the answer to your omni question is maybe and maybe not. there could be a all manner of things that cause it to cycle. From note name to note velocity to note volume and that's just the easy stuff. there could be some LFO that causes them to cycle. they could be using the that step sequencer ... if that is the case you could -maybe- tell it to stop stepping.... but all that can be very complex. That thing, I'm sure is like climbing Mt Everest if it's remotely like Absynth which I think it is regarding complexity.
But you have two issues here. The morph of Omni which you know is happening but you are able to identify the good from bad. So say ideally you wanted 60 bars of that one sound you like but you only hear it happen 12 times in teh full song length.
So you selected I/O points for your render and caught those 12 segments.
But now it seems you are saying, yes those were the 12 spots of good sound but upon replaying them as simply an audio wav now they simply don't sound the same. so you had....
OMNI - playing 12 instances of 'good' out of 60 total.
RENDER - those 12
12 Audio clips - when placed in project now no longer sound the same 'good' --- when in fact, they should have sounded identical.... that's a whole new issue beyond Omni. That is a technical Audio Render issue. At least imo, based on what I think you are doing/saying.
Those 12 snippets you captured should sound exactly like the parts of the Omni performance that you liked...
Exactly unless --- the output of Omni gets additional processing. Or the new snippets get processed by some means as you listen back to them.
IOW, you should be able to place those snippets on a blank track. Mute Omni and hit play. you should hear exactly what you heard in Omni minus the altered stuff you didn't like. then you could copy/paste those clips to fill your 60 tracks as desired.
That's why everyone told you to render your midi to audio. When you are making your track you change notes, voice, etc. Then when you say... ok, that's it, that's just what I want.... that's when you render it to audio to preserve it exactly as you hear it. As though a musician was playing it live on the track.
Do you see that even if you figure out Omni, you will still have a render issue ( if I am understanding you ) until you figure why your renders sound different from your developed midi 'performance' track? You should be able to have those two things.... the Omni track playing live ... and the rendered audio file from it.... and toggle back and forth on solo/mute and not hear a difference for any given exact same time segment.
Now on yours 01:05:28 may not sound like 03:07:22 for the same midi notes due to morphing patch,,,, but 01:05:28 should sound identical on the rendered audio AND the live Omni playing. ... it sounds like you are saying it doesn't. 01:05:28 RENDER = 01:05:28 OMNI --- no matter what the actual sound is at that time ... they should be indentical.
btw - OT: this is why I say it may be difficult chasing down what Omni is doing. This is the developer of Absynth and this is 'one note' ... https://soundcloud.com/brian-clevinger-1/dissolving-ice-tunnels and I'm sure Omni is equally as complex in different ways.
Oh, btw... LEAD SAW VIOLIN means it's a lead voice which is usually bold and it's a 'saw'tooth wave and obviously a violin which could be sampled or generated depending on how Omni does that... So in synth talk you will see things like "saw" pad, "square" pad, "sine" pad ... they are telling you the generalized character of the patch. Sawtooth, squarewave, sinewave. the "lead" tells you it's probably not subtle like say "atmosphere" ... like 'ethereal sine rain' kinda gives you an idea of what to expect