Is Ageia\'s PhysX Failing?

[twile]

how about you try to address the pros and cons of hardware vs software physics, using past and present products, as well as possible future scenarios? I'm sure it would be much more useful and relevant to the article this thread was based off of.

I already did. Please read the thread before making conclusions.
I have read the thread. If you're done making useful input, then fine. Just stop this pointless bickering because I think everyone is fed up with it.

What would really enhance the believability of games is if such destruction could truly be handled in real-time, and be completely variable. In a word, what would really convince people would be a small-scale scientific-grade physics simulation on a card. "

No offense,but this is as stupid as it gets. In order not to script events in the game, you have to make an advanced AI. Now,in COD 2,for example, you are running through the trench, while the (scripted) destruction takes place around you. Imagine the writing of the code that would allow a great number of your opponents to interact,change opinions and experiences and THINK!!! They would then be able to spot you,tell the tank commander about it so he could fire a shell directly at you. This is something we've seen on a small scale in F.E.A.R. and other games, but i guess you wouldn't expect it to be possible with 100 (or more) artificially intelligent objects (enemies). Or,if You have,lets say,an gigantic underground worm that's "chasing you".He's supposed to jump out on a certain place,and kick something off (usually,to devastate something completely).You would NOT be able to destroy it yourself before that,with or without the physics card!!! If there wasn't for the trigger zones in the games,they would not be possible at all.You just cannot build a perfect universe inside your computer and just join and interact.
Imagine a machine You'd have to have for this.Also,this opens an issue of enormous media able of storing and transfering this data, and,of course,the sheer time and resources required to make such a game.

I think you both have some merit in what you're saying. It would add to the processor load if you were to give AI more autonomy and fewer scripted actions. And it would be cool for explosions to look less canned. Maybe there's a compromise?

Say you're running across a field and you're supposed to get shot at by mortars. First the game devs want you to know you're getting shot at, but not get hit. So they define some region 300 feet away or more, and have the mortar crew hit a random spot in that region. Then they want to start freaking out you and your allies, so they hit in a closer area, where your allies are but you aren't. Then they want to give you a good scare, so they drop a mortar somewhere near you, just close enough to knock you back and make you dazed for a few moments. In this way, you don't have to actually add to the AI, you just add some randomness to where the mortar round goes and let the PPU do the math of the crater and debris. If the game devs think they can implement better AI that will work well in generatic those cinematic moments while not overburdening the CPU, that's all the better.

I think the limiting factor on physics simply lays in updating thousands of objects, both across the system bus and more importantly over the net. Most games deal with a few dozen objects and crap out with more.

I'm guessing the biggest growing market in PC games is MMOs or multiplayer and I can't fathom keeping thousands of objects syncronized over the net. Physics in games are good for a for a few interactive objects at a time and for graphical effects. I really hope physics improves, but I think the bottleneck is everything but the math itself.

Wouldn't this depend on how the game works during multiplayer? If you load up the map on everybody's computer but don't let any of the players join, it should be exactly the same on all the computers. Given what a player does in that level, his/her computer can figure out how the level reacts, right? So by the same token, given what all the other players do in that level, the computers can each also figure out how the level reacts. What I mean is, instead of synchronizing predictable things, such as physics effects, just synchronize the unpredictable things, the players. If every computer knows that Player A threw a grenade at x degrees, y feet per second, and facing direction z, then every computer can also tell what its trajectory will be, what it will land on, how far it will slide or roll, what (if anything) it will bump into, and as a result, where it will be when it explodes. If every computer knows precisely where the grenade is when it explodes, every computer can also tell what forces will be applied to the nearby objects, which ones will move, which ones will break, and what directions and speeds the broken pieces will have. All this relies on is the assumption that any in-game event, when done many times, will always be exactly the same; that way, any number of computers doing the event in multiplayer will be guaranteed to end up with the same result.

 

[SuperG]

I wouldn't not take the expection to high for very destructable and interactive and in high detail simulation of destructable enviorment.

i don't think such Killer game comes in the near future. I think such game take a third or Fourth gen of PPU to do in a resonable detail.

Destructable terrain is already done at very small scale but restricted. Red fraction. But I think what you want need a 10000 or more power solution. PPU is not.
It do wil give you more Destructive interaction but at what level is possible?

