[map] Titanic

Models gotta be solid. Put a default one in. Check the properties and find out which boxes to check off. I walked all over the boats and stuff in HL2.

---

The best things in life, aren't things.

Yea turns out in HL2 Im getting these in the console:

SOLID_VPHYSICS static prop with no vphysics model!

Googling it..

Got it. Turned out I had to give the entity a name and then set that name in the Physics Mesh parameter.

So before the limited use of Propper I've done, my map had some 7456 brushes, and now with most of the staircases converted (Grand Staircases, and both Second Class staircases are not yet) the number of brushes has dropped to 6989. And the model limit per map is 1024 so I think this will help a lot.

I havent had a lot of time to work on this over the last week so the overall progress hasnt really advanced, but at least now Im back to a state where I can keep working.

One area which I've had to deviate from the blue prints are the spiral staircases which allow crew access from D Deck to the Tank Top, because if I base myself on the plans, its too small for the player collision hull. Im still in the process of making it big enough to permit the player to use them.

I've made another "quick" video of the new sections that have been added since the last Youtube last month. You can check it out here:

http://youtu.be/T4Z9mdRj3Yo

I actually talk in this one.. I got tired of making labels.

I haven't posted a screen shot in a while, so here's one from tonight. Over the last two days I've been able to make a bit of progress. After getting the whole errors and model stuff out of the way, I've completed the layout of the spiral crew staircases all the way to the Tank Top, and I've also enclosed all the Boiler Rooms and designed basic boilers.

The isn't a whole lot of information about how the boiler rooms looked like and how the escapes up the fidleys to the boat decks were laid out, so there's going to have to be some guess work there. But I cant really build the escapes, or the boiler casings, until I finish building F and G Decks, so Im going to focus on that now. Once those are one, I can do the boiler casings, and once that is done, I can do the fidleys.

So, here's tonight's shot: The brush count is back up to 7643, mostly because of the boilers, each double ended boiler having 22 brushes, and there are 24 double ended boilers, and 5 single ended boilers on the ship. Those will get modelled later. At 22 brushes a piece, they're by far not detailed yet.

Had the chance to work a lot today, added a lot of the layout on F Deck in the forward half of the ship. You'll note some of the locations added from the screenshot below: The swimming bath, the Turkish baths, First Class baggage, Third Class Dining Saloons, the larger individual boiler casings.

The Ash Ejectors and Coal Chutes are located on F Deck but there wont really be much to be built for these. A few of the ejectors are in rooms that are accessible and those will be built.

The current brush count is nearly 100% now at 8181 brushes, though once I model the boilers, that'll eliminate about 500+ brushes. I'll be making good use of Omegaslayer's compiler. We'll see how that goes. Though being already at over 8100 brushes, and none of the passenger rooms are built, I'll have to split the map at some point.

Or maybe Source 3 will come out and the performance restrictions will be raised hehe.

Here's the shot from tonight:

You Dillweed. You are in Florida and you are working on a map? You really need to look up the word "Sightseeing"

---

The best things in life, aren't things.

holy cow you've made some beautiful progress since last i was here! It's so cool to see you're still working on this

(is there a way to subscribe or "watch" for thread updates?)

I just skimmed the entire thread, and i must say, there is A LOT of good information and tips in here for building big maps in source! (More than a couple things i've read here i'm going to try immediately in my 747 map, like changing brushwork to displacemens.. very obvious but i woould have never thought of it to save on map planes and brush sides!)

A couple thoughts i have from the things i've read:

Scale
Hammer scale doesn't really mean anything imo, but i would agree with Haymaker you are gonna want to scale things up for the space to "feel" the right size, and then again even more to make some areas more playable. E.G., i started building a map of a WWII destoryer(the sullivans) to scale from blueprints, and in the end, i had to scale everything up(1.3 to 1.5 i think) to make the interiors of the map not feel like a small closet

VVIS running for 4 days
Ya this is in no way acceptable, not for you or your map! Change everything not sealing off the map to groups of func_details, propper prop_statics, and displancements. You can also try func_lod/propper lod, which are like prop_statics with a fade distance, except they are brushwork.

