Corpse Crew Alpha Version 0.4A (Want your input)

[Satori (original)]

Sounds good, just don't get the 'just one more improvement' bug that creators get.

 

[Cleveland Mark Blakemore]

Does anybody here believe there is a cross platform graphics engine in the world that could make a $50.00 graphics card run the equivalent of FarCry/Bioshock/Fallout?

This is what is called market penetration:

http://www.nikkoid.com/ocean01.html

[Right click for fullscreen]

Now that's on a graphics card that is like some old crappy ATI 800 like mine. If you have a serious graphics card you can try this:

http://www.nikkoid.com/ocean02.html

... and that's just plain awesome. I worked on a military simulator using USAR toolkits and nothing they had could touch this engine with a ten foot pole. Simply the greatest thing since sliced bread.

EDIT: My own swim code was coming along so-so (looked almost convincing) but I dropped this shader in with a new convex plane for going underwater where you can see the ripples overhead, refraction, stippling of light .... it blew me away. Obviously this zombie game will have to involve some swimming through sewers, canals, rivers, etc. on some maps. Maybe a swimming pool or two.

EDIT #2: Somehow it is really disturbing to me to have had this graphics card all these years without realizing it was capable of doing stuff like this at a good enough framerate to support a game. To think I could have replaced it soon with a better card without ever finding out the card I had could be used for dev of a good quality PC game.

P.S. That web player has the framerate capped and is using slow timing to generate the refresh. On my machine here with all optimizations and framerate set to 200fps max the thing flies at an amazing clip between 64fps and 176fps easily in a complex scene. In the other study on that machine it was running at 200fps without breaking a sweat.

 

[Stereotypical Villain]

 

[sheek]

Well I'm getting about 5 fps on my MSI netbook, but I have about 30 other tabs open in firefox.

I played that zombie game on the Unity Engine the other day and it ran great. Amazing I agree. Got to level/room 3 before I ran out of bullets...

 

[DarkUnderlord]

That first ocean test was a little choppy but didn't look too bad at all on my current laptop:

• HP Mobile AMD Sempron (tm) Processor 3000+
  1.79 Ghz, 1.87 GB of RAM
  ATI RADEON XPRESS 200M

The second one looked much like the first and again ran a little choppy but not too bad. Not bad at all considering the laptop's four or five years old now and is only a typing machine.

And I found that Zombie game got boring by Round #3.

Got to level/room 3 before I ran out of bullets...

You're supposed to save points and buy the weapons on the shelves about the place. Either the Flamethrower in the first room, the shotty in the third room or the rifle in the fourth room.

 

[oscar]

 

[St. Toxic]

 

[St. Toxic]

#2: Somehow it is really disturbing to me to have had this graphics card all these years without realizing it was capable of doing stuff like this at a good enough framerate to support a game. To think I could have replaced it soon with a better card without ever finding out the card I had could be used for dev of a good quality PC game.

Ever do a range check on a TNT2?

 

[Korgan]

hai guise wats goin on in tihs thred

 

[Cleveland Mark Blakemore]

I think I am getting the mechanics down. It won't be your typical FPS shooter at all.

I like the idea of a clipboard where you plan your strategy each day, then execute it before sundown. By blocking this time out, we can simulate the urgency but keep the impetus of the game moving forward faster than real time.

I am working on the GUI for the clipboard now.

 

[Major_Boobage]

Wow sounds good Cleve, iam waiting for your games man! Keep up the good work!

 

[poocolator]

Yes... can't wait to see what further complications arise barring release of any actual evidence as to the existence of the game.

 

[xuerebx]

You promised screenshots 15 days ago - "Next week". Waiting...

 

[denizsi]

One problem with Unity3D terrain is you can't as yet punch holes in it. I think I can fake cave interior pretty well with it, however.

Are you referring to the the DepthMask shader solution? It seems to be as good a work-around as it gets, but it's not clear how much extra work it will require to make the AI (visibility checks, pathfinding) work like it would were it an actual hole. Have you thought about that yet?

 

[Cleveland Mark Blakemore]

One problem with Unity3D terrain is you can't as yet punch holes in it. I think I can fake cave interior pretty well with it, however.

