Distant Lands and OpenMW

General discussion regarding the OpenMW project.
For technical support, please use the Support subforum.
User avatar
Mistahtokyo
Posts: 139
Joined: 07 Sep 2013, 18:31

Re: Distant Lands and OpenMW

Post by Mistahtokyo »

It is worth nothing that while MGE's method has flaws, people in general absolutely love it because relative to the original game it is still a tremendous visual improvement. Basically, if people here aim for too much perfection in the system right off the bat, it'll never end up getting done, whereas something that works well enough can be enjoyed sooner rather than later.
User avatar
AnyOldName3
Posts: 2668
Joined: 26 Nov 2015, 03:25

Re: Distant Lands and OpenMW

Post by AnyOldName3 »

MGE XE doesn't really have enough access to do anything better than what it's currently got. That means there's absolutely no problem with them maintaining their current system. If OpenMW adds several not-quite-perfect solutions, that's a lot of code to maintain, and people have to spend time making sure any other new features are fully compatible. Perfect is the enemy of the good, but the antagonism goes both ways.
Chris
Posts: 1625
Joined: 04 Sep 2011, 08:33

Re: Distant Lands and OpenMW

Post by Chris »

Mistahtokyo wrote: 05 Mar 2020, 02:05 It is worth nothing that while MGE's method has flaws, people in general absolutely love it because relative to the original game it is still a tremendous visual improvement. Basically, if people here aim for too much perfection in the system right off the bat, it'll never end up getting done, whereas something that works well enough can be enjoyed sooner rather than later.
At the same time, if you're going to do something, do it right the first time. It doesn't have to be perfect or complete right off the bat, but it should have a strong foundation that doesn't need to be tossed out and redone. If it doesn't have a good foundation to build off of, it creates more work than necessary and further delays a proper implementation. So if someone interested in distant land and has the technical knowledge to help code it, it's strongly encouraged for them to work out a method that can stick around, rather than one that'll need to be removed.

Currently, the main problem with MGE's method is that the distant land needs to be pregenerated; for something like vanilla Morrowind, this is fine since the world is pretty much static (sans Ghost Fence, Fort Frostmoth, and Mortrag Glacier(?)), so you just need to do it once. However, this starts falling apart when you want to have a more dynamic world that has big landmarks that can change, or if you want to allow people to add and remove mods semi-regularly that change the landscape (which can happen if you play with multiple profiles or mod-sets for different characters) without the hassle of having to remember to regenerate distant land.

Now, that doesn't necessarily mean MGE's method fundamentally can't be used, but it should be an internal and automatic process that can update during play, rather than something the user has to do ahead of time.
User avatar
Mistahtokyo
Posts: 139
Joined: 07 Sep 2013, 18:31

Re: Distant Lands and OpenMW

Post by Mistahtokyo »

I personally like having to manually generate the distant terrain since it lets me, to a degree, fine tune what does and does not get created. An automatic system, depending on the implementation, can take away that freedom or make it more confusing for the user.
User avatar
AnyOldName3
Posts: 2668
Joined: 26 Nov 2015, 03:25

Re: Distant Lands and OpenMW

Post by AnyOldName3 »

You don't need to care about what does and doesn't get created with an automatic solution. It will always produce the correct answer. The huge lists of different meshes to include in MGE XE distant terrain all came about because it isn't able to adapt to a changing game world or even determine what might change. You've never opened up The Witcher 3 and been disappointed that a building suddenly disappears when you get too close to it because your distant terrain list didn't take into account that it only gets built half way through the game.
User avatar
Mistahtokyo
Posts: 139
Joined: 07 Sep 2013, 18:31

Re: Distant Lands and OpenMW

Post by Mistahtokyo »

AnyOldName3 wrote: 06 Mar 2020, 20:46 You don't need to care about what does and doesn't get created with an automatic solution. It will always produce the correct answer.
Well that is objectively false. What if I want to prevent certain statics from generating to improve FPS some? I trust that the automatic method won't screw up? Hope any multi-part buildings or similar won't get affected? Oh just have some blacklist/whitelist for those? Then you're still relying on manual input. Or what if some particular landmark or feature is something that I want to allocate more fidelity to (viewable from greater distances, with higher poly count, etc)? I hope the auto method works? Unless you have some hyper intelligent a.i. determining the distant statics even an automatic method will involve compromises, just as a manual one would. It's not a holy grail solution.
The Witcher is a poor example, tbh, since while it is an open world your movement across it (especially due to lack of flight and limited movement in certain crowded areas) allows for a different host of optimizations.
For the most part manual = more control, while automatic = relinquishing that control.
User avatar
AnyOldName3
Posts: 2668
Joined: 26 Nov 2015, 03:25

Re: Distant Lands and OpenMW

Post by AnyOldName3 »

No.
User avatar
sjek
Posts: 442
Joined: 22 Nov 2014, 10:51

Re: Distant Lands and OpenMW

Post by sjek »

there's lod levels what can be used. i don't see that automatic and manual would rule each others out.

in morrowind mipmaps are used for some object's textures and it's kinda easy to set up.

let's say you got a telvanni tower. rendering only parts of it would seem lunky and doing automatic polycount degrease while looking good could be costly (given that morrowind meshes doesn't necessarily need that but let's think it does or other high poly big mesh) so what solution could be used .?

mipmap levels with some predetermined way affecting polycount or some clever way using blender toolkit keeping texture same but deforming polymesh underneath. wouldn't play well with uv maps but maybe a possibility .?


edit:.. in the other hand. one way of doing would be conjure that tower when it's still under fog. determining good parameters would be field work and manual thought. or so would imagine.
would't necessarily work in games without fog.
edit2: then again.. morrowind weather system with fog distances would vary the performance in this case.
User avatar
AnyOldName3
Posts: 2668
Joined: 26 Nov 2015, 03:25

Re: Distant Lands and OpenMW

Post by AnyOldName3 »

When I say "No." I'm not necessarily saying that we'll come up with some magic system that makes everything work perfectly, but that the system we'll have will work perfectly with good input data. This will probably mean modders will have to do some work to get everything performing as well as possible while looking its best, but there shouldn't be a bazillion switches, dials and filters for users to set up. The polygon count of a whole Morrowind cell isn't wildly different from the polygon count of an overly detailed armour mod for Skyrim or Fallout 4, so we shouldn't need to be very aggressive with culling small objects from distant versions of combined meshes, and if something genuinely is causing problems, it'll probably be modded content, and adding a NiLODNode to the Nif where the near version has the mesh and the far version doesn't will be enough to exclude it from distant versions of any combined meshes it would be part of.
Xenuria
Posts: 45
Joined: 26 Feb 2017, 22:35

Re: Distant Lands and OpenMW

Post by Xenuria »

Running the latest build with All Grids loaded removes the needs for Distant LOD or statics. If you have at least 32GB of Ram and a SSD you can do it but it runs at 3-8 fps and takes longer to load a save than GTA Online. If you have a really good CPU and Video card you can load all grids with TR but I haven't been able to do that. TR is way more lore accurate and detailed so it's harder to do with all grids loaded.

Until we get to OpenVulkanGraph the biggest bottleneck is CPU single core clock speeds because DRAW calls. So if it's doable on a consumer grade PC to render the entire gameworld (without TR) you don't need distant statics. That avaricious german fellow working on RTX hacks is going to release a video showing morrowind in RTX and that is going to draw much needed attention to this community.
Post Reply