I understand the concept of visblocking, but in source, your compile will go on for eons and eons with too many worldbrushes in your map, MOST ESPECIALLY a map of this size.

max map planes/map map brushsides
A lot of times if you ever get this error, it dosn't mena you have literally run out of brushes, but that you have one or more really messed up brushes from a VMing error. (i know you have a lot of brushes in this map, but just food for thought from my experiences)

propper and prop_static models
Propper is amazing tool, and allows you to go way over the "brush limit" than you ordinarily would, but having too many models in your map will cause a major hit to fps, regardless of max/min view distance, so don't go crazy with models.

clipping
Clip out every non-enterable floor space to make the smallest "footprint" area hammer compile has to make calculations for. (i'm sure you already know this)

fog
Down the line when you have a playable demo, you might try this to increas your fps. I have yet to experiment with it personally.

Question:
Did you ever try the compiler the one bloke mentioned about increasing the hard limit on brushes in hammer? Really curious what the implications would be on performance!

Phew! sorry about the wall of text =P GOOD LUCK on your map and i look forward to future media! =P

I know I know.. (that was for Orph)

[edit]
Hey Captain Terror, nice post. You can "watch" a topic. If you do a reply there's a checkbox 'Watch this topic'. Not sure if you can subscribe another way.

I've found that using actual scales and measurements that there are some aspects - like you mentioned horizontally - that seem to feel a bit tight, but its hard to tell. Vertically though, the scale feels too large. But at the same time, the blueprints call for the heights from frame to frame, and then you add the floors and ceilings, which I havent really accounted for yet.

I've propper'ed staircases at this point, to avoid engine limitations, but once the basic layout is complete I do plan on modeling a lot more. Boilers are going to be a good candidate. They're by far not detailed, but they're 22 brushes each on the double boilers, and 24 of them, thats a lot of brushes. Im realizing though I'll definately have to cut up the ship into two or three maps for the final builds.

Im planning when the layout if finished, to paint all brushes with null textures and then go over and texture better. I think that BSP automatically nulls out any faces that arent visible, but Im not sure. I dont think it used to be that way.. (in HL1)

I'll check out the fog stuff, never played with that. My FPS has actually been surprisingly good, despite fast VVIS and VRAD compiles, but I know that will change.

I had tried the compiler that omegaslayer gave me but at the time I ran into errors. I havent tried it since, I should give it another shot soon though. Im at 8262 brushes right now :P

Haven't really tried so I dunno if its still working but, if you watch a topic, it should email you that something new is posted. Otherwise, it takes 3 to 6 seconds to come here and look. shrugs

---

The best things in life, aren't things.

ok thanks guys, i didnt see "watch this topic" in big yellow letters at the bottom, DOH! =)

Orph: yeah im not a regular, so having "watch" or "subscribe" function available jsut makes it really eazy for me to keep up on topics on various forums

Regarding fast vvis, i find there is no real savings in the long run, and normal vvis SAVES you on errors and catches things that fast won't. PLUS, your lighting will look entirley different from fast to normal vis. (for these reasons i never use it anymore) Just more food for thought, but it is VERY encouraging you have good fps at this stage, i must say1 =)

I null like a banshee but i'm beginning to think it's largely a waste of time, e.g., on my nuketown map i nulled like 2000 faces and saw not a bit of improvement to fps, but of course i still null wherever i can by principle because it's just cleaner that way.

Yea I want to try to get a full VVIS done soon because thus far the three times I've tried, its just gone on for hours and days. I dont have any lighting in yet though, not going to work on that until later.

i see you live in ontario whereabouts? (i'm just across the border in sunny buffalo ny =)

If you are inclined to release the vmf, i'd be willing to try on some compile optimization for you, but i FULLY understand if you're not interested

I just can't imagine waiting for days for vis.. i would never be willing to wait that long not ever. ='(

I might be interested in releasing a section of the VMF and seeing what you can come up with :P

Im in Ottawa, well, 'cept for this week. I think the biggest thing that'll need optimizations will be the exterior areas, the Boat Deck, A Deck, the outside areas of B Deck (Poop Deck at the rear and the Forecastle at the front), and the outside areas of C Deck (the fore and aft Well Decks).

Everything else is mostly small tight spaces, narrow corridors and such. There are a few areas that are more open though. Scotland Road on E Deck runs nearly the length of the ship straight out. On D Deck, the First Class Dining Saloon and Reception Rooms are large open spaces as well, and at the foot of the Fore Grand Staircase.

