GoldenEye 007 Nintendo 64 Community, GoldenEye X, Nintendo 64 Games Discussion

[MultiplayerX]

I can bet a guess on the blood question. In the 90's censorship on games was much stricter for one thing. Two: Parents (including mine) were very paranoid about violence in television and gaming UNLIKE today. Our culture has just about made a 180 in media and rating systems. Hell you can show people nude on tv now as long as you embed a MATURE logo at the top left of the screen. Nintendo was most likely concerned about being able to market GoldenEye to the younger audience SO smoke and sparks are almost as cool as blood AND are less realistically violent.
_________________
http://codelegends.proboards.com/

[Dragonsbrethren]

zoinkity wrote:

I might add they later stated Citadel was a very early test stage[...]

Actually, their description was even more accurate than that:

Quote:

Q: What's all this about a hidden 'Citadel' level? A: 'Citadel' was a very rough test level designed during the early stages of multiplayer mode. It's not in the finished game in any shape or form, and Oddjob and Mayday wouldn't be in it if it was.

Its layout confirms it was designed for multiplayer, too. So yeah, outside of the part about it not being in the finished game, this is 100% accurate information, and it was available on their site as early as 1998. You already explained how they might've believed themselves about it not being in the game, but I interpret that differently: I just assume they meant there was no way of unlocking it as a playable stage under normal means.

Since I brought it up, here's their accurate response about Line Mode:

Quote:

Q: Is the 24th cheat a line mode/pen and ink mode? A: No. The cheats were devised to try and make the game more fun in both the single and multi player games. How a line mode would have done this eluded us, so we kept it out. It might possibly be activated using a GameShark or similar device, and though it is a useful debugging tool, as a gameplay feature it is a bit lacking.

I don't feel like digging it up, but I know they had another explanation that just called it a glitch caused by GameShark, not anything official.

[PacmanPlush]

Dragonsbrethren wrote:

Quote: Q: What's all this about a hidden 'Citadel' level? A: 'Citadel' was a very rough test level designed during the early stages of multiplayer mode. It's not in the finished game in any shape or form, and Oddjob and Mayday wouldn't be in it if it was. Its layout confirms it was designed for multiplayer, too. So yeah, outside of the part about it not being in the finished game, this is 100% accurate information, and it was available on their site as early as 1998. You already explained how they might've believed themselves about it not being in the game, but I interpret that differently: I just assume they meant there was no way of unlocking it as a playable stage under normal means.

The state that Citadel is in, and what we know about the development of GE's multiplayer mode makes me wonder about the actual timeline of the game's production.

Everything we've been told suggests that the multiplayer mode was done close to the end of the game's development. Martin Hollis calls the multiplayer mode an "afterthought" ( http://www.zoonami.com/briefing/2004-09-02.php ). Grant Kirkhope recently claimed that it was done "in the last six weeks" (see this video, at the 25:35 mark: http://www.youtube.com/watch?v=t70l-9n1rCQ ). Steve Ellis claims it was started in "March or April of 1997" ( http://www.joystiq.com/2012/08/14/goldeneye-007s-multiplayer-was-added-last-minute-unknown-to-ra/ ). Most places say GoldenEye was released August 23rd in Japan, and on the 25th in North America and Europe. So, we're looking at a 5-6 month period between the start of multiplayer development and GE's release. However, the game would actually have been finished a month or two (probably closer to two) prior to August, since final code would have to have been run by Nintendo, the ESRB, PEGI, and so on. Additionally, it takes time to manufacture the cartridges. If development on multiplayer started on the first day in March, it could have been worked on for 4-5 months, at most.

We know that Citadel was intended for multiplayer--Rare stated this directly, and as Dragonsbrethren notes, the layout is good confirmation. We can go further: Citadel's name is located in memory alongside the rest of the multiplayer level names. Furthermore, it exists in two forms: "CITADEL" and "Citadel". The all-caps version is intended for the level select screen, and the lowercase version is meant for the multiplayer setup screen.