Are you referring to the the DepthMask shader solution? It seems to be as good a work-around as it gets, but it's not clear how much extra work it will require to make the AI (visibility checks, pathfinding) work like it would were it an actual hole. Have you thought about that yet?

It's not a real hole, though, is it? It's a shader applied in place of a texture map to create the illusion of a hole.

It's okay, because I have already done the code for interior/exterior portals and it looks fine. I have an underground sewer system now that is reachable by clicking on a manhole cover, which causes it to animate sliding to one side before fading out and fading in to the sewer map.

... and for those of you who don't think I am seriously at work on this game, last night in spite of already having some thirty custom zombie model assets, I decided to save them for another game because they were not Romerish-enough.Then I went out and bought an asset set of 40 zombie models male and female that fit the style of the game better. They are absolutely perfect and very low polygon so I should be able to run twice as many zombies at once in Unity3D. The other zombie models were kind of Silent Hill-or-EC Comics Zombies. I have other games in mind for these assets.

P.S. I have about 220 zombie "faces" I will probably use to touch up the faces of those RealZombs I bought. They're almost perfect but they could use some more details, wrinkles and scabs on the faces.

 

[Cleveland Mark Blakemore]

You promised screenshots 15 days ago - "Next week". Waiting...

Soon. First map almost completed. Adding vegetation instancing.

 

[SimpleComplexity]

You promised screenshots 15 days ago - "Next week". Waiting...

Soon. First map almost completed. Adding vegetation instancing.

Jajajajajajajajajaja.

 

[xuerebx]

I'll believe you for now If everything you say is legit, this could shape up into a great game.

 

[Cleveland Mark Blakemore]

Had another look at random cities and maps. I think I might just know how to do it. That would obviously be ideal and give the game infinite replay value. You can create any object at runtime with the "Instantiate" command (all the functions are excellently named) and a prefab can hold can entire mesh with subcomponents in trees.

I was actually working on a 2D version of Corpse Crew for quite a while to generate random cities with random buildings and random names for everything. Quite a bit of this code can be ported from Angelscript to C# in Mono.

It's the identical principles just applied in a 3D environment.

Don't laugh, but I think I may design the interiors of the buildings with FPSCreator, import the DBO to Collada DAE and then just add lights and static props then populate the interior with point markers for randomization ... i.e. "this spot can hold a couch, table, bed, cabinet, etc." and within the prefab itself have a point marker for an item that can be randomized.

The doors on the buildings will lead to any of several random interiors, a manhole or grate can be dropped randomly around the city leading to the sewers, etc.

 

[tozth]

Regarding procedural generation of random cities, I don't know if you are aware of this stuff, but it might be helpful.

L-systems

Subversion by Introversion Software is based on an L-system random city generator. There's some kick-ass videos and screenshots in the project blog:

Subversion Development Diary

The paper Introversion used for their procedural city implementation

Yoav Parish and Pascal Mueller: Procedural Modeling of Cities

Something interesting, but I didn't look into it.

L-systems considered harmful

 

[denizsi]

It's not a real hole, though, is it? It's a shader applied in place of a texture map to create the illusion of a hole

It's okay, because I have already done the code for interior/exterior portals and it looks fine. I have an underground sewer system now that is reachable by clicking on a manhole cover, which causes it to animate sliding to one side before fading out and fading in to the sewer map.

Using portals is a predictable solution and it should work good enough, but you aren't limited to it. The DepthShader acts similar to a portal in a way, only faster, but it has potential issues.

Basically, you place your indoors where you literally want it to be, as if you could actually cut holes in the terrain. You connect your indoor geometry to the shader, so the shader applied faces will only render that geometry, ignoring anything else. So it's not just an illusion. Then you place a collision cube on the face so entities, that collide with the collision cube, ignore collision with the terrain within the cube and can walk right in and out of the fake hole through the cube. It's a little more work than simply using portals, but it can be immensely useful.

