MusicVillan. Thanks for encouragement regarding my post. It was directed toward increasing variety rather than reducing variety which prompted me to post based on this comment in your first post;

<< "One major issue of BiaB, is Playing Pattern Inconsistency. Each RealTrack has a sub style A and a sub style B, under each sub style, there are different patterns being played and recorded, which are uncontrollable, and therefore causing inconsistency of the generated track, ruins the musicality.

Currently, the only solution is using Partial Regen to manually regenerate bar by bar until the playing pattern is correct, which is very time consuming.

Guitar arpeggio tracks, such as RealTrack 1831, is a typical example. You can't generate a track with consistent arpeggio patterns being played in each bar, unless you manually mess with the partial regen." >>

Your remarks are correct and RT1831 did give unsatisfactory results from what you were seeking. I see in the Bad Version SGU you posted that it required you spend time and effort to edit the generated track. My post intended to demonstrate that more variety results in less post generation editing. It can give you what you want when you generate a song without having to spend time and effort with what you described as inconsistent patterns.

My initial post may benefit users in the future but it was conceptual and may be more suited in a post of its own.

However, even though Andrew has some ideas to improve results using your workflow, I applied the steps I outlined in my earlier post and BIAB resolved your issue in less than 30 seconds. I loaded your SGU Bad file, opened the RealTrack Picker and did a search for the Artist playing RealTrack 1831 and the search found 32 RealTracks (folders) for Dave Cleveland. Well, the first RealTrack below RT1831 is RT 1829 which is Mr. Cleveland playing the same instrument, same setup and tone, correct tempo because it is an arpeggio pattern RealTrack that plays the exact same riff as RT1831. I didn't need to look any further.

Adding RT1829 to a vacant track and a single generation produced enough of the riffs without the strums that only mixing the tracks by bouncing between the two using RT1829 to be active in the measures that RT1831 strummed those measures. I matched volume 90, panning -45 and reverb -55 and the two tracks matched exactly.

The concept I intended to present was not that your technique of loading a single soloist and partial regeneration 40 times to edit is incorrect. It isn't. But BIAB has features that are not commonly known or used to get what you're looking for and by using more material from additional RealTracks that are similar in production to have BIAB generate what you need in a single generation rather than have to use post generation tools for editing to get what you need.

Audition RT1829 and you can hear the similarities between it and RT1831 and you'll get an idea of the concept I mentioned earlier. I'll be glad to answer any questions.

Single Generation Render with no partial regenerations

Attached Files (Click to download or enlarge) (Only available when you are logged in)
1.jpg (198.01 KB, 209 downloads)
2.jpg (368.7 KB, 209 downloads)
BIAB 2025:RB 2025, Latest builds: Dell Optiplex 7040 Desktop; Windows-10-64 bit, Intel Core i7-6700 3.4GHz CPU and 16 GB Ram Memory.