Look again at Rare's statement: 'Citadel' was a very rough test level designed during the early stages of multiplayer mode. Why bother creating a new map to test multiplayer, when many of the single player missions already existed in a near-final form when development on multiplayer is said to have started? Any one of them could easily have been adapted for testing purposes. Proof is found in IGN's screenshot archive: http://ca.ign.com/images/games/goldeneye-n64-1991 This is by no means a comprehensive collection of pre-release photos--for instance, Silo, one of, if not the very first levels to be designed, is not pictured here. I link to it only because every image is clearly dated. From this gallery, we can see that Facility, Frigate, Statue, Archives, and Caverns definitely existed prior to March of '97, though they are in varying states of completeness. We can also see that multiplayer was up and running by June 12 of that year.

As zoinkity notes:

zoinkity wrote:

I might add they later stated Citadel was a very early test stage, which from the completely foreign and highly-inefficient clipping file system used in it can be assumed true. Seriously, there's parts of the path networking and lookup systems that are incompatible with the linkage method. Besides, it was obvious the stage was not compiled with the final output but an older version was caught by the linker and mushed in, with all the old image indices and none of the other stage-specific settings set within other final-version files.

It seems a safe assumption that Citadel would have been created the same way as the rest of the game's levels. So, if Citadel truly was created as a multiplayer test map, it could only have been designed 4-5 months before the game's release. From this, we can assume that the rest of the game's levels would have been using the same "completely foreign and highly-inefficient clipping file system." Does it make sense that the designers would have changed the way in which the levels were coded just a few months before the game's release? I don't quite know how vast the differences are between Citadel's data files and those of the other levels, but it seems odd to me. To be safe, even slightly restructuring the way levels are coded would probably require weeks of testing to ensure no new bugs were introduced. Don't get me wrong, with a dedicated team, a lot can be done in just 4-5 months. What they did was risky, but the GE team clearly were not opposed to taking risks.

I can't remember where I read this, but someone once suggested that each of Citadel's four main rooms seemed to have been designed to test a different aspect of the game's engine--ramps, corners, collision, and so on. This always seemed to be a likely conclusion. Although Citadel's layout lends itself well to a multiplayer map, the actual content seems like a rough means of testing the game's movement and collision code. To borrow Zoinkity's words again, Citadel seems for all the world like a "very early test stage"--not something created 4 or 5 months before the game's release.

I don't quite know where I'm going with this, so I'll reiterate my main points:
-Could the developers really have revamped the way in which maps were coded 4-5 months before the game's release?
-Why spend time creating a new map (albeit an incredibly rough one) to test multiplayer, when many levels had already been designed for the single-player mode and could have been easily and quickly converted?
-Citadel seems more like an overall engine test, and less like a multiplayer-specific test map.

About the only argument against these last two points I can think of is that Rare would have had to create and test animations (and possibly movement code?) specifically for multiplayer. Think about it--in a single player FPS, you don't have to create animations for the main character walking around, aiming, switching weapons, and so on. In multiplayer, players need to see their opponents doing such things. These animations would have to be done largely from scratch--can't just borrow the guard's animations. Still, does this necessitate creating a whole new map?

There are five conclusions I can think of here:
1. Citadel was designed 4-5 months before the game shipped as a means of testing the newly-created multiplayer mode (despite the fact that they could have adapted a single player level) and all levels used its style of coding at that point.

2. Citadel was designed to test multiplayer, but development on multiplayer began much earlier than the 4-5 months before release that we have been told--hence, why Citadel's format differs from that of the final maps.

3. Citadel isn't a test map, but a rough and abandoned concept for a multiplayer level. This explains its abstract design, which bears some similarities to Complex.
3A. It was designed 4-5 months before the game shipped and all levels used its style of coding at that point.
3B. Development of Citadel and multiplayer were both started earlier than we have been told.

