Doesn't work if you need to manipulate the plugins gui.

If you assign CCs to physically manipulate the parameters, "live input" should work.
 
Does it not?
 

Anderton

Earwax
Because the guitarists in the bands cited in my examples are not recording gestures. They are recording sounds. They are recording sounds at the same time as the keyboardist and drummer are recording their sounds and some (depending on the VSTi and what performance tweaks they are using) of their gestures.  The live performance breaks down even in duo or trio recording without the ability to record VST/VSTi directly into Sonar. 

SONAR is recording the sound of the guitar.

Not if the final audio from the guitar is the output of an amp sim and VST effects in Sonar. If Sonar could playback and record output from VST/VSTi at the same time, from multiple sources, there would be no issue. It can't. In my examples, more than one person wants to record VST/VSTi audio at the same time. No way can Sonar do that.   

jatoth
I believe Scook has posted a work around for this many times.
IIRC it went something like this...
 
Place a note past the end of the song on the MIDI or instrument track
Enable "Echo On" on the MIDI or instrument track
Select the track(s) associated with the synth
Bounce the track(s) with "Fast Bounce" disabled, "Audible Bounce" and "Live Input" enabled
Play synth while the track(s) are being bounced.
 
This should allow you to "record" the VSTi while playing live.
 

Try doing what you suggest with 2 or 3 (or more) instrumentalists wanting to record VST/VSTi output at the same time. If you can get that to work, I'd love to hear about it.

jatoth
If you assign CCs to physically manipulate the parameters, "live input" should work.
 
Does it not?
 

 
There are synths that require user interaction via the interface which cannot be controlled by midi. 
Newscool example:  http://youtu.be/Lg_2WVNGN...u9stuavxCp5UtC7c4fmfWg

Earwax
 
Not if the final audio from the guitar is the output of an amp sim and VST effects in Sonar. If Sonar could playback and record output from VST/VSTi at the same time, from multiple sources, there would be no issue. It can't. In my examples, more than one person wants to record VST/VSTi audio at the same time. No way can Sonar do that.   

I'm really trying not to be dense here, but this doesn't make sense to me. SONAR can play the output from a VST and VSTi in real time (plus any latency, of course) and record the input to them at the same time, so it can reproduce the sound heard during the real time performance at a later time. Sure, it won't physically record the output onto a track, but why would that matter? The object of recording the real-time output would be so you can play it back at some point, right? But you can play it back now, with the same results as the real-time performance (with the exception of stochastic devices that produce variable outputs even with the same input). 
 
With non-stochastic devices, I'm just not seeing how there would be any difference in functionality or sound between hearing something in real-time and recording its output so you can play it back, and hearing something in real-time and recording its input so you can hear the same output you heard originally when playing it back. 
 
I'm really not trying to be argumentative or anything, I just don't understand how there's any practical difference between the two yet people here seem to believe sincerely there would be, so I want to know what I'm missing. However if this relates only to stochastic devices, then I do understand the difference.

You can record anywhere you can put a VST using this.
http://www.meldaproduction.com/plugins/product.php?id=MRecorder
 
You can also loopback outputs if you want either virtually or with cables depending on your interface.
 
And that said, I've really never needed it. Ardour for Linux has a feature to warn you when you have a routing that could cause an infinite feedback loop, but I like that Sonar doesn't have to warn you of this because it is basically not possible unless you implement hardware to do it.
 
This has not been something I need. Frankly, I like it without it. I can totally achieve it using the RME if I want and I really never do (at least for the purpose of recording).
 
I have used the RME Loopback feature in order to get the Main Outs routed into RME DigiCheck, but that's a massively different conversation.

Anderton
 I just don't understand how there's any practical difference between the two yet people here seem to believe sincerely there would be, so I want to know what I'm missing. 

Let's ditch the Hendrix example.  I was just trying to fabricate an example, but it appears non analogous to some.
 
Tell me how to do this WITHOUT physically patching my sound interface.

Scenario:
I'm making a video on one of your incredible amp sims and describing how cool it is. I'm ALSO using a screen capture program to record the video. Sonar for audio, for obvious reasons.  So..., I'm recording my mic audio with Sonar AS I'm playing my guitar into one of your amp sims.  I am making adjustments on the amp sim to show just how lovely and flexible it is. As I'm talking and playing, I'm also twiddling knobs on the GUI.  Obviously I'd like to playback the audio from my guitar/Anderton amp sim, along with my microphone, just as I had done on my "original performance".....which I will mix in Sonar and sync with the video in Vegas.

