» Thu Jun 21, 2012 4:17 am
I'd like to reiterate (yet again..) that NOT all Vanilla areas are affected by the NavMesh Bug. My Overlook Tower mod works just fine; as have other test-mods I made to expressly prove this. That mod adds NEW navMesh to Vanilla cells already containing some; specifically the cells behind Drelas' Cottage (other side of the Labyrinthian mountain). It connects to custom interior cells (two diff Van cells connected to two diff custom interiorCells)... everything works flawlessly; followers go anywhere I intended them to with my custom navMesh (some of which is on the peak of a mountain, and are "island" navMeshes).
What I do (and I've explained this several times already, though in the other threads I believe)... basically I use the procedure that hypno88 posted in the last comment; except I use TESvSnip to manually change the file header to be an ESM (not use the CK to somehow trigger that change, too shady for me when it takes10sec to change it manually). TESvSnip saves proper header data, but if the record count isn't manually changed to be correct (after adding/deleting records)... the CK is what changes the version # back to .85 - NOT TESvSnip. I use Snip v4.2, so it may have been updated since then - possibly breaking itself or whatever.
Another difference is where it's stated to 'Snip' ONLY the navMesh into the ESP; I Snip over ANYTHING having to do with Vanilla, most importantly ALL navMesh data (NAVI & NAVM records), as well as ANYTHING placed/added to or changed in a Vanilla cell. But yes, I make a copy of the original plugin's ESP; then convert it to ESM and Snip out anything NON-Vanilla (and vice-versa for the ESP.. Snipping out whatever remains in the ESM). This technique enables me to move or change Player-storage (in subsequent versions of the mod's ESP) without Users having to remove items before updating... THIS is one reason why I use ESMs (other than NavMeshery drama).
The other reason is to prevent anything potentially legacy from interfering - as in Oblivion, ESMs couldn't CHANGE another ESM (though ADDING to one seemed/seems to be generally acceptable, but MAY backfire). I'm fairly certain everything that is TES and FO are ALL based on previous versions, whichever was most recent at the time. If they 'disabled' code in a previous version somehow but 'mistakenly' reinstated it in a later game, it could cause issues not unlike what we're experiencing... with SEVERAL aspects of Skyrim, not just navMesh. (mannequins and their poses.. AHEM..)
THIS may be where some people's mods fail while others' work. I KNOW that ONAM lists were implemented in FO3 to enable ESMs to alter other ESMs, but that may or may not be working as intended. All I know is the way I do it, works every time UNLESS.... trying it in certain cells (see the 'Riften/Whiterun Experiments' posts in this/previous threads). Supposedly this ESM-on-ESM action taboo doesn't exist anymore... and maybe what I do is overkill, but my technique works (again, NOT for ALL areas), unless tried in an 'Anathema' area. (ONE cell WILL affect all those connected to it when navMesh is involved)
Regarding the memory issue: It cannot be defined universally because everyone's memory is being used/managed differently, depending on operating system configuration, background stuff, INI settings, etc etc - a MILLION things can be affecting this enough so as to make it 'non-universal'. Someone with less mem is presumably more prone to these issues than those with more RAM than the ghost of Steve Jobs (..too soon?). Personally, I'm not convinced it's the actual memory... my work with intense real-time scripting showed me that vidMem is NEVER max'd out on my system (even in Ultra default it only uses 3/4 of my 1gig vidRAM); and sysMem never even comes CLOSE (unless I run background progs like a 'full' CK, 3DSMax, Photoshop, etc). What my testing showed me was that the GPU is what causes script-timing issues... and LITTLE else (except for OBVIOUS system stress, induced before Skyrim is even started)
This can be extrapolated to explain Amethyst's issues with shadow lights.. 1.5gig vidRam with a decent GPU; yet two or more shadow-casters in highly-cluttered areas causes drama. My contention is that it isn't the lights per se, but the shadows they are casting that cause the drama... more objects means more shadows having to be drawn - NOT more memory being used (though that actually IS the case, more shadows=more vidMem used... but it never max's out, so shouldn't be an issue). I'd be willing to wager that the same lights +1 would work just fine if the # of objects affected by them were reduced. My GPU is ALWAYS max'd when I have issues with real-time script.. but if I reduce detail to allow less-than-100% GPU usage, it seems to work fine (unless affected by other known issues).
[Arthmoor: I never played or modded Morrowind... perhaps the supposed Obliv limitation on # of objects per cell was heresay or carry-over from MW? Of course it's possible that I'm simply remembering disinformation, or even experiencing faulty memory reconsolidation. I believe Skyrim isn't affected by a hard-coded #, but I haven't actually tried counting any cells... heheh. I HAVE added about fifty actors to a single cell in Whiterun... and nothing out of the ordinary happened (even when they ALL have onCellAttach scripts and displaying equipment). I only suggested it as some people's issues seem to only happen when in highly 'detailed' areas. While it may not be hard-coded, some Users' GPUs/etc may not be able to handle X amount of whatever.]
[EDIT: I'm not entirely convinced that what Amethyst just said really is the case.. regarding having 'clean' ESM/ESPs. Personally, I believe my files are clean, but issues in the game-engine/CK may be interfering with certain functionality (eg- LOD for large statics). I see no evidence that they cause drama of any sort... so how is it 'dirty'?]