4. Citadel was created early on as a dedicated test map, like those featured in Banjo-Kazooie ( http://www.youtube.com/watch?v=C0XOlcOW9J8 ) or Ocarina of Time ( http://www.youtube.com/watch?v=3Hb72KsGks8 ). It was updated throughout development, originally used to test single player and then, later, to test multiplayer. Ultimately it was removed (as Rare stated) but an earlier version of the map was mistakenly left behind.

5. Rare intentionally used the multiplayer mode as a means of showing off abandoned material. Think of unused characters like the Terrorist, Biker, and Helicopter Pilot. (Also, the last ditch attempt to include the old Bond actors as multiplayer characters!) Is it possible that the push-button code that temporarily unlocks the extra multiplayer characters (including the aforementioned characters that were intended for single player, but did not make the cut) was planned to unlock Citadel, as well? (In this scenario, I am assuming Citadel to be an early engine test map--not something specifically created for multiplayer.) Perhaps Rare got as far as coding its name for multiplayer use, then realized they lacked the time needed to create a new clipping file, set up spawn points, weapon locations, and so on.
_________________

[SubDrag]

Those are really good points pacman. Citadel is such a different format, it feels like it was actually the first level created. It combined the setup and bgfile formats, used floats instead, etc. There's no way in a multi mode created in one month that would've happened, had to be from way way earlier.
It looks like it tests intersections of rooms (using the pillars) and whether they show or not, perhaps clipping as well.

I suspect it was the first level ever made as a test, considered as a multi level since they were doing it ultra fast (yet it's also stage id 0x28...late in there, though Bunker was main one), but maybe not up to snuff or too hard to convert, so abandoned?

I also think, that while they say multiplayer was added in a month, I suspect there were 5 months or so until release (delay?) whereby they solved the hundreds of bugs created cause of it.

[PacmanPlush]

SubDrag wrote:

I suspect it was the first level ever made as a test, considered as a multi level since they were doing it ultra fast (yet it's also stage id 0x28...late in there, though Bunker was main one), but maybe not up to snuff or too hard to convert, so abandoned?

Agreed on all points. Citadel has to be the first level made--perhaps work was started on Bunker and Silo when Rare only had the SGI workstations as a reference, but Citadel could really only have been used as a means of testing various aspects of the game's engine.

Good call bringing up the Stage IDs, the rabbit hole deepens. To save time for anyone reading this, the internal level order, with the corresponding level IDs (assuming I'm understanding the Setup Editor, haha) are as follows:
0x09 - Bunker 1
0x14 - Silo
0x16 - Statue Park
0x17 - Control
0x18 - Archives
0x19 - Train
0x1A - Frigate
0x1B - Bunker 2
0x1C - Aztec
0x1D - Streets
0x1E - Depot
0x1F - Complex
0x20 - Eqypt
0x21 - Dam
0x22 - Facility
0x23 - Runway
0x24 - Surface 1
0x25 - Jungle
0x26 - Temple
0x27 - Caverns
0x28 - Citadel
0x29 - Cradle
(unused slot)
0x2B - Surface 2
(unused slot)
0x2D - Basement
0x2E - Stack
(unused slot)
0x30 - Library
(unused slot)
0x32 - Caves
(3 unused slots)
0x36 - Cuba
(3 unused slots)

There are a great number of oddities found in this list--probably enough worth discussing that we could easily start a new thread about it. (And probably should, given how far off-topic we've strayed--four years ago this thread was about an unused face!) We know for a fact that the earlier stages in this list were seen more frequently in pre-release media. Furthermore, we can see the list get less polished as it progresses, with more and more gaps between valid levels, which suggests that it truly does reflect the order in which the levels were added.