Another general overview of the ship as of now. At 8299 brushes. Most of the work on F Deck is now complete, and most of whats left is the Orlop and Tank Top decks. The Tank Top is the bottom most "deck" (actually the top of the double bottom). The Tank top mostly contains the propelling machinery: The engines, the boilers, the shaft tunnels. The Orlop Deck is split between refrigerated cargo, coal bunkers, and cargo.

[IMG]http://img815.imageshack.us/img815/7590/titanic48.png[/IMG]
[IMG]http://img195.imageshack.us/img195/8857/titanic49.png[/IMG]
[IMG]http://img259.imageshack.us/img259/2724/titanic50.png[/IMG]

Here's looking at the entire ship, at 8374 brushes total. When I get back home I will model the boilers, and the Grand Staircases and Second Class Staircases, that should eliminate some 500+ brushes.

You'll note the set of orange brushes between the 3rd and 4th funnels, at the Boat Deck level. This was ventilation for the reciprocating engine room.

At this point I think I'll focus on the Tank Top and Orlop Deck. Those are the only things really left to do. Mostly towards the rear of the ship. The Reciprocating and Turbine Engine Rooms, the Shaft Tunnels, and the various refrigerated cargo areas on the Orlop Deck above.

I also need to start finding some appropriate textures. If anyone is willing to help in that respect I'd appreciate.

WOW, this is very ambitious! I like it. Please do a CS:S compile of the map when complete so I can add it to my CSS server wild card!

Thanks G4mer,

Not sure if I'll make it available to CS:S, not sure why I'd want to, since its not going to be a CS type map. Im not really planning any kind of multiplayer play with this. Just mostly a walkthrough. I do want to get a full compile going though..

I've put in a bit of work onto the funnels now, the earlier "version" of them was more or less just a place holder for their actual dimensions and location. You can see the new ones in the shot below, as well as the fidley "grate" (will be grates later) which vented the boiler room's hot air outside. In front of the fidley grates (forward, or behind on the aft ones) were the air intakes to pump fresh air into the boiler rooms.

I've also started doing some work on the Tank Top in the electric engine room and the shaft tunnels, as well as the refrigerated cargo area on G Deck.

The total brush cound is now 8500 on the nose.

I've modelled the boilers and brought the brush count down to 8059 now, below the 8192 limit. I also detailed the foremast and crow's nest including the access within the hollow foremast. But now Im running into this guy:

Too many unique verts, max = 65536 (map has too much brush geometry)

Which as far as I can tell, simply means its time to cut up the map. Im betting the work to the foremast is to blame..

New profile shot, though nothing's changed since the last post. Havent had time to work on the ship. Changes include: modelled boilers, added crow's nest and hollowed foremast, G Deck at hatch #4 aft of Turbine Engine Room, shaped Electric Engine Room, new funnels and ladders.

Still running into the max verts error though.

I keep meaning to ask but forget. How close to scale is this boat?

---

The best things in life, aren't things.

full scale

I thought you were making the stacks from displacements?

Like G4m3er said they will be harder to make, but well worth the effort as they will be cheaper to render than models.

Also btw, that last hammer screen is really beautiful. I think you would benefit greatly from a multi-montior setup for this project, so you don't have to mouse-scroll-zoom for days and days to move from one end of the ship to the other

Since there are only 4 stacks and hard to make, wouldn't it be better to make all the decks and/or flat surfaces into displacements? At the very least, all the flat surfaces that no one would ever see? There has got to be sections/areas that nobody will be in.

Even if not, people have been walking on displacements in almost all maps so I see no reason to make a hard shape to save surfaces when you can more readily make easy shapes faster.

There has gotta be a bazillion flat areas crying to be displaced.

---

The best things in life, aren't things.

Im unfamiliar with displacements so I'll have to Google that. I didnt even think to model the funnels, I probably should. I'd like to get a multi monitor setup though.. Maybe some day..

There arent a whole lot of areas in the ship that arent player accessible right now. The few would be the boiler casings, fidleys, and stokehold vents. All of which will be player accessible eventually as the fidleys would have been used as emergency escapes for the firemen in the boiler rooms.

Think of HL2. All the dirt, rocks, and most anything else on the ground was prolly displacements.
They aren't real hard to make but they are hard to get to look right. That's why I suggested flat surfaces. You ain't gotta do much but plonk them down and make sure no one can see under them.