First destruction must be generated on the fly wich means the enviorment geometry changes so gets more complex. the more demolished the more geometri adds to it. So renderload adds to it. A flat wall could be implemented with one quad for back and front and if not flat with bumb maps. A nade shell in it would generate a complex structure of that wall with a more complex geometri hole in it and larger and smaller debri which the larger parts must sta in game. extra objects to and some Particles. A hit in a corner or side wil give more complications.

Geometri will grow's in destructable envirments games.

I think Destructable feature can be implemented on a larger scale but still Fix. Or at a tad more larger scale but more realistic simulated not Fix damage. Or a Mix of it.

Then in time this will scale up in more detail and more larger scale if there come nextgen PPU's.

The detail and scale they can choose from is large and the impact on Physics hardware power requierment scale up drasticly on wich detail you aiming for.

There is a large road to travel between what now is done on CPU to realistics reallive destruction.

So expect more destructable enviorment then ever before. But it would still be done on a more limited detaild scale.

As for AI Predefind Pathing routes this AI solution doesn't work more.
Would look stupid if you AI teammate don't follow though a large wall hole. But take a predefine path around it.
Ore a Path blokked and can't handle it and get stucked.

i think AI is not a large problem but it needs be done with a more adaptive solution. It could be alot more work on the AI programmer.

 

[twile]

I just remembered a point people made earlier about more realistic physics with more objects being demanding on a system visually, as it's more polygons to draw and shadows to cast. To that I'd like to add in the additional CPU load from increased AI requirements, meaning that the AI now has to anticipate and cope with physics-based attacks, intentional (such as starting a rock slide) or accidental (so they'd duck when an explosion sends a piece of shrapnel their way). If you go from having 100 movable objects in a zone to 10,000 movable objects, the AI has 100 times as many things to keep track of, in terms of their immediate threat ("are they on a collision course with me, and if so, are they dangerous?") and potential threat (such as someone setting up a phsics trap).

However, despite the additional strain on systems added by huge numbers of objects, I'd consider it to be worth it, given the amount of eyecandy (dust, small explosion craters, liquid droplets) and gameplay physics (destructable environments let you forge your own path, using a grenade's concussive force to blow away a smoke screen, using physics to create traps, etc) which could be added. After all, lots of the people here seem to be quite certain that CPUs and GPUs will be able to hand most anything you throw at them 😉

 

[SuperG]

In current games AI avoid nades. it's just a different avoidin. Blast radius and Cover blockin vs flight path of dodgable debri.

 

[Eviltwin17]

the way i see it is that ageia is going to end up like voodoo. They both had great ideas and used them but unfortunately went out of business because competition was too big. Ati and Nvidia have ideas for physics solutions that may be much cheaper than ageia's physx card. This will most likely put ageia out of the physics business.

 

[DrBlofeld]

I just remembered a point people made earlier about more realistic physics with more objects being demanding on a system visually, as it's more polygons to draw and shadows to cast. To that I'd like to add in the additional CPU load from increased AI requirements, meaning that the AI now has to anticipate and cope with physics-based attacks, intentional (such as starting a rock slide) or accidental (so they'd duck when an explosion sends a piece of shrapnel their way). If you go from having 100 movable objects in a zone to 10,000 movable objects, the AI has 100 times as many things to keep track of, in terms of their immediate threat ("are they on a collision course with me, and if so, are they dangerous?") and potential threat (such as someone setting up a phsics trap).

Not quite true. The additional CPU load can be negligible with efficient programming. An object is just an object, and boundaries can be placed around them to represent the object as a simple box. They can be categorized into different dangers. The difficulty in AI programming is not keeping track of objects as you imply, but with intelligently responding to the situation or objects at hand.

Here is another scenario:

The new AIPU ( Artificial Intelligence Processing Unit)

The new AIPU will be a PCI card that costs $300 and comes with an API for game programmers. It will feature a dedicated processor for solving multiple path-finding problems within complex 3D landscapes. It will be responsible for dictating actions through a motivation model that can accept multiple traits and inputs. The company will introduce a sample game that proves that you need an AIPU to run it. Someone figures out that you don't need the AIPU to run the sample game. Soon Nvidia and ATI will release an AI API that is GPU based and then there would be countless 10 page threads discussing it on the Forumz.

