i think there is an in-engine cap to things.... doesnt really matter how much you got, what matters is can the engine use it.....
So as far as adding more proxy meshes - in the xml file, there is a entityproxy=true
Because proxy meshes are allowed to have huge autohideDistance values, maybe changing all the meshes to proxy meshes would work?
@marvelmaster - could you try changing the meshes (of vivaldi's bank) to proxy meshes in the xml files, and then test a much longer autohideDistance value? I'm still exporting all the files, so I can't do anything in wcc_lite.
Also, when you export eredin, he's half a gig of models. Holy fuck.
Because proxy meshes are allowed to have huge autohideDistance values, maybe changing all the meshes to proxy meshes would work?
@marvelmaster - could you try changing the meshes (of vivaldi's bank) to proxy meshes in the xml files, and then test a much longer autohideDistance value? I'm still exporting all the files, so I can't do anything in wcc_lite.
Also, when you export eredin, he's half a gig of models. Holy fuck.
in the game files there's one XML and one .INI file that caught my attention. One is io_priority_table.xml located at Witcher 3/bin folder and the other is rendering.ini located at Witcher 3/bin/config/platform/pc folder. Both of these files have something to do with mesh loading priorities and values. I don't know if changing the value would do something in the game but it's worth taking a look.
Last edited:
i would also add that the level design in general changed a lot since the early builds and has been tailored to fit a certain performance " budget " ( and new zones that weren't there early are built just around that ), they reduced the meshes details in general for novigrad for example but they added more trees and grass in the zone and sometimes even objects , same for the swamp, ther's way more trees bush/grass in the final game, so you probably gonna have to reduce foliage in general to compensate
the real questions is how far the engine will go with this, how much " high " detailed meshes can it load at once without crashing
Unfortunately i dont know if i can get to my pc today...so progress might be delayed...
I can remember that i edited io priority a while a go but with absolutely no effect.
Carol might be right... if these xml realy describe whether a mesh is a proxy or not it would mean we just might have to edit the one xml of the main mesh
I already tested it a little..few days ago but didnt see an effect.
But i ll dig deeper into this
I can remember that i edited io priority a while a go but with absolutely no effect.
Carol might be right... if these xml realy describe whether a mesh is a proxy or not it would mean we just might have to edit the one xml of the main mesh
I already tested it a little..few days ago but didnt see an effect.
But i ll dig deeper into this
Last edited:
As soon as the big export is done I'll happily take a look into it (should finish up later today) but I asked mainly because you were working on it and was able to import
A problem though. I'm quite literally running out of space on my computer (I use a single 250GB ssd) and I'm not sure I'm be able to import anything without more space. I've already had to delete the entire characters folder (from export) in order to make room for the architecture.
If anyone else is planning on doing this, realize it takes over a full day and will eat 150+ GBs of storage.
I believe I have a hard drive (that isn't broken, like the last one) that I can use. I'll put it in my system once the export wraps up; I think I have enough empty space now that I removed the characters folder.
Honestly, if it comes down to resource management and trees have to be cut down , so be it. There's way too many trees in the game anyways just like there's too much fog. Look at this screenshots below. I think the first one has a good balance of trees while the second one has trees way overblown in the distance. (2nd pic was taken by Essenthy and his STLM mod)
http://giant.gfycat.com/ApprehensiveCompetentIchneumonfly.gif
http://i.imgur.com/hOF9yaa.jpg
, so be it. There's way too many trees in the game anyways just like there's too much fog. Look at this screenshots below. I think the first one has a good balance of trees while the second one has trees way overblown in the distance. (2nd pic was taken by Essenthy and his STLM mod)http://giant.gfycat.com/ApprehensiveCompetentIchneumonfly.gif
http://i.imgur.com/hOF9yaa.jpg
Indeed, summed it up quite nicely as I was going to reply a similar thing to @erxv
So yeah erxv you're mostly correct, basically it'll come down to the Engine > API > GPU interaction and just how far you can push that. VRAM won't so much be an issue, but just how much of a load the Engine/API can take before it implodes, or so to speak.
That would almost definitely require a proper world/environment editor, like a Redkit, and even then you couldn't be sure as to what positive engine related effect it might have (It'd likely be insane trial and error).
