StarDestroyer.Net BBS

Enemy 3 of 5 - it looks better in motion, the front claws are pulsing with energy and flexing slightly while the paddles at the back push it along;



This one will shoot extremely long range projectiles that can blow up the island from way out over the ocean. Simply intercepting and killing it would take too long, so you'll need to power up some obelisks and put them in 'ABM mode' to neutralise their attacks before you can mop them up. They will be escorted by another enemy I haven't designed yet. On the later stages I might have them start aiming their main attack at you, trying for a lucky airburst.

It would be cool if one type of obstacle or enemy looked like and moved like jelly fish. hundereds of floating space jelly fish would be damn cool.

salm wrote:It would be cool if one type of obstacle or enemy looked like and moved like jelly fish. hundereds of floating space jelly fish would be damn cool.

Hmm, nice image, but I'm not sure where to work it in. Possibly if I do a stage where you have to rescue an island that has already been half-eaten by these things, there could be lots of them around as dangerous obstacles (assuming they turn up only after the more aggressive variants have pacified the area).

Creature 4 of 5. Again, this looks better in motion, as it has four counter-rotating segments (two with spikes and two with paddles);



This is a huge and really nasty creature that will appear towards the end of the game. The spines will all fire homing projectiles, with an effect rather like a continuous Macross Missile Massacre (tm). A single one will be killable using the charge attacks, but even then you have to blow off all of its spines and paddles before the main body will die. If multiple ones appear your only hope will be to lure them into a killbox of fully powered defence obelisks.

Do you create these bastards procedurally as well, like the terrain?

salm wrote:Do you create these bastards procedurally as well, like the terrain?

Yes. That's why the geometry is all smooth trig and polynomial curves, rather than an irregular mesh; they're defined as a series of function calls to mesh segment generators, rather than a 3D model file. There's a small amount of noise applied to differentiate individuals. It's a quick and dirty technique that severely limits the kind of models you can produce, but I think it works quite well for this application.

I had no idea procedural mesh and texture generation was so advanced ! A lot of the grunt work in making a 3D game goes into the artwork. Why don't the game industry use same techniques as you do to cut down time and costs ?

Sarevok wrote:I had no idea procedural mesh and texture generation was so advanced ! A lot of the grunt work in making a 3D game goes into the artwork. Why don't the game industry use same techniques as you do to cut down time and costs ?

I guess because you simply have a lot less control over procedural models than over models crafted by hand.

It could still be helpful to generate most the mesh and then tweak the it to look the way you want. Say you are making your basic FPS soldier guy. By answering a few basic questions you get a generic armored dude with preset animations for walk, shoot, jump etc. Basic attributes like height, color, general look are derived from your description. Then you change bits here and there to make him to look like a Stormtrooper, Replica, Overwatch etc.

Sarevok wrote:It could still be helpful to generate most the mesh and then tweak the it to look the way you want. Say you are making your basic FPS soldier guy. By answering a few basic questions you get a generic armored dude with preset animations for walk, shoot, jump etc. Basic attributes like height, color, general look are derived from your description. Then you change bits here and there to make him to look like a Stormtrooper, Replica, Overwatch etc.

Hmm... is there a way to procedurally create meshes as complex as characters with the properties required for a gaming character?
You need a low poly count for games because of performance and you need a mesh with the right kind of edgeflow in order to make an animation look good.

Most modellers would never model a character from scratch either. Everybody´s got a whole bunch of so called lowmen with different qualities to work off from. Lowmen are just generic looking models with little detail. You basically don´t model that much. You just take a lowman and put it into a scultpting program and sculpt. And i doubt that you could procedurally generate the detail you can create with a good sculpting pass.

I had no idea procedural mesh and texture generation was so advanced !

This is nothing; take a look at this for what you can achieve when you invest some serious time into it.

Sarevok wrote:Why don't the game industry use same techniques as you do to cut down time and costs ?

