Fallout 4 Optimization and Performance Systems Explained

[NeinGaming]

finding the culprit by a process of elimination

Disable one half, if it doesn't crash, you know the culprit is in the half you disabled, otherwise in the half you didn't disable. Repeat the process with the half you know contains the culprit, reducing the number of candidates by half each time.

 

There is a proper term for this which escapes me, binary tree something maybe, but at any rate, it's a huuuuuge time saver in cases where you can apply it.

[84Cronos]

 

finding the culprit by a process of elimination

Disable one half, if it doesn't crash, you know the culprit is in the half you disabled, otherwise in the half you didn't disable. Repeat the process with the half you know contains the culprit, reducing the number of candidates by half each time.

 

There is a proper term for this which escapes me, binary tree something maybe, but at any rate, it's a huuuuuge time saver in cases where you can apply it.

 

I guess I'm searching for a lion in the desert.

[NeinGaming]

if you have 128 things to check, it's just 7 steps (2^7), if you have 1024 things to check, it's 10 steps (2^10). With this technique, the dune buggy gets faster the bigger the desert you have to search becomes :)

[84Cronos]

Little interesting update: I picked the most suspicious custom NIF in my interior, replaced it with the same NIF without collision and the precombines generation was successful. It turns out the CK doesn't like it when you copy over combined collision from a SCOL to a normal static NIF. I would have never thought that precombines take collision into consideration. Or maybe something else was wrong with the NIF.

[VIitS]

Little interesting update: I picked the most suspicious custom NIF in my interior, replaced it with the same NIF without collision and the precombines generation was successful. It turns out the CK doesn't like it when you copy over combined collision from a SCOL to a normal static NIF. I would have never thought that precombines take collision into consideration. Or maybe something else was wrong with the NIF.

Yeah, using an SCOL nif with a STAT record is known to be problematic (can't remember if it causes any problems outside precombined generation). It should be fine placing the SCOL there and generating with that in place, so you still have the right collision.

 

 

And collision is taken into consideration in the most annoying way: a single CELLID_physics.nif file that contains most (sometimes all, sometimes just some of the precombineds*) of the collision data for all precombined REFRs in that cell. It makes precombineds just that little bit less modular (if collision was kept in the corresponding nif, there would be ways to mix and match potentially, though it would not be very user friendly and you would still need to regen previs, so not really a big loss that you can't do that, just annoying).

 

*No idea what makes it leave collision in a precombined nif rather than merging it into the physics.nif file, but it doesn't seem to be that common (only a few of the vanilla precombined nifs I have looked at had the collision data in them)

[greekrage]

Little interesting update: I picked the most suspicious custom NIF in my interior, replaced it with the same NIF without collision and the precombines generation was successful. It turns out the CK doesn't like it when you copy over combined collision from a SCOL to a normal static NIF. I would have never thought that precombines take collision into consideration. Or maybe something else was wrong with the NIF.

yeas this is a big problem ive faced but not generating precombines only.... Ive had issues even where i needed to use collision from a similar object and the CK kept crashing afterwards... . Also seen the problem with embedded collision data when trying to edit a model in 3ds and then exporting it...

Its like devs used different methodology in some cases...just to annoy us :tongue:

A prime example of collision madness is the light house where even though the spiral stairs are a separate shape the collision is embedded in the outer wall part of the nif....

Edited April 16, 2023 by greekrage

[SoulesFury]

ive mad my first real mod and uploaded it here on nexus No More sanctuary hammering at Fallout 4 Nexus - Mods and community (nexusmods.com) what ive done is removed the 7 markers in sanc, where npcs hammer, some one pointed out that i maybe need to look in to the precomb i am currently watching this video and working my way through the steps great video here(4) Fallout 4 Creation Kit Precombine & Previs Tutorial - YouTube i am very new but love doing this . is this needed though for removing the markers and not actual rendered objects, is it a concern? i have no ill will about starting anything over and making sure its right i think thats how i learn lol. any input would be appreciated thanks

[niston]

You can convert a SCOL to a "regular" nif with niskope.

Convert the BSFadeNode into a NiNode and set all node flags to 14.

[SoulesFury]

i am working on fallout 4 mainly the sanctuary bridge im replacing it and ive gotten everything done however theres a spot on the side where it seems i am colliding with somthing invisible i open collision geomety and it shows a red outline where somthing is but isnt its like one of those small things thats not realyl small and i cant finish it out till ive gotten it 100 percent any clue as to what i prob may be? 



i know this doesnt talk about collision but i got here by watching trosskis videos and ive gotten the pre comb vis down pretty well but i just cant figure out this i tried to post a pic on here not sure but i can provide a video both from ck and ingame of whats going on 

 

Edited August 11, 2023 by SoulesFury

[SoulesFury]

i dont know why my last post there looks like that so ive surmised now that what it is is the prewarnorthbridge collision thats whats not matching up in the one spot how can i maybe use the collison info from the broken northbridge and apply it to the prewarnorthbridge? im so close and it looks great i just cant quite figure out this collison