 

[3lfk1ng]

Your so right

 

[SuperG]

the way i see it is that ageia is going to end up like voodoo. They both had great ideas and used them but unfortunately went out of business because competition was too big. Ati and Nvidia have ideas for physics solutions that may be much cheaper than ageia's physx card. This will most likely put ageia out of the physics business.

So wrong. If they go the Voodoo way it means First a couple years of Sucses. until they get taken by the competition in the very longrun.
Wich means Ageia is goin to be sucses full and a bad situation like Voodoo for Ageia is to far in the future.
The question is for now will Ageia kickstart the new market. As it looks like the competiton kicks in very early compared to 3DFX. So will they have a or more sucsesfull years?

nV/ATI frequently run second place. While that's no problem if you are a big OEM player. 3DFX is not its purely aimed at the Retail Gaming market. Who runs behind there get's a blow. For 3DFX it was devastating. When nV runs upfront a bit to long.
Now does Ageia falls in that picture.

And If I recall 3dFX had some years of no competition. Glide games.

it could be even worse. Ageia Kickstarts a new Market and wake up some money smelling big ones ATI and nV so the Market is there started. But could be kicked out sooner then later.

But its to early to conclude that.

 

[twile]

Here is another scenario:

The new AIPU ( Artificial Intelligence Processing Unit)

The new AIPU will be a PCI card that costs $300 and comes with an API for game programmers. It will feature a dedicated processor for solving multiple path-finding problems within complex 3D landscapes. It will be responsible for dictating actions through a motivation model that can accept multiple traits and inputs. The company will introduce a sample game that proves that you need an AIPU to run it. Someone figures out that you don't need the AIPU to run the sample game. Soon Nvidia and ATI will release an AI API that is GPU based and then there would be countless 10 page threads discussing it on the Forumz.

Funny you should mention that! I heard talk about an AI card some time in the past year which was designed to do just that. Well, at least the path finding and such. At present I don't really see much of a need for something like that, I see it as a different situation than physics has. With physics, you can add tons of destructable objects to most any first person shooter, and those add depth to the game. That's fun and good. AI is different however, because in most games the player is up against many foes who are equally strong. An AI card would be needed if the game were to require a much larger number of enemies or allies (which might be useful in WWII-type games with large numbers of enemies) or if the AI were to become significantly more advanced. However, modern AI implementations seem to do pretty well, and as the CPU mostly has to do AI and physics, it wouldn't have much to do...

As far as both AI and physics cards go however, I'm going to wait this one out for at least a year. Crysis looks like it has good AI, responding to dynamically changing environments (such as taking cover behind trees you just shot down), and good physics as well (shooting down trees and breaking stuff, responsive foliage). Seeing as that doesn't require a physics card or make use of GPU power for those things, it should give us a good idea what more advanced physics and AI look like on dual/quad core processors, with regards to overall detail, scale, and performance.

 

[Ragger]

A PPU handles AI too btw.. and why would you need to add super ai to games because you add more realistic physics in a game?

how objects and stuff react doesnt require some revolution in AI too all of the sudden 8O or does my logic fail me and we need to have 60" plasma screens when using a PPU because my pink elephant says so?

edit: it doesnt handle the AI(but maybe it could), i put the wrong words for it but it helps AI and can improve AI in lots of ways so a CPU would still handle the AI but stuff like movements and reactions etc... would be taken over by the PPU, movements and reactions is like(my gues) 80% of gaming AI anyway so lol.. maybe i was right almost.

 

[MilanChe]

What I see as a REAL advantage is possibility of complex sound interactions and even weather patterns.With a really mighty PPU. Maybe third or fourth gen.

 

[Lord-Ilpolazzo]

I think Ageia's product is a great idea. But yes $300 is a little high. $100 would be a better price point for mass adoption. I realize they have to sell it high now because they're not selling in mass yet. With the proper adoption by game developers, the product would be a great investment.

The issue is not whether we need physics acceleration in games, but how we're going to implement it. The article said it best that the beauty of GPU accelerated physics is that even if the game doesn't support it, you can still use both cards for rendering. But the nice thing about Ageia's card is that you can still use both GPUs for rendering, and have a dedicated physics processor thats designed to do only that.

