Whoa. The more CIG talks about its crappy planetary generation system the worst it sounds and the more I appreciate FDEV´s or HG´s elegant proc gen client based solutions to be honest...
The 400bn stars are overkill gameplay wise but the generation itself makes a very believable world.
ED generates a world full of stars. 1000 players converging on a single planet is a different matter and more how they make matches with their networking solution which is quite nifty too.
Well, indeed, the comment was addressing one of the main issues acknowledged by CIG there in that quote, namely the server streaming limitation when all players in that server are spread out. Of course in the case of SC the issue is with all assets in general, planets, biomes and all content and textures etc in them being a subset therein. But that is precisely one of the issues that FDEV´s and HG´s solution manages to very elegantly solve with client based proc gen, as opposed to planetary content based on actual memory loaded assets managed by a server for all players in it.
Yeah, CIG´s "hand made" planets feel even more fake than NMS ones to me tbh.
In what way exactly do they feel fake? They're sci fi sure, but nothing too outlandish except the pink gas giant maybe.
. But don’t feel like we’ve got conclusive info at this point.
Yeah, in terms of asset handling, memory loading required and limits thereof I think it is quite clear cut, no? That is precisely the main value proposition of proc gen and the main reason Delamar had to go and the reason behind the CIG comment that started this discussion. That was the case in 1984, and it is still today. If the client can handle the planet asset generation thanks to proc gen streaming, that is a ton of things less in memory for the server or clients themselves to handle and likely allowing for higher tick rates.Yep that’d be the thing. Groupings driven primarily by proximity etc, I believe. (And therefore none of the same issues with having to handle all planet-wide activity etc)
But, y’know, instancing. So… * spitting noise *
I dunno, I think it’s still an open question. Would like to see a pro breakdown of how the techniques differ in terms of performance load, and how both might perform under SC’s proposed networking, for sure. (There seem to be some comparable issues that mean they might not differ hugely. Both need to handle texture at scale, scatter generation, flora to varying degrees, for example. Both will need to generate nav & collision meshes for large areas etc.)
It does seem strange to me that CIG have to network the collision mesh itself. (You’d think that could be client side, with just player behaviours networked etc). Would certainly make a difference if a proc gen system definitely could be purely done on the client. Would be good to hear all this from the horse’s mouth though.
TLDR: I wouldn’t be surprised if CIG had done another number on themselves in the pursuit of fidelity
Agreed. I just wish atmo flight and re-entry was more visceral and realistic, requiring at least some amount of piloting skills, not asking for full orbital mechanics like KSP but maybe aero effects such as lift, mach compression, the effect of center of gravity vs center of lift, etc. and justify the "aero optimized" ships, vs those which are clearly flying bricks. That would bring more personality to each ship, also it would help put people in different roles than just flight jock with full control of all the weapons and promote a bit more team play, since SC is a lot less annoying when playing with a group.
)I think, by this point, the key failure of those of us who are skeptical of CI-G is that we have never actually been able to nail down what timescale the estimates for delivery are on.
