Outerra forum
User mods, screenshots & videos => Other => Topic started by: HiFlyer on March 17, 2016, 10:56:50 am
-
If you want to cover the entire planet with a realistic facsimile of what is there in real life, the only viable system to use that would not require 750GB+ of data is one based on the Olsen landclass (LC) approach. This allows a whole-earth sim to be shipped in about 10-15GB worth of files. In FSX/P3D this Olsen LC system is limited to about 128 classes - far too restrictive to depict the vast terrain variety on earth.
There has been lots of banner waving and espousing of supposed current-tech engines like Outterra, Unigen, Unreal (insert name here), but those engines simply cannot scale to whole round earth models and push the insane level of geometry to the horizon as the 'old' FSX engine can. Despite 10 years of progress in gaming engines nothing can touch the FSX engine for far-horizon geometry rendering; the other engines are all mostly designed for limited area arenas. Even the excellent DCS engine is limited in its scope and relies on vast amounts of data per area.
There is no need to use a new engine if the fundamentals are right. Tech can be layered on top of 25 year old code and work perfectly well. I don't get some of the naysayers and blue sky dreamers on various forums who are bleating for new engines, they don't understand that you need a proper foundation to build on first.
Strong words, and you can see that somebody definitely has some skin in the game on DTGs new sim. However, my post here is not an invitation for some rally around ye' old flag and defend Outerra posts, but I would like to hear from knowledgeable people regarding the technical merits of the argument.
-
That's John Venema's old argument against modern engines, though it's unclear why he keeps throwing Outerra into that, when he then refers only to "limited area arenas".
Obviously the existing engines aren't designed for the whole world, and even though they tried to implement efficient streaming for large worlds (like Crytek's Realtime Immersive), it didn't go very well because of tons of content to stream (and create), while also attempting to keep the graphical details of the games made for the original engine, or else it would make little sense to try it at all.
I would say you can't go too far from the original design, if it wasn't designed to scale that far in the first place.
Just as the FSX engine can't scale down into finer detail well.
-
Can I laugh now?
-
I would say you can't go too far from the original design, if it wasn't designed to scale that far in the first place.
Just as the FSX engine can't scale down into finer detail well.
A good point.
-
The FSX engine IS a great piece of software....but you have to remember it's 10 years old now: It was designed at a time when the expectation for hardware was higher clockspeeds and single CPUs and before internet delivery of software took off (at least for multiple GB of data). Furthermore Outerra doesn't (yet) handle the vast array of subsystems that FSX does (i.e. AI flightplans, AI shiplans, AI road and boat traffic, Air Traffic Control, landclasses, integration of higher quality DTM, aerial photo or landclass data, seasons and weather etc etc.
Apples and oranges really but the guts of FSX + fractal refinement and lighting of Outerra would be awesome.
-
The FSX engine IS a great piece of software....but you have to remember it's 10 years old now: It was designed at a time when the expectation for hardware was higher clockspeeds and single CPUs and before internet delivery of software took off (at least for multiple GB of data). Furthermore Outerra doesn't (yet) handle the vast array of subsystems that FSX does (i.e. AI flightplans, AI shiplans, AI road and boat traffic, Air Traffic Control, landclasses, integration of higher quality DTM, aerial photo or landclass data, seasons and weather etc etc.
Apples and oranges really but the guts of FSX + fractal refinement and lighting of Outerra would be awesome.
I'm not the most tech savvy person out there, but I'm pretty sure that anything AI shouldn't have an impact on current performance as the CPU should handle that (and it's pretty free in terms of used computation power), having more models being rendered doesn't seem to tax the performance a lot either (at least based on the tests I've seen here made by other forum memebers). Same with the rest, performance test so far clearly show that Outerra is highly efficient at what it does.
The problem on a straight comparison is that Outerra is a WIP engine and FSX is a finished simulator with years of development on top (both from MS and Dovetail). The later will inevitably become redundant eventually because that's how technology works.
If starting -sort of- from scratch to properly handle new technology (be it graphical features or whatever) wasn't necessary, Valve would still be using the GoldSrc, Bohemia wouldn't be trying to come up with a new engine to handle larger landmasses+longer range rendering and Fallout 4 wouldn't perform so horribly bad (because that's an old engine with a lot of glitter on top).
Just my opinion from a player's perspective, I could very well be wrong, but that's what I can see from here.