I'm using it on the birds and enemies for two reasons; because I'm not a very good 3D modeller (it really is easier for me to define cubic patches by hand than to use a 3D program to make them) and to keep the download size down. For the landscape it is a time saver even if you are a good 3D modeller, but I can only get away with it because the landscape details aren't critical to gameplay the way they are in an FPS or driving game.

Birds and these weird alien monsters can easily be generated from lathed sections glued onto each other. However a lot of game-relevant shapes are much harder to generate procedurally. In principle nearly any type of object can be created this way. However most games companies would rather just hire a crew of cheap, replacable 3D modellers to grind out a load of content, rather than take the high-risk option of hiring top-quality programmers and getting them to program a procedural shape grammar. Since most games have a relatively limited environment, the extra variety isn't a huge benefit, and in any case most creative directors are control freaks who don't like the idea of the player experience differing in any way from what the playtesters see.

Steady progress is being made though, mainly by the increasing sophistication of plug-in middleware.

Sarevok wrote:Basic attributes like height, color, general look are derived from your description. Then you change bits here and there to make him to look like a Stormtrooper, Replica, Overwatch etc.

Character customisation is usually implemented by a combination of two techniques; combining pre-made pieces (e.g. equipment) into a single character mesh at runtime, and extending the bones animation system to do things like facial feature customisation.

Hmm... is there a way to procedurally create meshes as complex as characters with the properties required for a gaming character?

Procedural generation of humans is an active research area. It's been used in movie CGI rendering applications, for large battle scenes. I'm not aware of any games that use a fully procedural approach, though several use animation bones to alter proportions and facial features (hence creating multiple characters from one mesh).

Implemented the final type of enemy, with a very basic flapping animation. These things are small (only slightly larger than the hawk) and trivial to kill individually, but come in flocks of 10 to 30, with one flock escorting each claw artillery enemy or ray bomber pair (the nightmare stars will have multiple flocks escorting them).



They have no ranged attack, but move very fast, turn tightly and try to ram the player as soon as they get close. The best way to kill them is to power up an arch that grants you the chain lightning attack, which will fry the whole swarm in one shot. Hopefully I'll be able to make that look cool. They're extremely low poly (they have to be, since I have hundreds of enemies on screen at once), but normally they move too fast for you to really notice.

Initial version of HUD;



The radar isn't working yet; it will work as an effective 360 degree fisheye view of enemy & waypoint positions, rather than a plan position indicator. The HUD elements will only appear as needed; the altimeter and airspeed display will be toggleable in the options. The central ring is the lock-on hotspot for the default (threaded ball lightning) attack.

the fisheye radar is great, I loved playing with it in Tachyon: The Fringe because it was actually useful and told me exactly where the enemy ships were.

Are you going to have some sort of melee attack? Would be fitting for the first level, when you're just some normal bird and have to face down these supernatural beings. And then later on maybe some sort of "Power Claws!" upgrade/weapon for shredding whatever you want to call those enemies.

Hawkwings wrote:Are you going to have some sort of melee attack?

Unfortunately, no, for two reasons. Firstly the physics required to do it realistically are very complex, on a par with the main flight physics. Secondly it would be an immense control challenge in a game that already has challenging controls. I'd have to put in some special stabilisation and lock-on AI to make it work, and it's a lot of work for very little benefit.

Would be fitting for the first level, when you're just some normal bird and have to face down these supernatural beings.

I have gone with human characters whose magic is based on changing into birds (which you will see in the cutscenes only - you can't turn back in gameplay) because that makes it easy to characterise them and relate them to the environment features (villages to defend, magic structures to use). Having actual sentient birds is a whole barrel of writing issues, especially since this is now supposed to be accessible to average gamers. Kind of a cop out I know but as a time-limited personal project I'm on a bit of a complexity and innovation budget.

Various friends and family have agreed to contribute free voice acting. Perhaps I will post the script here for comments, if and when I get that far.

