Improving VST Performance with Workflow

[S1GNL]

This might be not interesting for those who already have their own "workaround", or a system which is just powerful enough, but it seems that some people here are struggling with using a lot of VSTs without performance problems. I'd like to suggest the following "workflow" to avoid running into CPU overload.

--- VST Instruments (also recommended for CPU hogging Props instruments and REs):

1- Create an instrument with the VSTi of your choice.

2- Don't touch/change anything on its mixer channel. It's fine to touch the gain knob for unity, but you could also use the instrument's volume/gain controller to adjust its output level.

3- Route the created mixer channel to its own, dedicated bus.

4- Use name prefixes for a better channel overview.

What I personally do is calling the instrument's channel "o|synth1" (as in "original") and its dedicated bus "b|synth1" (as in, you guessed right, "bus").

I also use a coloring structure. The "o" channels get the lightest tone, the "b" channels a darker tone of the same color. It really helps finding the right channel much faster.

5- Whatever you want to add to the original sound on your channel (mixer comp., mixer EQ, inserts, sends, volume automation etc...), you do it only on the "b|synth1" channel!

6- So, now that you have your composition and the virtual instrument's settings (and maybe instrument automation) right, you set the loop position and length, and then you bounce the "o|synth1" mixer channel.

7- I'd label the exported audio track's channel "a|synth1" now. Place it next to the "o|synth1" channel and choose the same color.
Then you route that "a|synth1" channel to "b|synth1", mute the "original" note lane (NOT the track! The lane mute is independent and stays on mute even when muting/unmuting all tracks) and just disable/turn off the VSTi.

This way you can use loads of VST instruments in one project without running into CPU problems. That's an old story, I know! Here is the big bonus:

Whenever you feel like you want to change either inside the MIDI clip or the VSTi, you just mute the audio CLIP (don't delete the track!), enable/turn on the VSTi, unmute the note lane and do your thing.

Afterwards you just bounce the "o|synth1" channel again, replace the old audio CLIP with the new one, delete the now created audio TRACK, mute the original note lane and disable/turn off the VST again. During the changes, your mixer channel's processing stayed as is it was. You can decide if you want to work "online" (VST active) or "offline" (VST inactive) without changing anything else. There is no need to copy, adjust or whatever...

(A tip on the side: when bouncing channels the created audio track appears at the bottom of the sequencer window. I think it's annoying to grab and slowly scroll upwards when there are already 30 tracks or so... just select the audio track and cut it with ctrl+x. Scroll with your mousewheel to the "original" track, select it and paste the created audio track with ctrl+v directly below it.)

It sounds pretty tedious when describing the process lol, but I promise that once you've got used to that it will be a matter of a few seconds only.



--- VST Effects:

This is a tricky one. It's not really handy (not really an improvement) if you're using the VST effect permanently during mixing (like EQ, compressor, ...) or if you're using the VST effect AFTER you've already engaged the mixer channel's compressor, gate or EQ.

But if you're using VST effects pre-mixer channel compressor/gate/EQ then you can save a lot of CPU resources by applying the same process. Here's an example:

So, I use this UJAM IRON VST (CPU demanding!) to create a clean sounding rhythmic guitar track.
Then I use AmpliTube4 (CPU hog!) for the distortion and amplifying and the analogue chorus (because I really like how it sounds!). Afterwards I route the processed guitar sound into Slate VTM (another hog!).

1- I create the instrument/channel "o|gtr1", loading IRON.

2- I route "o|gtr1" into the bus channel "f|gtr1" (as in fx). I put AT4 and SlateVTM into the insert slot of "f|gtr1".

3- I route "f|gtr1" into the bus "b|gtr1".

After getting the composition and AmpliTube4/Slate VTM settings and sound right:

1- I bounce "o|gtr1". I disable the VST instrument and mute the note lane.

2- I route the created audio track's "a|gtr1" channel into the bus "f|gtr1".

3- I bounce "f|gtr1". I disable the VST effects.

4- I route the created audio track's "p|gtr1" (as in "processed") into "b|gtr1" where I do all the mixing stuff.

Of course, you could put the VST effects into the insert slot of "o|gtr1" and save one bounce and one bus, but I prefer to be able to keep the VST instrument disabled if I e.g. just want to change the highgain level of the amp or the speed of the tape.

You can get really "creative" with that and "stage" instruments and processors the way you like. Just keep it well organized and the color coding helpful and logical.

Drawbacks:

Yeah, it's not perfect lol.
It takes some time to get used to it so you're really quick with applying it. And your mixer will get bigger!
Each bus channel needs CPU resources, but compared to one active instance of Omnisphere or 5 active instances of Albion One (Kontakt) or 4 active instances of Decapitator (pre-comp./pre-EQ) the additional CPU usage caused by bus channels is a joke.

Like I've already mentioned it before: this way you can use probably more VSTs than you need! It's not the perfect solution if you e.g. want to use FabFilter stuff as your go-to channel strip instead of the Reason mixer Comp/EQ. Because you'll probably use them permanently. But applying the process as described above you'll save a ton of resources to actually be able to use FabFilter, Waves, Metric Halo (my favorite!) or whatever as your channel strip instead.