Bunker 1 and Silo were definitely the first levels to receive a significant amount of work, as they (along with Archives) were among the first to be shown publicly--there is no shortage of pre-release video clips and screenshots of these three levels. Statue is well represented in screenshots, no video that I know of. Strangely, despite falling in between them, there is only one pre-release screenshot of Control--the one on the back of the box. (And, if I recall correctly, the in-game mission picture shows the glass in the control room was at one point untinted. So, technically we've seen two pre-release images of Control.)

This would seem to suggest that Citadel was (re-)added to the list toward the end of development--I'd wager a guess that it used to be much higher, and represents one of the missing numbers before Bunker 1. If Citadel truly was used as a multiplayer test level, it would have been put back in specifically for that purpose, and so we can date the beginning of multiplayer development to sometime before the addition of Surface 2 and Cradle. I don't know of any prototype screens of Surface 2, but some images of Cradle show Bond's blue cuff ( http://web.archive.org/web/20080602131307/http://goldeneye.detstar.com/beta/betacradle.asp ). If these images can be dated, and if we assume that Citadel is a reliable means of estimating when development on multiplayer started, then we'd be close to solving this mystery. If not for one little issue...

As we'd expect, four of the six multiplayer exclusive maps (namely, Basement, Stack, Library, and Caves) are located at the end of the list (ignoring Cuba), which would seem to confirm the developers' statements that multiplayer was an afterthought. However, Complex and Temple come much earlier than expected. Complex apparently predates Facility and Caverns, and Temple is listed right before Caverns! I'm singling these levels out, because, referring again to IGN's image archive ( http://ca.ign.com/images/games/goldeneye-n64-1991 ), Facility and Caverns were in the game as of February 4th, 1997. Multiplayer development was allegedly not started until March or April...

Several possibilities here:
-Complex and Temple are early concepts for single player missions. Though these designs were scrapped, enough work was done to justify their eventual inclusion. (If this is the case, my money's on Aztec for Complex and Temple for Egyptian. Further backing this up, Aztec appears before Complex, and Egypt appears before Temple... perhaps after work shifted to the new design, the old ones were polished up shortly thereafter and re-added to the list?)
-Complex and Temple were always meant to be multiplayer maps. If this is true, then development began on the multiplayer mode and its maps earlier than we have been lead to believe. Facility was in the game as of February 4th, 1997, yet Complex is three slots before it. For simplicity's sake, let's assume that Facility was added to the game on the same day that the pictures were posted. Obviously this is not the case, it would have been added prior to that, but it means that multiplayer was present at least one month before Steve Ellis has said he started work on it.
-two levels were removed, and Complex and Temple were placed in their slots. However, they were actually developed toward the end of the game's development.

This last point might seem to be the most logical, but then, wouldn't the developers prefer to keep the multiplayer levels together? The unused slots after Stack and Library could easily have been filled with Complex and Temple. It's also strange that Stack and Basement come before Library. (In fact, Stack and Basement are separated from Library by "lue", an invalid slot.) This is of course just speculation, but I always figured once Statue and Cradle didn't work out for multi, Library was split into two levels that would replace them. (In the order as seen in game, Library comes before Basement and Stack. After these two, the borrowed single player levels begin.)

That reminds me--is there any explanation for the missing IDs between Bunker and Silo, and Silo and Statue? Similarly, is there any reason that the ID list starts at 0x09? My assumption is that these missing IDs represent levels that were removed, but unlike the invalid slots that come later in the list, these ones were taken out more completely.

Something that just occurred to me--in his retrospective, Martin Hollis said that during the first year of development, Rare had access to the film sets used during GoldenEye's production. The Bunker set was almost certainly one of them. Could Bunker have been the first level created because it was used as a proof of concept to show Eon Productions and Nintendo? This would explain the painstaking level of detail put forth in the main room--the big screen, the hanging monitors, the reflective "eye" on the back wall, and so on.