In the beginning of the thread you said something about eneimies burning the trees. Does the player have a way to extinguish the trees if too many of them catch fire?

salm wrote:In the beginning of the thread you said something about eneimies burning the trees. Does the player have a way to extinguish the trees if too many of them catch fire?

I am planning to add a swan maiden character, whose special attacks are 'monsoon' and 'typhoon'. The first one removes toxic sludge (dropped by the rays), kills grounded shadow slugs (by dissolving them) and also puts out fires. The second one will create a persistent tornado like effect that sucks in and kills any enemies that get close (though it only works over water).

Here I thought you were just making an amusing gadget side-project about how one can fly an owl and wanted input from other people. Instead, you're making a full-fledged (if freeware) indie game. That's rather awesome.

How do you plan to make a cutscene? Drawn still images or just hastily-assembled 3D models talking.

Here I thought you were just making an amusing gadget side-project about how one can fly an owl and wanted input from other people.

Initially, I was, but it was just so easy and pleasant to program (compared to AI stuff) that I got kind of carried away. Then various people who saw it started making suggestions as to features and characters...

Instead, you're making a full-fledged (if freeware) indie game.

Well, if it turns out good enough (e.g. 10 hours plus of decent gameplay, adequate production values), I may put it on Xbox Live for $5. Even if I do that I will make the basic 'free flight' mode a free PC download.

Zixinus wrote:How do you plan to make a cutscene? Drawn still images or just hastily-assembled 3D models talking.

While 2D images would be cool (see: Ace Combat 4's awesome painting-based cutscenes),
I am insufficiently skiled as an artist to make acceptable drawings. Thus it will have to be hastily assembled and barely animated 3D models.

I am supposed to be doing real work right now, but I will do some more on it tonight.

You could always make a post in AMP asking someone with talent to make some acceptable drawings for you.

Lusankya wrote:You could always make a post in AMP asking someone with talent to make some acceptable drawings for you.

From past experience, collaboration on a project of this type is something you have to take seriously if you want it to work. If you're asking people to work for free, then if they don't share your enthusiasm or vision for the project, it'll likely end in tears. There are a lot of people posting in 'services offered' on the XNA forums, but I'm reluctant to get in touch with any for this reason.

The one thing I might ask for help with is the music. I was just going to buy a bunch of tracks from somewhere like royalty-free.tv, but if someone can do better than that I would be interested. Unlike cutscene art, music is more or less independent from the main game.

Playtesting is another story, hopefully it'll be ready for people to try in another week or so.

Well, I spent the last two evenings refactoring the code to use four cores rather than just one, and reducing some GPU call overhead, rather than doing any new features. I did have time to address a legal concern though.



Specifically the game development law instituted in 1999 that requires all outdoor games to have excessive sun glare effects. With soft dimming behind foliage of course. No 'angel rays' though, so sadly I won't be competing with Project Infinity any time soon - at this point gameplay takes priority over any more eye candy.

Got the radar working;



I uploaded a video of the enemy creatures to Youtube, but the quality came out horrible. The format should be correct for it to generate a HD version, but it isn't working.

I'd suggest putting an oval-shaped overlay on the radar, to mark the point where things cease being in front of you and start being behind you. In other words, showing the forward 180 degree hemisphere. Also, perhaps another overlay box on the radar to the extent of what your main screen displays.

Hawkwings wrote:I'd suggest putting an oval-shaped overlay on the radar, to mark the point where things cease being in front of you and start being behind you.

An oval would be problematic because the radar is based on a relative compass heading and an absolute pitch angle; it considers your absolute yaw only when projecting enemy positions (IMHO this is the most intuitive layout). Instead I've added a couple of lines showing where your horizontal view angle ends; this also minimises clutter.



As you can see, my previous statement that no more eye candy would be added was an outright fabrication. I put in these drifting low-level clouds to make the airspace seem less empty - I had to add a particle system for the explosions anyway, so they weren't too much work.