indeed requires a lot of CPU resources to run several instances. You may decrease the strain it puts over the CPU by increasing its internal processing latency along with the audiocard block size. The only other way to resolve CPU load problem is to use either of the aforementioned approaches.

Using Pristine Space in Cubase SX 3 is a matter of inserting the plug-in on the track insert or send FX - just like any FX.

[tcovem]

What a lot of people do is use three instances of the convolution reverb on the aux busses. One for near, one for mid, and one for far within the hall. Do panning in DSP in your sampler or sequencer, and assign each track to either the near, mid, or far instance. I haven't been able to figure out if I can mix between the instances (assign some near, some mid, and some far to the hall for example) or dial in differing amounts of one of the reverbs (leave some dry signal and increase or decrease the wet signal through the send). That's what I would do with a normal reverb but it seems that one is supposed to set dry signal to nil when working with a convolution reverb, at least GigaPulse, and set the fader to "pre" so that the reverb handles the sound exclusively. See my other topic where I ask about this... I don't know the answer to that part of the puzzle. If I can only place something near, mid, or far, it doesn't give me as much control as I have with traditional reverbs.