Digital Amusement Game Development
  
|
Superman_b4 has grown so large, the compiler is choking. My machine which is a AMDxp2000, 1 GB PC2100 ram, is having many difficulties swallowing the huge amount of data. It can take over a half minute to do a simple bsp compile. Light -fast is under 40 minutes. Last night I ran "light -fast -super -patchshadows -bounce 2". When I check it this morning, it didn't take and I don't know what the error was (cmd window was gone). I don't have enough internal processing and throughput to work with this much data in a reasonable time, which sucks. However I'm a month short on the major upgrade. Meanwhile I can still design and do brushwork. My only fear is that it will become too big... and the compiler will refuse to compile. That would break the August date, and I can't allow that to happen. Fortunately the compiling issues are not hindering gametesting much. I did encounter a problem last night where the bsp compile took about a half hour and crashed. This was not looking good, but a reboot cleared up the memory fragments and allowed for a compile. The offender was the water shader... when I was testing out tesssize, I knocked it down from 1024 to 4, meaning a texture shader surface had went from forcefully being broken into 1024 unit tri's, to 4 unit tri's. I had somewhere from 10 to 100 thousand extra triangles in the water because of this. Ouch. On the plus side, I changed it to 512 which resolved another problem of overbreaking the surface, everything fits nicely now. The water shader looks beautiful, one of the best water shaders I've seen (and yes I made it from scratch, including adding the reglow effect). I've also redone the stoplights. Much much better looking now. Still some slight changes to be made, but that's kinda standard with all the objects at this point. I am looking hopefully to the polishing stage, after everything is built and into public gametesting.
Also I fixed the door textures (oh yeah I have new door texs for the hotel and military building front doors). I downsampled the sweet-looking originals, then applied them and they looked horrible! I went back to photoshop and resized the originals to a nice X^2 by 512 size, then rebuit the doors to be X^2 by 512 as well. I won't get to see the rendered result til later, but so far seems like it did the trick. You'll like the door eyecandy... friggin texs are large enough too. But sometimes certain details require that extra bit fat. It makes a big difference between "it's a door" and "/say wow that's a good looking door!". Another thing I'm trying to scheme up is a way to manipulate the lightmap to give me depth shading when compiled, not as a shader pass. I'd like to do this by creating a b/w source tex and telling the compiler to alpha filter the lightmap according to that source tex. Might be something in the q3map2 manual, I'll look later.
I've done some reworking of the water shader. Experimenting with "deformvertexes wave" to add some life to the water. This is similar to what was used in druglord2 and 3 to make the water wavy. I want some damn good looking water, dammit! Also am working on redoing the streetlights, which have so far proved to have a successful prototype. My woman didn't understand why I was out taking pictures of parking lines... at first. :) Tonight I'll hopefully have the full ground plan done. I need to draw up all the streets and buildings that are outside of the gameplay area. I'm thinking the highway is going to split to the west and get grounded to the east. Rear of the hotel there's going to be a squarish building with an empty center. Second will be mult-tier. Building next to the park is going to be another radical design, thinking something looking like a futurish empire state building. Flypoints to be added to hotel upper roof, top of military building. Lt1 has a tv newschopper (another combo point) in the works which will hover over the highway. Found a great architecture pic to design the maintenance floor against (soon to be renamed). I've got a bunch of cosmetic ideas to fix up the map, but I really have to remain focused on the basic gameplay first. I have roughly 8 weeks to have the map in a brushwork/lighting playable state, followed by vis work, touchups, additional shader work. After comes the playtest version for two weeks - expect multiple version releases... i.e. superman_b4_testrelease1 etc. just as I did with b3. There were 5 or 6 releases like this if you remember correctly. The result was a very polished beta map. Also in the beginning of June I'll have my upgraded machine to do the heavy, demanding compile tasks. Instead of 14 hour compiles, hopefully I can knock them down to a couple hours.