The main things Ageia needs to do right now is get their costs lower and also switch to a PCI-E bus for the card. PCI is too slow. Another step would be to work with Nvidia and ATI at seeing if they can sell their chips directly to both companies, and put the chip right on the GPU. Then they could sell millions of chips which means they can offer the chip at low price to the big boys. It also eliminates the slow cross talk between the CPU and the PCI bus.

"switch to a PCI-E bus for the card. PCI is too slow. "
Who are you to say that PCI bus is too slow for the card? neither You nor i have any relaistic idea how much bandwidth such a thing needs. Natrually, Agea know EXACTLY how much it requires, and they deem PCI to be sufficient. Therefore it probably is, and people should stop griping on about what interface it uses and cosider its output/performance alone.

I cant see any reason why the would cripple their flagship device with the "wrong" interface, and indeed i suspect there is little difference in the cost of PCI-E or PCI on card interface hardware, which rules out the issue of cost.

Personally i assume that a physics card takes in various equations, operands, co-ordinates object nets? and they goes on to perform complex operations on them. Which would mean you give it realative;y small ammount of data, which it then goes on to do a helluva lot of calculations on before outputtting a relatively small ammount again as the result. If it works anything like that... that is.

 

[twile]

A PPU handles AI too btw.. and why would you need to add super ai to games because you add more realistic physics in a game?

how objects and stuff react doesnt require some revolution in AI too all of the sudden 8O or does my logic fail me and we need to have 60" plasma screens when using a PPU because my pink elephant says so?

edit: it doesnt handle the AI(but maybe it could), i put the wrong words for it but it helps AI and can improve AI in lots of ways so a CPU would still handle the AI but stuff like movements and reactions etc... would be taken over by the PPU, movements and reactions is like(my gues) 80% of gaming AI anyway so lol.. maybe i was right almost.

I... uh... don't think that's right at all. As we've seen, the Ageia card barely has support for physics in games right now, and it's optional in both GRAW and CellFactor. Reason for this is that currently there is no physics standard. Nor is there an AI standard. So far as I no, the PPU does absolutely nothing with regards to AI. Offloading physics to the PPU might give the CPU more power to do the AI, but that's not hardware AI support.

The point I was trying to make is that the AI won't have to get more complicated, it will just have a lot more to do. Consider if you have a group of 15 enemies in an area. In most modern games, the only movable objects they'll react to are grenades, and that's pretty simple. If the grenade's trajectory crosses within a certain radius of the enemy, look for cover or run. And there are usually, what, a couple grenades in play at most? If you move to physics-based damage and AI that responds to that, you have more stuff to do. First the enemy has to check if a moving object's path will cross its own at the same time, i.e. if it's going to hit the enemy. Also, the speed, size, shape, and material of the object factor into whether it does damage, so an enemy won't freak out when you throw a teddy bear or pencil at it, but it will avoid shrapnel and throwing knives. Now, that may not be complicating the AI too much, but if every object in a room or environment is movable, and there are a much greater number of objects (thousands instead of dozens) we're talking several orders of magnitude increase of how many objects each enemy has to analyze and react (or not react) to. Just how much AI is taxing to a CPU right now, I don't know, so this might not be an issue.

If you want to add more to the AI, such as actively using the new physics features, it gets more complicated. Now the AI has to realize that its cover might be destroyed, moved, or simply shot through, and react to that. It has to actively look for physics traps, not just imminent threats, so it needs to be able to analyze configurations of objects which might be able to fall on, trap, block it, etc. Even more complex, it could be able to set traps such as these. Path finding could change as well, the AI needs to recognize that the best way to its objective might be forging its own path by moving objects or destroying obstacles, such as walls, wooden doors, trees, crates, etc. Adding in AI support for such features would of course be optional, but could result in some humorously bad AI if not used (putting a huge cardboard box over an enemy and having them not know they can shoot through it or throw it off would be pretty funny to watch, you could incapacitate the mightiest foe with recylcable materials).

 

[Ragger]

A PPU handles AI too btw.. and why would you need to add super ai to games because you add more realistic physics in a game?

how objects and stuff react doesnt require some revolution in AI too all of the sudden 8O or does my logic fail me and we need to have 60" plasma screens when using a PPU because my pink elephant says so?

edit: it doesnt handle the AI(but maybe it could), i put the wrong words for it but it helps AI and can improve AI in lots of ways so a CPU would still handle the AI but stuff like movements and reactions etc... would be taken over by the PPU, movements and reactions is like(my gues) 80% of gaming AI anyway so lol.. maybe i was right almost.