---

The best things in life, aren't things.

Dunno if you subscribe to Nat Geo, but

http://ngm.nationalgeographic.com/2012/04/table-of-contents

April issue.

That's a really good idea orph i never thought of using displacements like that

WC i'm quite a nub with displacements myself, and i'm really just starting to use them in my maps.

Thanks for the link haymaker. I'll have to buy that issue. I checked out the high res pics of the wreck.. Simply amazing.

I finally discovered visgrouping tonight, and its made viewing a lot easier. And also allows me to finally post a top-down view of each deck:

Just discovered vis grouping. You do know that we are online every night, right?

---

The best things in life, aren't things.

LOL ya, visgouping is the bomb. I'm so dependent on Hammer 4 visgrouping now, going back to do anything in gsource is really painful

Also great screens btw! (i need a dual-screen setup just to view half the ship properly =P

Actually, Im pretty sure visgrouping has been around since the HL1 days, but I only now started using it!

I kinda wish I could work expense this project lol. Already had to shell out 40 bucks for FRAPS and 100 for the books from Bruce.

Well, its time to start planning the splitting of the map into multiple maps. We'll see how this goes.

[edit]
Started the split working on the Boat Deck first. And will continue top-down. I will be splitting each deck into a separate level as best as possible I think.

With that, I've listed phase 1 as complete, and I've changed the scope of phase 2 to be simply the level splitting and basic optimization which will allow (hopefully) for full compiles of each.

Keep an unbroken version. I bet there are gobs of optimization you are unaware of.
Hell, you prolly haven't even tried the displacements yet.

---

The best things in life, aren't things.

Yea. I didnt do much. But the error Im getting seems to have only as solution to remove detail from the map. And given that none of the actual rooms are built yet, nor are any of the details put in (ceilings, railings, lighting, furniture, etc). I know I'll have to split it

I'd still look into optimization. There are some tuts here. Think of it this way. Some of the maps in HL2 were HUGE. They had way more detail than you currently have. If they can do it, so can you. WAY MORE THAN YOU.

I am betting you would be better served reading some tuts instead of figuring out where would be the best place to chop the map into bits without ruining the goal of your project.

---

The best things in life, aren't things.

That is very true. I spent maybe two hours or so working the Boat Deck and it is a bit of an annoyance to split the map..

But I need to fix that error..

I've tried turning a lot more into models with no effect, still getting the error. I've reduced the brush count by around 1000 or so and faces by a lot more.

I propper'ed the following:

• all funnels
• dome over aft grand staircase
• dome over fore grand staircase
• mainmast
• foremast
• railings at fore grand staircase at boat deck

I dont quite like modelling though as the models dont end up being the same dimensions as the original brushes. For instance, the funnels are 582w 305l 992h, however the model ends up being 582.5w 305.5l 992.5h which is a tad annoying as for some of the models, such as the railings, they dont fit perfectly like brushes did.

In any case, the error is still there. If I block out parts of the map, I can get the error to disappear, and then I get a Tjunction error: Too many t-junctions to fix up! (3048 prims, max 32768 :: 65541 indices, max 65536) which as best I can tell is caused by too many func_detail brushes touching world brushes.

Im well below the engine limits as far as I can tell... Now at 7707 brushes and 46799 faces.

I was able to get a normal compile (fast VVIS) but I had to block out all decks bellow D Deck (D, E, F, G, Orlop, Tank Top). Not exactly a solution.

[edit]
On a slightly positive note, I was able to get a full compile (full VVIS and full VRAD) with allow the decks bellow D blocked out, and FPS never dropped bellow 300 except for at the extreme ends of the maps where they flirted with 299. Not sure if thats accurate or not.. I couldnt get mat_wireframe 1 to work though.

What other sites have you posted this on?
I am sure there are solutions, you just need the right people looking AND commenting. Right now, the vast majority of Snarkpit users look and little else.

---

The best things in life, aren't things.

Posted this on VDC: http://forums.steampowered.com/forums/showthread.php?t=2617039

But looks like the answer there is (surprise surprise) to split the level..

The responses I've gotten from VDC have mostly been unanimous, so I'm going to continue working on splitting the levels. Im finishing up the Boat Deck now, going through a full compile.

