Awesome film and the type I really like to see more of. Your workflow is pretty much matching the same conclusions I've set on as well.
Creating layers in your editing program is a must. For sound layers, i'm currently trying out this mix: interior/exterior/vox/score/fx/misc
Setting up the faders during course adjustment mixing can take out some emotion, but the goal is to prevent too low/loud sounds form jarring viewers. Expect to add fine tuning to bring back emotions and a follow up step.
I also had Vegas crash on me all the time during sound editing. I've learned to save my work all the time and with a new file name daily/hourly. We've all had programs teach us this lesson.
As for banding, my current guess is youtube transcoder will probably reintroduce it, but I've still not tested this part of the process. I use a video noise debander plugin, it removes the banding, but also introduces fuzziness, so I then apply it only to the sky color to limit side effects. I've always had banding issues, and would use some kind of filter to remove it, but it would come back in the final render back in the days when I had to apply lots of compression to get file sizes under 100mb. I think I used a gaussian blur on the blue hue via a mask to get rid of it in Vegas back in version 6. 8-bit video will have this issue, especially if you start adding video effects. The game itself has it.
YT's encoder settings have always been subpar for those looking for high quality. I chose to avoid YT and use WMV2 with my own encoder settings when YT first got started. One could get better quality with WMV2 than with YT's 480p settings, a codec built-in to Windows. YT has gone through multiple upgrades since. I think their VP9 and av1 settings are descent, but their avc1 settings seem biased towards space/bandwidth savings. Even Netflix and Amazon had low quality settings for years in their video or audio settings. So I think 1440 and 4k videos might be best bet to get YT's better encoder settings.
For video capture, I use OBS, only because FRAPs stopped getting updates and fell behind. The builtin GPU encoders NVENC are mostly fixed in settings, but they are lightning fast and don't take up much resources. Might want to give this a try during your testing. I use CPU encoding during final film renders though to get best quality.
Mixing can be infinite. As far as peaks and loud stuff when the rest of the sound file is lower, that's what compressors are for, to pull that peak down so you can bring the background up. That said, using them wrong has bad results, using them right makes a big difference. IE,...each scene has a comp. I running those at 3:1 ratio with very little threshold like -1.20, when norm is around -3,.. just topping the loud peaks,....3:1 still retains dynamics.....on my main output I run more strict but light,.. at 5:1. We stair step them like that so the listener never detects changes,.. but they are a great tool. If you don't understand them I can give ya a quick lesson. That scene where the 51 suddenly drops in from above and loud (4:22 again),... that went into full compression so I could get the distant sounds up and audible, but was still loud...and attention grabber. Evidently, Vegas bought rights to an Ensonic (I think) digital compressor, which in their consoles is one of the worst. They have to be layered or they distort easily.
For outboard (of Vegas) audio editing I use Cool Edit Pro for decades now. Adobe now owns it and called Adobe Audition. Very good program. I was going to use a 360 degree stereo rotation panning on a file,....sounds like the sound is spinning around your head. lol. I though it was too over the top. lol
YT actually didn't hurt it much. Unfortunately the banding issue is on my end. It banded before I uploaded. I THINK what happened, my worst banding, in scene starting 4:22. that is two scenes rendered together as one. I think I made that early in the game and was using the Sony codec or processing, before switching to NVENC. The fact I rendered it as one scene and then rendered it as part of the video (twice) MAY be what happened and my own damn fault. Vegas viewer is lame, you don't catch that stuff until rendered and watch on larger screen.... I had enough of that film, wanted to move on, and just wasn't going to go through all that redue. other banding is minimal,.. but I want 0. I hope the game is still around when I figure it out lol.
I don't use vox unless the convo is part of the immersion. Sometimes the chatter just doesn't fit. I probably will use more if I do the next scenario.
How you mix the game is how your capture audio will sound. I spent time on that to get more dynamics and not a default straight line mix. unrealistic engine, flaps, buzzer and guns volume really mask other things that bring the game more to life. It's the only way to mix individual sounds before production.
Soon as I get this last banding issue solved I'm up and running. More videos in a lot shorter time. The only flaws would be slopping scenes/angles. I think you're better at that and history than me. I did barrow your style at the end with Peddy.