I've also been extensively reading up on game development and design. Wikipedia is my new online best friend. Being a game designer or producer is NOT easy, it's not standardized workflow, requires a lot of figure-it-out-yourself. Sounds like my kind of gig. I've decided that superman final will be released before August. After that, I will begin learning a new game engine that involves extensive use of physics. Right now, Unreal Tournament 2007 has my attention. They've also built the engine to take use of the new Aegia Physx PPU. Remember kids, physics are fun! The other direction I would move would be to gameboy type games. I really want to check out Geometry Wars on the xbox360. I have a feeling retro gaming is about to me a huge comeback and I'm thinking this would be the time to jump on that bandwagon. I just watched a keynote speech by nintendo's president, released yesterday. He has confirmed -exactly- what I've been talking about with the new Revolution game system. The use of the nunchuck controller configuration will enhance the FPS genre greatly. That's right, we're talking about the next generation of first person shooters, and on a console? Nintendo? How is this possible? It's great! He also addded the nintendo virtual console for playing retro games will not only include the library of 8 and 16 bit nintendo games, but also genesis and turbo-grafx games as well! In a time when pretty graphics are becoming old and gameplay is taking the spotlight again, I foresee nintendo as dominating sony and microsoft in the console war. In the current generation of consoles, nintendo is seen as "the other guy", next gen, the foot of ass-kickin is gonna deliver. Note this though, each console will have it's distinct advantages: microsoft has xbox live - conecting the console community (however sony and nintendo are taking on their own services now); sony has an all-in-one media system, including backwards compatibility; nintendo has a unique controller for a new gaming experience, plus the features of the competitors. In two years, I then predict the console war will stand with nintendo as the leader, sony number 2, microsoft last. However, microsoft is a sneaky, aggressive, and bloodthirsty company, so I'd expect them to steal as much of nintendo and sony's innovations as possible. Of course the flipside is that microsoft created the online console community... Back to the original subject. Work on superman continues! I have approx 3 months to create a leveltest release. It's on.
I redid the upper parking garage again. This time I like how it's turning out. In the tradition of the rest of the buildings, the 2nd upper floor and rooftop will be accessable. The highway now sports construction like any good highway has, indicating the launchpoint with the cones. You can see new streetlights, different water tone (still working on this), addition of more street paint, a sign at the gameplay edge of the highway, new building in the back. Not pictured - new bus stops next to highway, Y-intersection behind military building, evolving front of Pond Publishing, modified bridgeway behind military building, military building window texture mods and optimization. Removed front building lighting to be replaced with something else (spotlights maybe?). Running around in the map continues to get denser in the way of enveloping me in a full, vast urban city environment.
A friend of mine has been recruited into NCSoft, makers of Lineage 2 and City of Heroes. Also as I recently read in Game Developer magazine, many previous members of Origin Systems (makers of the Ultima series) including Richard Garriot (a.k.a. Lord British) reside there. I've logged many, many thousands of hours into Ultima 5 (imo the best rpg ever made), Ultima 4, and Ultima 6. These games are one of the prime reasons I wanted to get into game design, My envy is high, and my ambitions burn hotter.
Otherwise, I've been spending a decent amount of time talking game industry politics with friends. Such as direction of the industry, current console generation, WW2 FPS's, etc. There's been some buzz around me and my workplace about me doing modding and level design. I'm finding out there's more gamers around me than I realized. It's pretty cool when the people I normally escalate issues to come to me asking about game stuff... that ends up being a lot of fun. I took a long trip in the past week, gave me time to do more thinking about the layout of superman. I figured out a bunch of new details to be added, maybe a bit more on the direction of the map. Still have to go through curve hell for the frontside of the west building... not looking forward to that but the results will be beautiful. I still need to finish the q3 pacmans. I have no other maps in process of being worked on, although the thought of doing more work on temple2 is tempting. Yes yes meanwhile I will be doing work on superman. Keep your pants on. Nothing else to speak of for now, getting r/l solid then gaming may commence!