I plan on having the following levels:

• Boat Deck
• A Deck
• B Deck
• C Deck
• D Deck
• E Deck
• F Deck and G Deck
• Orlop Deck and Tank Top

I think I should be pretty safe that way.

The upper decks will be the more difficult ones to make, especially the Boat Deck, A Deck, and B Deck, as each can see a lot of each other, along with the Forecastle and Poop Decks, although those will only be accessible on the B Deck level.

Its a bit of a pain.. but at least this is progress

To retain the vertical feel, you're gonna chop it up "up/down" right? Surely not horizontally.

---

The best things in life, aren't things.

To keep the vertical feel, Im keeping some areas of other decks. For instance, on the Boat Deck map, fore and aft, I've retained all the external structure of A, B, and C Deck. At the Grand Staircase, I've kept the surrounding layout for all decks.

hmm it would take a fair amount of work but have you thought of converting whole sections into models? I don't think there is a major restriction on model size in Source ( there is an entire coastline model in the HL2 GCF )... you could maintain your runtime engine efficiency by planning where to begin and end these models, and use nodraw / areaportal / render trigger brushwork accordingly. You mentioned that there are fractional differences in sizes sfter conversion, but this is easily hidden by any number of workarounds, props don't necessarily need to sit exactly flush to brushwork etc.

The major downside is going to be compile times under -staticproplighting and -staticproppolys, but you are in for a loooong wait regardless I'd say. As for textures, I'm not 100% sure, but can you not edit your model textures and have the changes show up on game reload? I know you can for brushwork, just applying lazy logic is all.

One of the problems with doing that is that models cannot seal the map. And practically all of my hallway brushes, especially once you get to the lower decks.. seal the map.

That slight size difference it also an annoyance.

I've got the Boat Deck map completed with the exception of putting in noclip brushes (to prevent the player from jumping to lower decks which arent in the level) and the changelevel entities. Im doing A Deck now, then I'll noclip and changelevel both levels and take them for a spin.

So far, the Boat Deck level has 2099 brushes and 12,990 faces. I've been able to do a full VVIS and VRAD compile, although there is no lighting yet.

I havent been able to get 'mat_wireframe 1' to work in game though. Am I missing something for that?

mat_wireframe 1,2,3 can only run under sv_cheats 0.

I was kind of thinking of portaling / sealing / visblocking / what have you by way of thin brushes inside the modelwork. This way you can probably divide each deck into quarter pieces and still have good framerates. If you're really on top of it you could aim for the 1024 auto-cut lines. Individual rooms need not be solids as you can propfade any small modelwork to fade out past the doors kind of thing. So you wouldn't end up with multiplayer-fast visleaves but they would certainly be manageable, and brush count would be reduced drastically.

Of course this is all in my head XD it's your time and sweat.

Don't forget your 3d skybox as you keep planning too

You gotta turn cheats off to enter developmenter mode?

---

The best things in life, aren't things.

"Orpheus" said:
You gotta turn cheats off to enter developmenter mode?

Oops duh sv_cheats 1 my bad

Well I got the Boat Deck and A Deck levels pretty well done and I have one (albeit ugly) changelevel in between the two at the Fore Grand Staircase. Both maps compile quickly on FULL compile, which is really nice. And the changelevel works perfectly also.

I'll of course need to add changelevels for every area (all staircases) that go between levels, as well as noclip the areas I dont want the player to go (such as jumping over railings to areas that arent accessible in a particular map)

I'm noticing Im getting a lot of degenerate faces though. Log shows 3 on the Boat Deck map and 5 on the A Deck map. Previously, I had I think only about 5 or so in the entire ship. I did notice an artifact on the roof of the bridge on the A Deck map. Despite the brush being a simple rectangle, in game there was a triangle cut shape in the middle of it that I could see through.

Gonna figure that out.

Boat Deck, A Deck, and B Deck will be the more difficult maps to level transition as they all have exterior layout and can see a lot of each other's areas. The lower levels will be easier.

Once the ship is rebuilt in individual levels, I'll be able to continue actually working on the ship.

w00t!

Haven't had a lot of time the last week to work on Titanic. Put in about a half hour tonight on the B Deck map.

Friday is a holiday and then its the weekend so Im hoping to have practically everything finished - in terms of splitting levels - over the weekend.