Threaded Mode | Linear Mode

Razorwing · 03-24-2009, 01:06 AM,

I disabled my local BSA archives, double-checked that I had the March 18th versions of the glow maps, drag-dropped a STIDCandleBlackBowlRed01Orange128 light object in the CS. It crashed. So it's safe to assume the crash bug is still there and was likely caused by unknowingly using a faulty version of Nifskope. There's been a few versions of the program that corrupt nifs that way and it's happened to me a few times as well.

I didn't test the static versions of the candles. I don't recall anyone mentioning they were bugged, so they might very well work fine. Smile

But what should we do about the lights?

TheImperialDragon · 03-24-2009, 02:42 AM,

Quote:Originally posted by SACarrow
TID: I seem to remember that, like me, you were one of the last holdouts against using CS v1.2. Could it be that these were not tested within that version? I think the candles work in CS v1.0 from what I've read.

Steve

I believe Razorwing might have been using the 1.2 version of the CS when he tested the candles before (as he was also using it to find errors with other models), but I could be mistaken.

Quote:Originally posted by Razorwing
was likely caused by unknowingly using a faulty version of Nifskope.

I don't think so. One of the things I tried before was completely redoing the candle NIF set from scratch using a different version of NIFSkope. It didn't work. Confused

hrug:

Could someone else who experiences the crashing try replacing a black candle NIF (Meshes/ST/lights/candleblackfat01.nif) with a Oblivion candle NIF (Meshes/lights/candlefat01.nif, from the Vanilla BSA)? Just copy the Oblivion candle to the desktop, rename it so that it replaces the black candle NIF, and then put it into the appropriate folder. Since it's the exact same as a regular Oblivion candle (except for the filename), one might expect it to work just fine in the CS... however that did not happen in my case - the CS crashed. :eek:

Razorwing · 03-24-2009, 02:57 AM,

Heh, you're right! :eek:

I did as you suggested, renamed a copy of the Oblivion nif and replaced your nif. The I drag-dropped a STIDCandleBlackFat01Orange512 light object into the world window and BAM, the CS crashed. This is interesting! So... the problem is with the Construction Set entries themselves? If so that could certainly explain why it works fine in CS v1 but not in CS v1.2.

TheImperialDragon · 03-24-2009, 03:03 AM,

Either a problem with the entries, or the CS itself. :confused: I can't check right now... does it look as though there's anything weird with how the candles are set up? At least in comparison to some of the Vanilla candles?

Razorwing · 03-24-2009, 03:17 AM,

Quote:Originally posted by TheImperialDragon
Either a problem with the entries, or the CS itself. :confused: I can't check right now... does it look as though there's anything weird with how the candles are set up? At least in comparison to some of the Vanilla candles?

You're a far better modder than me, so I don't think it's my place to say. But as far as I can tell, there isn't. I've barely scratched the surface of the settings for light objects, not much more than getting my models to emanate light. I guess it's not impossible that CS v1.0 saves some unimportant information (possibly garbage) that got filtered out in the code for v1.2.

***sandor*** · 03-24-2009, 03:18 PM,

Quote:Originally posted by Razorwing
Heh, you're right! :eek:

I did as you suggested, renamed a copy of the Oblivion nif and replaced your nif. The I drag-dropped a STIDCandleBlackFat01Orange512 light object into the world window and BAM, the CS crashed. This is interesting!

I decided to start an investigation (CS 1.20404) because this problem is quite weird since the candles work in v1.0.

A lot of subrecords in this case for the LIGHT record have a poor description, so I decided to check the LIGHT records for TID's candles and compared these to the OB counter parts.

For this type of lights a "bound radius" (MODEL subrecord) is required.

I changed the bound radius for a number of TID's lights and the CS stopped crashing (screenhots). Smile

If you rename a light without a MODEL subrecord like an AR type the CS will crash (assuming the new entry requires the bound radius).

If you create a new entry and use for example a "candelfat01.nif" the CS will crash because it requires the bound radius setting --> It's an OB model so it's not the nif that causes the crash.

If you rename e.g. "CandleFat02Yellow256" (which initially used the "candlefat02.nif, using a bound radius) the bound radius will be transferred (screenshot, changed to "candlefat01.nif") --> No crash.
Renaming an entry with a bound radius will transfer the bound radius to the new entry.

How did Beth add a bound radius to the lights?
Perhaps they removed this functionality from the CS?

SACarrow · 03-24-2009, 03:31 PM,

It must be a newer version of TES4Edit; mine still lists MODB as unknown. I've been looking at that as well and discovered that setting MODB to something allowed the CS to accept the object. That settles that then: MODB for LIGH is definitely a float. I'll change that in Gecko.

***sandor*** · 03-24-2009, 03:35 PM,

Quote:Originally posted by SACarrow
It must be a newer version of TES4Edit; mine still lists MODB as unknown. I've been looking at that as well and discovered that setting MODB to something allowed the CS to accept the object. That settles that then: MODB for LIGH is definitely a float. I'll change that in Gecko.

I am currently using v2.13 Beta (released because of FOTES4Edit which I used before that), I can send you a copy. Smile

EDIT
I attached it.

SACarrow · 03-24-2009, 04:15 PM,

It seems that if a light has a NIF (MODL), it must have a non-zero bound radius (MODB). I've enclosed a hacked version of the light section of the Gecko form ID report for ST with MODL & MODB info.

HTH,
Steve

***sandor*** · 03-24-2009, 04:58 PM,

Quote:Originally posted by SACarrow
It seems that if a light has a NIF (MODL), it must have a non-zero bound radius (MODB). I've enclosed a hacked version of the light section of the Gecko form ID report for ST with MODL & MODB info.

HTH,
Steve

Thanks, it's confirmed by my tests.

I am currently checking the relation between the radius (CS) and the MODB.

Do you think MODT (texture file hashes) has any effect?

blue[/HR]
Could someone test the lights from the attached text file using the CS v1.2?

These are lights from TID and have a bound radius >0 (merged a long time ago).

On my end the CS didn't crash.

When confirmed I'll change the bound radius settings. :check:

TIA.