I... uh... don't think that's right at all. As we've seen, the Ageia card barely has support for physics in games right now, and it's optional in both GRAW and CellFactor. Reason for this is that currently there is no physics standard. Nor is there an AI standard. So far as I no, the PPU does absolutely nothing with regards to AI. Offloading physics to the PPU might give the CPU more power to do the AI, but that's not hardware AI support.


i think AI is very off-topic in this thread, i shouldnt have responded.. that a PhysX not barely supporting anything atm is not so strange and have been answered several times in this thread.
there are alot of physics standards, PhysX is one that takes it to hardware form for games, theres alot of software standards too, and there will be other standards using GPUs, theres also tons of professional standards used by nasa and other companies, in car simulations, crash tests stuff etc.. the tools are there, and the dev-kits are there now for game devs.
theres 60+ games comming in the first wave with PhysX supported games, they will have minimal support and most likely only "eye-candy" use of the PPU thats barely noticible.
the 2nd wave will try to explore more options and make more use of the PPU, the 3rd wave might explore the "invisible" physics and start using the PPU what it was meant for(thats in like 10+ months).
the 4th wave might be awesome games using a PPU efficiently, metal sheet doors flex realistic when you bump into them, plexiglas ingames react like real plexigla, realistic car crashes where every little part has its own realistic mass/values, game play never before possible and not thought of today.
dont judge the first waves of games with support for PhysX, they WILL probably suck and the support will be minimal, its natural and how this goes. it will be atleast over a year minimum before we can see what a PPU realy can do. gamers wanting to have stuff right now will complain about the card, it will get bad reviews, it will get smashed to pieces. but eventualy it will come throu all that if it is a good product.

my opinion about GPUs being used have been said a couple of posts back, but since they have "big names" and tons of cash, and generaly private armies for advertising and all that stuff the GPU physics might be ready first while the PPU lies dorment waiting for its time. when ppl understand how limited GPUs as physics processors are and how bad that is (except for teh "grafix" physics thats basicly all Ati/Nvidia will develop). the power with PhysX is in stuff you do not see, its all about game feeling and how stuff reacts and behave. its not about big explosions but the things behind them. Grafix and PPUs handle big explosions, PhysX and PPUs handle how realistic the stuff around/in the explosion behaves and reacts realisticly. a physics GPU will "smooth" out explosions with grafix and make it look realistic the grafix way(cheap and simple trix etc), while PhysX makes it behave realistic from the code. thats IMO ofcouse and we will see how it turns out.. in 1+year i might add :twisted:

 

[twile]

there are alot of physics standards, PhysX is one that takes it to hardware form for games, theres alot of software standards too, and there will be other standards using GPUs, theres also tons of professional standards used by nasa and other companies, in car simulations, crash tests stuff etc.. the tools are there, and the dev-kits are there now for game devs.
theres 60+ games comming in the first wave with PhysX supported games, they will have minimal support and most likely only "eye-candy" use of the PPU thats barely noticible.

I wasn't making fun of the fact that they don't have any good demos of PhysX out yet, I know they've got a bunch of titles in the works which should take advantage of the hardware somehow. I was merely saying that before they start on the AI, they'll probably want to get around to the Physics stuff first.

 

[SuperG]

AI isn't off topic. Like Joint task force is also implementing PhysX but have a choice to do. EffectPhysics vs Gameplay PhysX.
The run into a AI Pathing problem to take debri into acount.

I guess physX has a very big impact on AI pathing. Gives AI programmer/scripter a lots of head aches.

Fix pathing doesn't work in its current state. it must adapt to blokkings and new path made by demolishing walls.

 

[twile]

AI isn't off topic. Like Joint task force is also implementing PhysX but have a choice to do. EffectPhysics vs Gameplay PhysX.
The run into a AI Pathing problem to take debri into acount.

I guess physX has a very big impact on AI pathing. Gives AI programmer/scripter a lots of head aches.

Fix pathing doesn't work in its current state. it must adapt to blokkings and new path made by demolishing walls.