Too much blue fog?
Not for me^^
Anyways Carol its realy sad that cdpr just released such a poor modkit...
If it wasnt like that you would have been finished in an hour or so.
Not for me^^
Anyways Carol its realy sad that cdpr just released such a poor modkit...
If it wasnt like that you would have been finished in an hour or so.
Last edited:
I love the thick trees. It makes woods actually feel like woods. if you look at that shot, you can see through what little copses of trees there are. those wouldn't feel like woods.Honestly, if it comes down to resource management and trees have to be cut down
http://giant.gfycat.com/ApprehensiveCompetentIchneumonfly.gif
http://i.imgur.com/hOF9yaa.jpg
---------- Updated at 11:42 AM ----------
I'm praying that we won't run into issues with crashing. I know that some engines (cryengine 3 is my best example) deal with overloads on polygons really well - it just CRAWLS. (like a couple seconds per frame crawls) at least if it reacts like that it would be easier to diagnose problems - we could see which areas would be hardest hit by the mod. if it crashes outright it would be more difficult to test.
I have a pretty beefy rig (3570K and 290X) but I'd like to enlist someone with at least a titan X (preferably 2 or 3) to help do testing. Vigilance, you mentioned that you've got 12GBs of vram, would you mind helping once we've gotten to a point where we need to do performance tests?
---------- Updated at 11:47 AM ----------
I mean, if MeshRenderingDistanceScale and the LOD value in user.settings worked we wouldn't have had this issue...
if CDPR were to just unlock that value we could all move on to other mods (I've got stuff cooking in the back of my mind, but It's all scripting and I need some time to get used to the syntax)
The performanceimpact should not be that hard.
If i remember right you can edit the lod and hidedistance of the proxymesh freely!
So we can switch to low poly again when the object is realy out if sight
(But need to test again maybe my memory is wrong^^)
If i remember right you can edit the lod and hidedistance of the proxymesh freely!
So we can switch to low poly again when the object is realy out if sight
(But need to test again maybe my memory is wrong^^)
Last edited:
if you can't test today I'll get to it after the export (finally) finishes.The performanceimpact should not be that hard.
If i remember right you can edit the lod of the proxymesh freely!
So we can switch to low poly again when the object is realy out if sight
(But need to test again maybe my memory is wrong^^)
So if i am right put in xml:
Lod0=10
Lod1=200
The mesh should then unnoticably switch to lod 1 after 200 meters
Lod0=10
Lod1=200
The mesh should then unnoticably switch to lod 1 after 200 meters
Absolutely. I'm always the first on-board with trying all these graphical related mods, especially when it comes to something so egregious like the LOD/Pop-In issues. And testing is the least I can do, considering I sadly don't have the prowess to actually create these things.
I've got a metric shit ton of Hard Drive space, and 2 Titan X's, so more than happy to oblige.
for the proxy mesh? I believe so...
ultimately I'd rather make all meshes technically "proxy" and then just increase ranges and LOD ranges as necessary. If we can do that I can just write a script to change all the values for every xml file within a directory.
Awesome! That'll be perfect for testing more ridiculous values. Of course, at a certain point it's more for posterity that any realistic need to increase the view distance that far, but It's kinda future-proofing the game for when resolutions scale up large enough that we'll need to have detail at that range.
Tell me about it. I play the game @ 4K and the extra clarity of the Image Quality on the larger resolution/screen only serves to make those LOD issues even more insanely severe, as it's near impossible to miss the big pops.
Anyway, best of luck to you guys working on this, and I'll certainly be here to help test.
No, I'd assume he's just an adult with expendable income.
Once I'm out of Uni those kind of purchases will be more reasonable
So i tested just editing the xml file of th main mesh to entityProxy="true"
but had no effect at all...
seems like something else tells the engine what is proxymesh and what is main mesh
but had no effect at all...
seems like something else tells the engine what is proxymesh and what is main mesh
so setting it to true didn't enable longer maximum distances? well shit, I was hoping that would work.So i tested just editing the xml file of th main mesh to entityProxy="true"
but had no effect at all...
seems like something else tells the engine what is proxymesh and what is main mesh