When designing superman beta 3, I had been looking all around downtown to see what makes reality "real". I found it's the suttle details that blend into, everything. First, streets are almost never level. It takes effort in real life to make flat areas, and those don't stay flat for too long... erosion has it's way with human meddling. We should not see a single flat map even if it's a parking lot. Heck, those are engineered to not be flat as to move water in rainstorms. I'm finding that a level designer should have a few conceptual understanding to accurately portray their work. Objects have mass. The ability to portray mass creates respect to your surroundings. With time, erosion occurs. Heavy things buckle, crack, or break. Areas with traffic, movement, or interaction incur wear. Cars rub curbs with their tires. Grass edges get worn to dirt. Garbage and litter accrue. Finally, an element I find to create the most classy details are representation of engineered compensation. This entails details for things that will never occur in your map, but if time were persistant, it would or could. Eventually, it's going to rain. You're going to want to eat, sleep, or pee. Make a phonecall. Some kid is going to climb that obsticle and someone is going to get sued when they fall. Street lights have control boxes somewhere near. Where do firemen find the nearest fire hydrant/water supply when your building is burning down because it's nothing but a box that doesn't follow city codes? And how are people getting out of the building? Are there enough exits? Signs? Street paint? I had a thought that blew my mind while making superman: should I make the curbs A.D.A. compliant? Followed by a thought of WTF am I thinking?? But it's stupid little details that add so much life to your map. Am I preaching these details as necessities? Certainly not, these are hints as to how to figure in more lifelike details. Basically, if you think real physics and problems, you can add simple details over existing details that show you really are paying attention to how things actually work. But do you really need to know how it works? It's hollywood... the more you know, the better the detail works. A well placed minor scuff or couple extra brushes to de-flatten a surface can mean worlds of detail.
In other UrT development news, FSK405 Bah is tearing it up with his first map. It's REALLY big but has a lot of neat little places to Do Stuff with. We like things that do stuff. I'll keep you on top of this aspiring new mapper. Keep up the great work Bah!
Think about this. The .map source file format is so human readable, it could in fact be generated much the same way GenSurf works. I could potentially create a script that could create the basic brushwork in a map, texture accordingly, fill in some details, I tweak the rest. Yes, I could essentially create a ut_Casa with a script. That how damn simple most first party really are. I'd create an algorithm that makes one area, create three paths that interconnect with each other twice, and connect it to a second area. All 90 degree turns. Maybe some raised/lowered areas. This generic formula works so well it's sick. And it's old. How many maps use that exact formula? I would expect more gameplay quality out of a 1st party map. See, this simple formula almost guarantees an even playing field. This brings me to a possible solution. Since these 1st party maps seem to be being built with competitions as a focus rather than pub play, try this idea. Build your map lopsided. In the real world (oh no!) one side is almost always going to have an advantage over the other. Standard competition cycle is 3 rounds. Ok, 3 rounds it is... 2 games per round. Each team cycles sides per half-round. Had the advantage before? Now you don't. 6 games in the same time of 3. Feel the pain. Anyway, back to the issue at hand. I'd really like to see stronger landscape and more innovative structures from first party maps. Pull out google earth and take a look at various urban environments. Many curves and irregular paths. Ditch The Formula and use your environment to institute gameplay, something that's not often taken advantage of rather than simple paths of least resistance with a couple details. Don't complain that creating those types of environments are hard - that's like a programmer complaining about not wanting to use advanced data structures. The beef is done.
POST-MORTEM The breakdown: Saturday night while playing on the FSK405 pub, talking on TS, this thought came to mind... make a pacman map... I thought about it for a while, then in one fall swoop I dropped everything and got straight to work. First I had to find an image of the pacman layout. The in GtkRadiant I had to make some reasonable guesses to what the appropriate dimensions of the map were going to be... this was critical in determining if the map was going to be any fun. I picked 64x64 unit grids (approx. 