Indeed. See, the defining difference bteween Effects and Gameplay physics are that Gameplay physics change the way you play the game. If your massive amount of physics-driven objects and particles don't change gameplay, they're just Effects physics. If shrapnel flies at you and does nothing, it's effects. If it damages you, then yay, it's Gameplay. Now you have to dodge shrapnel and pebbles from explosions, and the game dynamics are changed. If pushing objects around is just for visual pleasure, then it's effects, but if by pushing them around you can make blockades or weapons, conceal yourself, reach new areas etc, then again the game dynamics are changed, and it's a gameplay physics change. Now, this relates to AI in that the AI has to cope with it. Consider if you added a weapon to a game and the AI had no way of countering it. You add a flamethrower or a shotgun and the AI doesn't go out of its way to avoid getting hit, and when it does get hit, all it "knows" is that it's been damaged, not by what, and it doesn't try to get out of range or behind cover. After a few minutes of laughing at confused enemies running around in flames, players would probably complain that the AI couldn't cope with the weapon. Similarly, if you add in physics weapons, shrapnel, destroyable materials and the like, the AI has to be able to realistically interact with all of these things. If you can slip on liquid spills, the AI has to be able to recognize them (or not recognize them, if lighting is poor) and avoid them. If shrapnel can hurt, it has to be able to analyze incoming objects for their destructive potential, and change its path to avoid as many as possible while not doing stupid things like jumping off cliffs or onto mines. If traps can be set the AI has to be able to set and avoid traps. If objects can be rearranged, the AI has to be able to "remember" how they were placed so if a guard is on patrol and a bunch of crates have been moved or objects on a desk rearranged, it can notice them. Properly adding physics support isn't just about adding friction, gravity, inertia and breakability; it should be backed up with the other features, such as revamped AI, to take advantage of it. Again I have no idea how taxing current AI is on a system, but the changes imposed by having a dynamically changing and realistic environment would have to include more complicated AI.

 

[DrBlofeld]

Indeed. See, the defining difference bteween Effects and Gameplay physics are that Gameplay physics change the way you play the game. If your massive amount of physics-driven objects and particles don't change gameplay, they're just Effects physics. If shrapnel flies at you and does nothing, it's effects. If it damages you, then yay, it's Gameplay.

Right. This exact idea makes all the difference for the AI engine. This difference was explained in the article as well. You can link to it or just copy-paste it instead.

Again I have no idea how taxing current AI is on a system,

Absolutely Correct. The AI for reacting to new types and more numerous objects, "Teddy Bears and Pencils" as you seem to focus on, is negligible for AI. You can categorize objects into into two or three threat levels. The bottom line is the user perception. The graphics card is the one that gets taxed in this situation, not the AI.

SuperG brings up an excellent point with regards to the new path planning required when existing pathways are blocked. This can be a huge CPU hog for AI, especially if it happens numerous times. An AI character can seek cover behind an suitable object next to them, even if all the shrapnel and teddy bears are flying towards him. It does not need to path around it to get to the table as it does a suitable job for the user perception. However, if pathways are blocked, the AI will fail very noticeably. This happens in Oblivion. If you stand at a high point in the map, the AI does a poor job in planning a path to your character and the player can use this loop hole to make the game easier.

 

[Empty_V0id]

The Ageia PhysX Accelerator failed from the day it was announced. Unless you want to pay 200 something dollars for a card that will only handle about twenty games by the end of 2007 and only play GRAW for the rest of your life with the eye popping physics. So, what things crash and bend better and the fire looks more realistic and the particles have more detail and the shell casings bounce, no matter what the product will suck.

 

[twile]

I briefly redefined gameplay and effects physics and came up with my own examples so I could demonstrate gameplay physics that might have an impact on AI. HL2's "Gravity Gun" was gameplay physics in that you used it for a great deal of problem solving and creating makeshift weapons (my favorite was always a cinder block to the head, or a car engine to the chest) but it doesn't demonstrate gameplay physics that are particularly taxing on the computer for the physics calculations or the AI response. Since in previous posts people have said things along the lines of "read the article and STFU n00b" I'm tending to lean in the direction of completeness, and rather than dig through the article for something I know, I prefer to restate it in conjunction with an example.

