openmw.org

I want to edit the Better_Sounds.esp plugin -- attached below as compressed zip -- for compatibility reasons. Specifically, I wanted to revert the 2 edits to magical effects, and while I am there fix the 3 errors spotted by the verifier. Here is the problem: as soon as I do these changes, and then load it for testing, all the inns in Balmora suddenly become very dark (the mod adds various sounds, including to inns). Running through tes3cmd, it now complains that the previously cleaned esp now has junk cells, duplicated instances and redundant cell weights And if I do go ahead and clean them with tes3cmd nothing changes, the inns are still very dark. What the heck is going on, that completely unrelated changes have these kind of effects?

note(s):
- running openmw nightly, a relatively recent compilation made a few days ago, in arch linux. Any more info needed, just ask.

Edit:
- attached the edited omwaddon with just the 2 reverts to magical effects Also started a new game to make sure it was not some weird savegame interaction, but no, with the edited omwaddon the inns are much darker.

It's not something supernatural. ESM reader/writer doesn't work well with plugins that lack cell lighting subrecords (i.e. cleaned plugins), in resaved plugins the subrecords are resaved anyway but with zero values (black colors).

The problem is, there's no information in the plugin as is that could allow the esm reader to reconstruct the cell lighting.

Capostrophic wrote: ↑31 Jan 2020, 01:59 It's not something supernatural. ESM reader/writer doesn't work well with plugins that lack cell lighting subrecords (i.e. cleaned plugins), in resaved plugins the subrecords are resaved anyway but with zero values (black colors).

That makes sense, thanks for the info. Grabbed the original, un-cleaned version of the mod, made the edits and then cleaned (there is more junk than just lighting information) and thinks seem to work out fine.

What happens if the same kind of edit is made in the vanilla CS? Is this a bug/missing feature, or a computationally intractable problem people are just going to have to deal with?

(full disclosure - not a dev and haven't modded in a while)

I *believe* that the behavior in vanilla morrowind is to default to an interior lighting setting when the interior ambient lighting setting is not found or else may simply ignore ambient light settings when more than one found. There is evidence that more than one found causes issues/bugs due to this being something that the modding community has deemed to require cleaning from mods. I say all of that lengthy explanation because the link here

above link wrote: Another great thing cleaned by TESTool that alone should make it a must for any modder, is that it strips from your mod unnecessary changes to interior cells ambient lights wrongly added by TESCS.

and this link here

other link wrote:This removes the so-called 'evil' GMST entries, duplicate object instances, duplicate records, redundant ambient light & water height records, and junk cells created by a CS bug. The mod is now considered safe for use.

Now, assuming the behavior in vanilla morrowind is that it runs fine/does NOT have extremely dark interiors, then there may need to be some method for openmw to detect a missing (as in, user cleaned) interior ambient lighting setting, and/or check if multiple settings for this have been found, and default to something other than 0 or whatever value is extremely dark.

I realize that there is a lot of assumptions in the above theory, but I hope that this helps to determine if there is a bug in openmw.

ST the assumer

Ok, I just tested this. Loaded the cleaned esp attached above, opened into the TES-CS and went into the magical effects tab. Could not find how to revert edited records, so I just made changes to the fire damage effect (which I *think* are functionally equivalent to reverting), saved, inserted this copy into the load order. Balmora inns (e.g. the Council Club or that other one just in front of the Silt Strider) are just fine and dandy, not darker -- this should be reproducible by the way, and if it is not, then the problem is on my end. Somewhere.

Edit: opened an issue on gitlab for this.