On a related note, is the "empty Bunker" a separate setup file used on the same map? I notice in the Setup File list, UsetupsevbunkerZ appears twice. If this is the case, is it possible that this setup (now used when any invalid level ID is entered) was created specifically as a means of showing off the level to executives/shareholders without having to reveal any unfinished AI, glitchy objects, and so forth?
_________________


Last edited by PacmanPlush on Wed Jun 12, 2013 11:00 pm; edited 1 time in total

[mistamontiel]

PacmanPlush wrote:

I don't know of any prototype screens of Surface 2, but some images of Cradle show Bond's blue cuff ( http://web.archive.org/web/20080602131307/http://goldeneye.detstar.com/beta/betacradle.asp ).

Wow, sorry offtopic, but what's up with that GS code to have the chopper with Natalya in it on the Dam ? It causes the emulation to hang, PJ64 v1.6
_________________

[Rey]

@Plush---The second bunker setup. Is that not supposed to me a "beta Bunker"stage? Even though it looks the same and shows in the game.
_________________
www.youtube.com/gamerrey23

[zoinkity]

The 'debug bunker' you get when you use an invalid ID is Bunker I loaded without a setup file. By default you load at (0,0,0) and no text file is loaded.
So long as (0,0,0) is a valid coordinate (on top of clipping) you can load a stage without a setup file. There were some very, very old MP stage codes that did this.

Before AI was developed you'd have to do tests with multiple players. Believe it or not, that's remarkably easier than coding any sort of AI. Most likely a bit of this code was reused when MP was reintroduced.
They needed to show some sort of progress, and having two guys run around firing weapons would be very impressive. Mind you, the inventory was probably manually triggered.

The 4-5 month timeline doesn't match up. They've been very consistent with the 1 month timeline. Plus, there are a lot of bugs in MP, with each subsequent release (J and P) removing more of them. PAL's MP code is streamlined and isn't the total mess in E. MP specific settings and structures are all in an entirely different format. In general, PAL structs are smaller and arranged differently--skies are 1/3 the size.

The solo stages converted to MP all had background features removed from them--particularly sprites. Also, there's a stack of frustrations getting a few of these stages working in MP at all. Two existing MP setups weren't used due to stage blackout issues. The dedicated ones use vastly less memory.
_________________
(\_/) Beware
(O.o) ze
(> <) Hoppentruppen!

[MultiplayerX]

Something else that I believe they are either HIDING or were never asked ..... is the fact that the PICTURE values that are selectable deep in memory make me question that they had actually tested DAM, RUNWAY, FRIGATE, DEPOT, and STREETS as MULTIPLAYER possibilities. I'm curious if streets was one especially considering it's SOLO level is an extension of OBJECTS. If you deleted them you'd have a pretty nice little multiplayer level.
_________________
http://codelegends.proboards.com/

[Rey]

MultiplayerX wrote:

Something else that I believe they are either HIDING or were never asked ..... is the fact that the PICTURE values that are selectable deep in memory make me question that they had actually tested DAM, RUNWAY, FRIGATE, DEPOT, and STREETS as MULTIPLAYER possibilities. I'm curious if streets was one especially considering it's SOLO level is an extension of OBJECTS. If you deleted them you'd have a pretty nice little multiplayer level.

Yeah,I noticed that when I made my first Solo level, High Stakes Street's.
That was why I did a another part!!.
_________________
www.youtube.com/gamerrey23

[MultiplayerX]

cewl!!!
_________________
http://codelegends.proboards.com/

[PacmanPlush]

zoinkity wrote:

Before AI was developed you'd have to do tests with multiple players. Believe it or not, that's remarkably easier than coding any sort of AI. Most likely a bit of this code was reused when MP was reintroduced. They needed to show some sort of progress, and having two guys run around firing weapons would be very impressive. Mind you, the inventory was probably manually triggered.