Now...the problem occurs when I go to play back my audio after my performance.  I've recorded my guitar "clean" into Sonar, granted while hearing the amp sim while doing so.  But on playback, the amp sim is NOT changing, as it did originally, when I was turning the amp sim knobs.  Had I been able to record the virtual audio, I'd be OK. (because it would also reflect what was done in the video).

IF Sonar can record and ALSO be in "write" mode...to duplicate my knob twiddles, I'm ok.  But that's not possible either.
Therefore the solution is to record the virtual instrument audio at the same time.  

Now... I've actually done this type of video....., physically repatched my audio interface, and of course muted where needed to prevent feedback.
A "flip of the switch" inside Sonar would be nice.  Apparently the RME interface above can also do this, My interface cannot without physically repatching.  And I DON'T want to use the audio in my video screen capture program.  I want to mix it in Sonar before importing the audio mix to my video editing program.

________________________

All that said, Earwax's band example playing virtual instruments AND recording in Sonar also "hits the nail on the head"....  MY computer CANrecord two people, WITH low latency, using virtual instruments...  BUT I have to do some patching and feedback prevention.  Unfortunate for the person who hardwires their audio interface, and is unable to do this.  
But wait !!!!   if the "routing" could be done in Sonar, problem solved!

________

I'm struggling to explain why this is different, and unfortunately, hypothetical examples are not being clear.  Hopefully the TWO most recent examples above, illustrate what Earwax and I are trying to say and WHY it would be nice.

Craig...btw, here's a video where I did what I described above.  I HAD to record the audio of the virtual amp, as you will see, as I made changes that wouldn't/couldn't be duplicated later on playback.

(I don't expect you to watch the whole video...but in the middle of the vid, when the amp sims come up, you'll see what's happening.)
 https://www.youtube.com/watch?v=BlxxTnFGRtk
 
EDIT... PLUS I recorded Toontrack EZD2 live as well... However NOT using MIDI and lining up again.... as that would have been too complicated at the time as you can see.

mixmkr
IF Sonar can record and ALSO be in "write" mode...to duplicate my knob twiddles, I'm ok.  But that's not possible either.
Therefore the solution is to record the virtual instrument audio at the same time.  .

Sonar can do this.

Set up audio track pointing at the right interface audio input.
Put amp sim in fx rack/pro channel.
Enable amp sim (or other plugin of choice) or other controls as required for automation recording.
Enable track for recording.
Hit "record" and play/twiddle away.
Result = an audio track plus automation envelopes duplicating control movements.
For MIDI simply use a MIDI track instead of audio.

Now, if a plugin has been designed such that it doesn't expose it's controls for automation (or, unless you only need to change one control at a time, to MIDI control) then that obviously won't work. There is however no built in limitation to Sonar that prevents recording automation and audio/MIDI at the same time.

I do it very frequently when working with hardware synths, and if the synth controls send MIDI I generally tweak them and record the output (as MIDI) rather than editing in curves afterwards. Give or take a little oscillator drift or free-running LFOs that don't start at the same point in the wave cycle every time I get identical results playing the MIDI back to when I recorded it. With the added bonus that, unlike my voltage controlled stuff, doing a punch-in if I'm not happy with, say, a filter tweak, is pretty easy.

mixmkrMY computer CANrecord two people, WITH low latency, using virtual instruments... BUT I have to do some patching and feedback prevention. Unfortunate for the person who hardwires their audio interface, and is unable to do this. But wait !!!! if the "routing" could be done in Sonar, problem solved!

Maybe I'm being dense, but unless you're using both microphones and open monitoring rather than headphones, what feedback?

Routing two monitor sends to a couple of people playing VSTis can be done as follows.

Two incoming MIDI ports for the incoming MIDI, two sets of MIDI tracks each receiving from the right MIDI port on the right channel -> two audio tracks with the VSTis on them should work. Enable input echo for monitoring, mix via the track faders (which control output volume when input echo is enabled) and there you go.

If you need different mixes for two musicians then you need an interface with multiple outputs and two busses, each pointed at a different hardware output (rather than master). Use a send from the VSTi audio tracks to both busses then use the sends to set up the mix like on a hardware mixer. Routing done, and, apart from you can't get two mixes out of an interface with insufficient outputs, all in Sonar.