Uncertain problems are visibility checks made by AI, pathfinding -though that might not be an issue at all with predetermined navigation-, and rendering anything that is inside the indoor geometry for the time being, but isn''t connected to the interior geometry itself. If you have NPCs and items laying around inside the indoor geometry, they won't be rendered on the shader applied face, and you will see an empty room from outside. This could probably be resolved with some creative thinking, like dynamic parenting but I don't know if Unity supports string variables in shaders, yet.

<edit>I'd still rather use portals for places that aren't in constant sight</edit>

<edit2>I'd rather not make any sewers at all. You have to refrain from sewers as much as you can! Sewers have long become the most overused and horrible clichés in games, especially in 3D games, and especially in FPSs. They are fucking boring, and nauseating, especially when you have to go through entire levels based in and around sewers. Fuck sewers</edit>

They are absolutely perfect and very low polygon so I should be able to run twice as many zombies at once in Unity3D

Look into sprite based LOD modeling. You can literally render 100s of characters with an insignificant load on the renderer, and you could leave the LOD distance setting up to the player for immense scalability. You can also use pixel shaders on the sprites so they won't look like attack of sprites from 1992.

For distant models, you'd only need to render as many triangles as twice the limbs (counting feet, lower/upper legs, lower/upper body, lower/upper arms, hands, neck and head, that's 32 triangles per character) without losing significant animation quality (there is also an animation plugin for Unity for multiple animation instancing or something like that), and I don't think AI would be a problem at all with zombies. At worst, you'd need to look into some form of group based AI. I almost forgot: there's also a plugin to render multiple sprites (though I don't know the maximum limit or how scalable it is) in a single draw call.

edit:

They seem to have done very good job with procedural generation on Submersion. This video alone shows what seems to be on the fly generation and it looks beautiful:

http://www.introversion.co.uk/blog/20081223/test2.avi

 

[poocolator]

You guys provide some awesome reference links. That stuff is alluring.

 

[Sue Sweet]

Sounds great

Cleve throw us a bone and post some screenshots

 

[Cleveland Mark Blakemore]

I have this great biohazard decon guy with a flamethrower I desperately want to use and I got to thinking about random events in the city during the day the way Fallout 3 throws them into the Wasteland every once in a while. For example, you come around a corner and a bunch of madmen on a decontamination crew are flaming everything - zombies, civilians, you, anything they see on the street without partiality. This model has a big fuel tank on their backs, so a sniper shot to their flamer fuel should provide some hilarious real-time physics fun.

I was also thinking about random events like a tank of soldiers passing through, also on a rampage to kill everything ... or gung-ho civilians on the roofs of buildings sniping at anyone who approaches their doorway, etc. ... one in a while you could find a corpse on the street with keys or other useful items in their pockets when searching. Once in a while some jets could fly over and do an airstrike or drop a thermobaric bomb.

I think I can even do random street names by writing to an offscreen texture used on the street signs, in addition to random building names and random character names when you bump into civilians or soldiers.

As originally discussed, the best thing to do is to determine the flavor of the entire game (escape to cruise ship evacuation, find helicopter pad on roof of skyscraper, etc.) and then everything proceeds from there tailored to that storyline. The player always starts in the middle of the randomly generated map and will have to work their way to some location on the edge within X number of days.

 

[Solivagant]

Cleve, have you got any of that stuff documented? Like all the things a player can do, weapons he will get, etc etc etc? Cause its sounding like you're just adding stuff as you dream it up, and I think it would help if you started just writing what you really want the game to be about, rather than decide on features based on how easily you can implement them or get 3d models for them. Of course design based on what you can get is good, but if you keep going this way you're going to have:

-swimming sections because the swimming looks good on Unity3d
-more zombies than previously planned because the new blister pack of zombies you just bought have lower polys
-random cities and maps cause it makes for infinite gameplay
-clipboard to plan the defense

And who knows what else! The randomization alone is a tough nut to crack, worthy of a game totally devoted to it, especially from a one-man team.
I urge you to write down what features are essencial and narrow your scope a bit, otherwise how are you going to release Corpse Crew and Grimoire this year?

I appreciate your extensive prototyping though, that's a good practice and your game will gain a lot from it. Just don't start tacking every single "cool" feature you can think of, because it will probably bog down your work.