2 players wide/long), and the walls to be 96 units tall. Next was the wall thickness. Also I made the upfront decision since I was going to do these maps right, all edges needed to be curved/rounded. Teleports would work as expected. There would be breakable pac-dots. Had to get ghosts and fruits in there. The minimap would look like the arcade. Custom textures. The startup levelshots would be custom art done by myself. Bot support. All gamemodes (minus bomb). See there's really a lot that goes into the preparation of even simple looking levels. Doing the brushwork was cheesecake, just had to keep referring to the arcade map for correct sizing. Ran a quick test so UrT would generate the minimap. Cut the icons from the arcade map. Adjusted the colors on the minimap, added the icons. Sometime in here I was creating the levelshot (startup image). Began creating the textures in photoshop, this was pretty easy since the texs were just spackles and cloud gradients. The I built the shaders. Skybox is all black, so I ripped that shader from another of my maps. Cut the ghosts and fruit from the arcade map, created 128x128 non-aliased textures. Decided the dots would not be lightmapped (that would potentially cast shadows and light making spots look weird when you busted them). Began importing and editing sounds (thanks FSK405 Bah for finding them!). Added triggers for sounds, rigged up the teleports. I got the idea for the fadey thing the teleports do from Bah's first map he sent me, forgot quake did that. Added clips, decided I didn't want people climbing up the walls... in a playtest that made the map pretty weak. Added ALL the dots. This proved to be a problem later. Added the ghosts and fruit. Rigged up all the triggers. Thought I had myself a complete map... and right when I thought it all came together - it all came crashing down. Couldn't load the map, Urban Terror spat back Maxmodels_exceeded. Never seen this before. Worked backwards and tore out some triggers. No good. Tore out the ghosts and fruits. It loads. DAMMIT. Well I try fighting some bots. Weird things going on, stuff not there, playing on a team in spec... it's kinda jacked up. By myself it was ok... fought with this for a few hours. Finally gave up said fine no ghosts or fruit, how disappointing. Packaged the map. This is why you make test versions before you just up and release a map. Got people playing... no one could shoot anyone. Oh no! What the hell? It all made sense at that point. The error told me what I knew but didn't understand, too many entities in the map. Entity count wasn't much, all I had was some triggers and the dots... the dots!!! Too friggin many breakable dots!!! Ok, now I had to do damage control and recall the map. Jumped back into the map, got constructive as to dot placement whereas they looked good and was under the entity limits. Dropped the ghosts and fruit back in there right away too. Sure enough. Decided to get a little fancy and add some hinting and run -vis. Didn't help much, but I wasn't going to put much work into that. It was time to release this bad boy! Total time was about 16 hours, 24 total hours elapsed. Not bad for concept to completion!!! The first night of playing was a big success. So big that I said, hey I can use the pacman resources and make... Ms. Pacman! And not just one level, no I wanted to make all 3 and release a mappack. I got straight to work again. Decisions made here: consolidate resources by creating one shader file, reuse textures, reuse brushes and groups, reuse sounds while adding new ms. pacman sounds. I figured the only thing to do was going to be adjust the textures, redo the minimaps, redo some sounds, and of course move all the wall brushes around. I was mostly right. I hit one snag that was perceptually unacceptable. I went much farther with hinting these maps. The floor is semi-transparent. Problem that was occuring was the skybox floor was not being drawn in some places in the map, others it was, and would cause visual fragments. Took me about an hour to figure out why it was happening. How to fix it took a little longer. I added surfaceparm trans to the floor shader. This means when -vis does visibility calculations, it will treat a brush as see-through, thus not blocking visibility. However this still didn't fix the problem. I realized the other 5 faces of the brush were caulk, and needed to become something else that had surfaceparm trans - like weapclip. It's a fully clipping texture/shader. This fixed the problem for good. I ran a test of the maps with FSK405, checked out well. Last thing I had to do was the levelshot artwork. Level 1 is my favorite, representing an art style not too many other recognize. In total, roughly 48 hours elapsed. Completion of all 4 levels in 4 days. I was kinda dedicated to the project... staying up for 2 days with a 3 hour nap. The first image may have been the result of sleep depravation, perhaps with the assistance of Major Vodka and his trusty steed Red Bull - probably why it's the darkest of the 3. Yeah. :) Maps packaged, tested, shipped! From this experience, I learned a bunch of new things making these maps. First, I'm going to have to watch the breakable entity count in superman. Learned more about -vis. Learned more about my own limitations. Some of the good things that have immediately come forth is I'm seeing people's interest in mapping/level design popping up. This is a Good Thing! I'm going to focus more on teaching and tutorials for right now, need to take a break after that stint. Hope this post-mortem was comprehensive enough, and I also hope to see work you've done. Biggest issue I hear of is the embarrassment of being unable to do simple things that I can do with ease... I'm not here to judge, and I've been there, and I haven't forgotten. FSK405 Spider asked me "I want to start a map, what do I need?" and I said "a lot of patience!". It's true. Once the mapping bug bites you it's tough to stop. FSK405 wily duck :V Next page |