Makes sense, especially given what we know about GoldenEye's early Virtua Cop style of gameplay. For a while, there was no actual AI--just a set of scripted movements for each enemy. Before and after this stage of development, using a proto-multiplayer mode would, as you say, have been quite valuable in testing and demonstrating the game.

zoinkity wrote:

The 4-5 month timeline doesn't match up. They've been very consistent with the 1 month timeline. Plus, there are a lot of bugs in MP, with each subsequent release (J and P) removing more of them. PAL's MP code is streamlined and isn't the total mess in E. MP specific settings and structures are all in an entirely different format. In general, PAL structs are smaller and arranged differently--skies are 1/3 the size.

As Martin Hollis noted in his GDC postmortem, GE's multiplayer was in fact "developed in one month." This means that Grant Kirkhope's six week estimate is not far off. The 4-5 months I'd been using represents the period between the start of development on the multiplayer mode, according to Steve Ellis, and what I assume to be the end of development--roughly placed at 1-2 months before the game was released at the end of August. At the very most, development and bugfixing of multiplayer occurred between March-July or April-July. I'm not sure how long manufacturing/content review would have taken, but I'd say work most likely stopped in late June/early July.

Using the IGN screenshots as a reference, All Bonds was still in the game prior to June 12, 1997. On June 27th, a picture of the final character select screen (post-dating the one seen in the game's manual) was posted, which indicates that the removal of All Bonds came very close to the end of the game's development. I'm basing this on nothing, but it seems a safe bet that there was no more than 1-2 weeks between the screenshots being taken and being posted by IGN.

I'd imagine that the "one month" milestone represents the time taken to produce a stable version of the multiplayer code--however, as you've pointed out, the NTSC-J and PAL versions of the game show that bugfixing continued after the NTSC-U version was sent off for certification.

zoinkity wrote:

The solo stages converted to MP all had background features removed from them--particularly sprites. Also, there's a stack of frustrations getting a few of these stages working in MP at all. Two existing MP setups weren't used due to stage blackout issues. The dedicated ones use vastly less memory.

What data files differentiate the solo version of these stages from the multiplayer version? Would I be correct in assuming the map itself is the same, and the only changes are made in the multiplayer-specific setup file? I know a separate setup file is used to define the multiplayer spawn points/weapon locations, and I assume it is responsible for locking out the backzones and removing the objects normally found within them, too. Would the sprites that get removed be part of the BG file, or are they controlled in the setup file as well? I'm assuming the latter, but I'm not too familiar with the way GE's stages are set up.
_________________

[zoinkity]

Don't ask me how I know this, but:

Random Source on the Internet wrote:

All software licensed by Nintendo must be submitted for approval testing by Nintendo. This approval testing process normally requires approximately 2 1/2 weeks for its completion. Individual testing will vary depending upon the software submitted and peripheral compatibility. However, the overall process includes a thorough testing of the software on all game platforms for which it is intended to be used. Game play is performed by various testing personnel in order to test all areas of game play. All software submissions to Nintendo of America Inc. must be forwarded to the attention of NOA Product Approval Manager, Engineering. Otherwise, the submission's placement into the testing queue may be delayed. To help reduce a submission's turn-around time, licensees must assign a primary contact person for software submissions. This person is responsible for dispersing information within your company. All communications with NOA concerning a submission's testing status should be forwarded through this individual. When a submission is approved, your company's primary contact will be notified immediately in writing. When a submission is not approved, NOA may send a videotaped copy of the programming problem(s) which prevents the submission from being approved. This is intended to assist the licensee in analyzing the cause of the software problem. It is the licensee's responsibility to send a copy of this tape to any developers of the software. NOA strongly encourages that copies be sent to the software developers as quickly as possible.

It was already mentioned they had to change a couple memory allocations to pass NTSC submission, so upwards of 5 weeks passed after initial submission. Plus, you can't count from time of production; you'd have to count from the deadline set by Rare themselves. After all, one of the submission fields is scheduled release date.
_________________
(\_/) Beware
(O.o) ze
(> <) Hoppentruppen!

