Back in an old video, @LoLeif used a Roland GP-10 with MG2. If I remember correctly, he set up six instances of MG2, one for each string’s output. I believe that he did this for tracking purposes.
I have an RMC Fanout Box with UA Apollo audio interfaces, so I can output hex guitar into to my DAW.
My question is, will there be any advantage from a tracking accuracy point of view?
Also, I think there might be some advantages to using a hex setup.
Say you want to create a multi-timbral patch rather than layer two or three instruments. It might be easier to create a few instances of MG3 only on string outputs as appropriate.
Using mono-timbral hardware synths as single string voices, so you can use each synth’s pitch bend range individually.
I did set up with a hex PU and a Roland GP-10 on six channels like you describe, to achieve what I hoped to be a vastly improved tracking. And tracking was absolutely improved in the following two areas:
The minor second issue we all struggled with in MG2, became a non-issue, and
2). I could finally play big chords with whatever voicings i could imagine without having issues with beating notes, and the issues that would follow from that.
But if you remember from those videos, I also always felt a bit uneasy with the dynamics, which I felt at the time was suffering somewhat. I have since come to understand that this is most likely an issue stemming from the way the hexPU are built. There is still a degree of crosstalk between the six microphones in the hexPU that needs to be dealt with in one way or another. And it seems to me that the standard solution for MIDI Guitar setups have been to heavily gate the signal at the quieter dynamics end of things, and also to overly compress it to achieve maximal sustain from what is left. Overall, this has resulted in a tone from the hexPU systems I have tried that isn’t really dynamic nor especially sustaining.
But is there advantage from a accuracy point of view?
Maybe, on a theoretical level. If you experience a lot of note beating, and that the MG3 tracking doesn’t pick up on everything you play (however clean you play it), maybe there is a case where the hexPU and six instances of MG3 i(n Mono) would be more accurate. And certainly if you are looking to use MG3 for sheet music. Most of those software reliy on you providing separate string input.
But from a CPU resource POV I would probably wait for a JamOrigin dedicated HexPU solution, rather that to set up with six instances of MG3. But, hey that’s just me.
Personally I have no issues with the polyphonic tracking accuracy. Especially with the way we can use MPE nowadays, which feels so natural for stringed instruments.
I will probably reasess my stance on the hexPU use once there is an dedicated MG# solution, but until then I can safely say that the dynamics, sustain and accuracy in MG3 as it is today, all together is way more than I dared dream about back when I was setting up with the GP-10 and the Ableton templates with 6 channels of everything.
i believe this concern does not apply in @scratch17’s case. the rmc pickups do not suffer from the limitations of the gk2/3.
setting the gates properly was the trickiest bit for me. after that it was smooth sailing (with mg2).
the only argument i have against it is that it can be tedious to maintain. my main argument for - using slide with piano patches. i haven’t tried that with mg3 single instance yet, but with 6 instances of mg2 it is pretty flawless.
You are probaly right. I haven’t tried al kinds of systems. I only have GK2, GK3 and and a ghost Hexpander MIDI interface setup on my MOOG guitar to go by. And the problem hasn’t necessarily been the pickups themselves to begin with, but that the MIDI conversion to be done presupposes that the signal has these traits generally and treat the signal as if it is this bad.
So this is of cousre not a problem for anyone that uses a breakout box directly into an audio interface, thus skipping the “bad” interpretation.
Thanks to all for the replies. My Brian Moore i213 is ready to go.
But my Hamer Duotone used a GK3. So I’m considering a second pickup option; the Nu2 hex pickup from Cycfi. Nu2 has a FRFR hexaphonic line level (low Z) output through their Nexus breakout box. I plan to add one to both guitars.
And, potentially one of their specialty Nu2 pickups made for acoustic steel string guitars for my Taylor 710 BCE. MG3 with a truly awesome acoustic guitar. I can imagine creating a layered patch with a 12 string (using DivisiMate with the Transpose plugin) and a nylon string guitar.
Why add the Nu2 to the Brian Moore when it already has a hex output? To add the wonderful sound of that pickup to my DAW. And a second Nu2 won’t add significantly to the cost (<$200).
@LoLeif, I never knew about DivisiMate. Thanks for the heads up.
I spent today looking at videos of it in action. It seems to me that combining MG3 and DivisiMate can make for an amazing tool for creation. For example, by using the Low, Voices, and Mel ranges to create fretboard splits. For example, you could use the ranges for a Unison grouping on the Mel range; and a triad on the voices range or low range. I think the plugins in Divisimate could also really add to the tools in combination with MG3.
Measured against its potential, MG3 is still a sleeping giant.
MG3 can identify the fingered note very well and output it in MIDI. HexMG3 identifies the string of the fingered note. In my opinion, this information is the “missing link” for significantly expanded playing possibilities, especially for fingerpickers and bass players who also play rhythm parts in small bands.
The algorithmic evaluation of the additional string information opens up further playing possibilities.
the position of the fingered note on the fretboard of the guitar can now be determined explicitly; explicit guitar tabs, e.g. for GuitarPro, would then be possible
the unique position of the fingered note on the guitar can be used to create different keyzones, which can be used for sounds, but also for keyswitches and other switching operations.
guitar audio can now be switched off per string, allowing for more differentiated playing.
wind or string parts can be played in a more differentiated way, especially for fingerpickers
the additional information about which and how many guitar strings are involved in the current playing can be algorithmically analysed and used for additional playing control, e.g. individually played strings in a single sound voice, several guitar strings played can be output in an orchestral sound, for example.
Everytime I have tried to use GP10 or VG99 output into MG3 I have obtained a mixed bag of results. So to me it’s better to use guitar direct into Audio Card or as I am doing lately use the Guitar out to feed Audio Card (should be a quasi-clean passthrough).
Warning : This has nothing to do with using six instances of MG3 in a DAW but rather substantiate that if you use processed guitar (Virtual or real) even in a mono output, MG3 get confused about dynamics. Lofileif in this respect is absolutely right and I wanted to post this because thanks to his posts now I know that my ‘gut feeling’ had technical roots.
i replaced the 6 instances of mg2 in my ableton live setup with mg3 and the cpu usage is much higher.
mg2x6 was 36%, mg3x6 (no display) 50%, mg3x6 (display) 65%.
by display i mean the vst window is open, although only the current channel’s instance is displayed.
i believe the new visuals are the key factor. might it be possible to have a lower rez / less -animated option (ie: the cables are shown but don’t wiggle, meters are more rudimentary)?
I’m no coder but it seems like the early beta phase of these software things usually go through lots of optimization right up to and beyond final ready to release stage. So you may see it get better. Plus there is a Hex version in the works too
i do expect that the shipping version of mg3 will be optimized.
this issue is very low priority, the audience of potential users is very small.
but i brought up the possible link between graphics and cpu usage because the are probably many users who never look at the mg screen, as the action is elsewhere in the daw.
those users may prefer to allocate those cpu cycles to synths or efx instead of graphics.