search

KochieBR / Avalanche-Plumes

Solid Rockets Waterfall Plumes
Other
2 stars 1 forks source link

Low performance with multiple M55A1 boosters. #3

Open JacobB094 opened 4 months ago

JacobB094 commented 4 months ago

As a 3090 owner I don't often complain about performance, so the problem is acute. The current FX for M55A1 (Minuteman first stage) is particle-heavy, which causes lag when there are multiples of them. The problem, I believe, is particle-based plumes combined with the fact it's a four nozzle motor. Seeing as a Saturn INT-19 can possibly carry up to 12 of them, it results in 48 plumes if they're all ignited at liftoff (normal Saturn II configuration ignites 8, which is enough to cause lag). I would suggest moving this specific plume to a Waterfall-only setup, which should run better, hopefully making that Saturn II variant usable for people without a VR-grade rig like my own.

I already made an issue on BDB github pertaining to adding a single a smoke transform for the smoke plume. I'm not sure if anything can be done about it until that happens, save perhaps temporarily disabling the smoke.

KochieBR commented 3 months ago

Hello Sorry for the delay your problem is not related to the smoke plumes generated by smoke screen as they are limited by the global particle limit i have tested an saturn INT-19 build from BDB and i had a severe fps drop lighting all 12 boosters and even just 8 up all as you stated but that performance drop is caused by the amount of waterfall plumes rendering and on that i dont have any control

JacobB094 commented 3 months ago

FYI, the M55A1 has a smoke transform now. While it does seem that particle limit alone won't solve the performance issues, it certainly would make the plume look much nicer if the smoke wasn't being culled as aggressively as it is now. Currently, it's a problem even with INT-14 (just 4 boosters), at least at liftoff, when the culling is quite visible.

BTW, it should be quite possible to model an effect that would fuse all four plumes into a single model. Of course, it would be quite a complex one, but it might improve performance, as well as allowing some nifty plume expansion effects.

KochieBR commented 3 months ago

As I said, I have tested the performance, and even without any smoke, it still has performance issues. However, I will be updating to a single transform, and I will also experiment with your suggestion of fusing all four plumes to see if it helps the performance