Destructable objects could lead to a lot of interesting potential solutions for both the player and the AI. If somebody is on top of a stone pillar and you can't reach or shoot at them, tipping it over is an option that players and NPCs should be able to consider. Or if they're hiding in a tree, simply chopping it down could be a suitable solution. Certainly a step up from the AI wandering around in hopes of finding a path straight up.

 

[SuperG]

"obviously these are pre setup destructions " i think if this is the case they must pre setup everthing that is destructable. For a full destructable enviorment that means everything in the map.

Well If you wondering what headaches destructable wall bring. It could be a bit clearer with some examples.

RedFraction
Soldners.

In redfraction only a few walls where destructable. And very recognisable. Because
It had a homogenious texture so it would represent also a homogenoius structure for simplicity. It means the point where the hole comes and new generated geometry and debri can take that same simplistic wall Texture. Its Destructable but only this simplistic lookin walls.
Forget Brick wall. Bumpmap textures plus the Bricks are like Red strong cells with more slightly weaker cement spacy to stick them together. The wall hole break point must follow bricks and texturing and goemetry generation off the wallhole get a whole lot more complex in generagion and texturing.

But this simplistic wall method has the feature of a hole where it's been hit. Thats cool.

So if you want to make all wall destroyable. This means coming with solutions for complex walls like Brick walls or walls with a specific complex texture or material structure.
Or a map consisting off only some flavors of simplistic walls.
A dull map.
Volume textures and Voxel Texturing comes in mind as a solution . Sure It gives develeopers and level designer some head aches and more work to do.
Making wall consist out of more Objects with lot of Fix brake point joints. Like a wooden wall and brick wall making the Bricks Large.
Could help.


Soldners
In this game more is destructable like terrain to.
1 ) But just like Redfraction walls in this case the terrain looks simplistic. Not detaild. So You get a hole or crater in homogenious terrain. but trees near that crater will fall down. Nice. Makes you think of crysis jungle but without the fancy graphics and jungle.
2 ) Buildings are destroyable and every wall of those buildings. But very FIX doesn't matter where you hit with the tank shell. The hole is in the middel of that wall section. It looks so unbelievable but well its better then none. And debri fanish in explosion?

I think if they gone implement Destructable enviorment on a large scale it would be more of the Fix type.

Fix destructable You get the effect of Joint strike fighter. wich means The map can keep the more complex walls. But will be fixed destructable. But doing right not like that fix hole but falls appart in fix parts, it could be not that much disturbing. But could still be way better.

Problem is generating each damage stage of a complex wall believable. Needs a lot of thinkin and may be a new technik to do it in large scale.
In realtime
For level designers and content designers. All Fix states of those structures must be design. The wall as a new part. And each of its damages stages. Also the parts it breakes in must be design. Off course a game that uses this fix destructable method on a large scale, would have tools to break appart structures so it would be less work. The damage would then be generated in the design tool for a fix state.
But still more work then without it.

As for the impact on level design.

On the fly generated would shift a lot of the headache of the destructable feature to the Game engine design and feature set. If destruction is generated on the fly. Problem would be Textures for volume materials.
Also the ammount of materials used in a map.
Just keep away of Bricks but take metal for walls could be thinner but they got a bend and tear damange. So a other isue comming up. But there is a PhysX steal bend demo.

I don't know the source engine. But if its fix. You must set the damage states and content for each destructable wall.
A level designer must give every Physx or destructable object is PhysX attributes but also it damage states.
Of course a lot of it could be the same fix damage model. But Maybe the put some variations. And if there are some or even a lot different materials. A lot more variation is needed thus more work to make a level or Map.

In the reality builder for Cellfactor you can change Prefabs into breakable objects. You import a Mesh and textures and make it destructuble to design it breakable parts out off the original object. this object can then set to be instanciated. mutiple times.

 

[SuperG]

Nice.
Yes the demolition kit makes it very fix. But only one central blast to that outer wall. I would like to see a hit in the corner of that wall. But yes its fix.
It was more a show case of the whole map with real ingame play.

But destrucable objects is not new its even very old. But only a few selected objects. those fuel tanks. So it is still on a low scale low detail and fix.

All walls look the same material and simplistic. Nice little map to let RF geomod loose on.

Well i'am not a modder but lookin in to the realitybuilder from the reality engine wich Cellfactor is used for.

Tried to put the flag in the play field.