That tradeoff is real. Pre-rendering flat clips reduces complexity and can make a heavy project easier to edit. The cost is that every pan, horizon decision and field of view becomes baked into the intermediate.

If the project is still changing, I prefer to keep reframing editable as long as the software and hardware can handle it. If performance is the problem, proxies are usually a better first compromise than permanently flattening every shot.

Multiple delivery formats push me even more toward a late reframe. A composition that looks good in 16:9 may put the subject too close to the edge in 9:16.

For social versions I usually create a separate reframing pass rather than crop the finished widescreen export. The source contains enough view information to make a deliberate vertical composition.

VSDC's direct 360 workflow is useful when the goal is to keep the virtual camera inside the actual edit. You can cut first, then refine the view for the shots that survive instead of spending time reframing material that will be deleted.

For a weaker PC, I would test one representative native 360 clip with a proxy. If the reframe remains editable and final export still uses the original source, that gives you both responsiveness and revision flexibility.

What about stitching and orientation? I have had cases where a converted file looked correct but another application no longer understood the horizon the same way.

That makes me hesitant to add unnecessary conversion steps before I know which app needs the original camera metadata.

Exactly. Keep the vendor original as the archive master, especially when stabilization, horizon or orientation metadata matters. Use GoPro Player when it solves a specific compatibility/export need, not automatically as the first step.

A good rule is: flatten early for simplicity when the framing is final; keep 360 editable when the framing is part of the creative edit.