[Dragonsbrethren]

Since we're speculating, on Temple falling between Jungle and Caverns: Temple's slot name is dish. Martin Hollis's post mortem lists Transmitter dish as part of the Arecebo (Cradle/Cuba) mission. While we'll never know if any background was developed for it, I bet the slot was reserved for that mission, and Temple was later stuck in there.

Interestingly, Citadel is cat, and Cradle's original name was evidently Transmitter Catwalk. Cradle's slot is crad and comes right after that, I don't know if there's any significance.

(The slot after Cradle is "sho" - Showdown, perhaps? Fighting Trevelyan was originally planned as its own mission.)

-Edit-

Here's the image of the old mission plan: http://www.joystiq.com/screenshots/goldeneye-007-postmortem-gdc-europe-2012-1#/10

Yeah, Severnaya was originally intended to be the first mission. Makes perfect sense that Bunker was the first background they did.

MultiplayerX wrote:

Something else that I believe they are either HIDING or were never asked ..... is the fact that the PICTURE values that are selectable deep in memory make me question that they had actually tested DAM, RUNWAY, FRIGATE, DEPOT, and STREETS as MULTIPLAYER possibilities. I'm curious if streets was one especially considering it's SOLO level is an extension of OBJECTS. If you deleted them you'd have a pretty nice little multiplayer level.

They're not hiding anything - they said they tested every background in multiplayer. The text file for the menu makes it clear even Aztec and Frigate (somehow) almost made the cut - I wonder if those two were stress tests for the engine, actually.

-Edit 2-

Also you might be making too much of file formats changing. Remember Rare didn't have any specific editor for this game, it was all done with converters and source files. Updating the converter to export the same geometry (assuming they did something sane like that for stan tiles) in a new format wouldn't have been that big of deal. We know other game engines changed like that over the course of development, there's no reason to doubt GE was any different.

I also have no reason to doubt that Citadel was developed specifically for multiplayer based on what evidence we have. Duncan Botwood and Karl Hilton cranked out Complex, Temple, Caves, Library, and Egyptian in that month of multiplayer development - guys were level design machines! Sub, when you interviewed Botwood, did you ever bring Citadel up? Curious if he could shed any light on it. (Ignoring the test elements, it's a pretty boring layout compared to the rest of his arenas.)

[zoinkity]

Technically assigning the name Citadel is speculation. There's nothing physically tying those strings to particular stage numbers. As you've already pointed out, the name snippets don't actually correspond to anything in final. They could even be completely useless, left in from some older demo.

MP setup files are looked up by extending solo file names by "mp_". Oddly, due to the way they're compiled they include the full set of all solo presets. Library, Basement, and Stack all have a complete set of their presets. Background files and clipping are shared. Other attached data (environment settings, memory allocations, etc.) are unique to any given combination of stage + #players.
Sprites in question only appear in beta photos. Bunker II and Archives had hanging lights. Facility's MP area does not have second layer warning stripes; this is only found after the locked doors, and are common throughout the rest of the stage. Caverns and Egyptian may be exceptions, or at least nothing is obvious. Blackouts and collision issues are common in Statue and Cradle, but to be fair they occur in Caverns and Egyptian on 4 player.
These would be contained in a given room's second BG file, along with things like railings, grates, warning stripes, and other non-essential elements. In general, anything that involves some kind of transparency is in there.

The old format didn't follow the rules. They could be in any which order, linking wasn't against faces but against whole surfaces, they did not have to make direct contact, and faces could extend over the edge of other faces. It's more akin to PD clipping than GE, and this causes serious problems with GE's path walker. Using float math modified by scale is also significantly slower than simple SLTs. Besides, there's obvious magic used to offset through tiles.
So yes, I contend they'd have to make significant modifications to the engine to use the older clipping.
_________________
(\_/) Beware
(O.o) ze
(> <) Hoppentruppen!