A New Way of Looking at Damage

[Kodiak]

I think Mistef was getting at the fact that specific, location-based damage could be generated, not from pre-set levels of damage but from ones calculated specifically for each unique impact, could be achieved. He's describing bringing things up a notch or two in complexity. This may indeed be out there today. I haven't played Crysis or any of the newer shooters (my computer and wallet can't handle those anymore), so I don't know if anything like this is in effect right now.

So why would such features be 5-10 years down the way? That's a tremendous amount of time in terms of capability development. That's the same sort of spread as between Pong and GATO (OK, more old-school references), except with the potential for an even more dramatic change in scope, given the exponential improvement in processor power over this time frame.

It brings me back to my original question: what would it take to be able to pull this off?

[toyoka]

A lot of time, knowledge, enthusiasm and Hot Stuffs.

But in all seriousness, It would take quite a while to program all the different combinations of effects and such and I can't imagine how much time that will take, not to mention the game itself and the technology to put that game in the hands of "causal"(ie. pro/extreme in this case) gamers of the market.

[Adrir]

Originally Posted by Kodiak

I think Mistef was getting at the fact that specific, location-based damage could be generated, not from pre-set levels of damage but from ones calculated specifically for each unique impact, could be achieved. He's describing bringing things up a notch or two in complexity. This may indeed be out there today. I haven't played Crysis or any of the newer shooters (my computer and wallet can't handle those anymore), so I don't know if anything like this is in effect right now.

Hit Zones are incredibly easy to implement from a programming and modelling perspective. You simply associate multiple collision bounds with a single entity, each with different properties that enable the calculation of different damage values.

[Protector one]

Originally Posted by Adrir

Hit Zones are incredibly easy to implement from a programming and modelling perspective. You simply associate multiple collision bounds with a single entity, each with different properties that enable the calculation of different damage values.

And it has been done since at least '97. Goldeneye, anyone?

[Kodiak]

Agreed, however, these are broken down into major components of a model.

This 'Dvoxel' idea essentially turrns every single 'atom' making up an object or a character into a hit zone that can be affected on the fly.

So really, it's only an evolution of the hit zone concept.

[Protector one]

Originally Posted by Kodiak

This 'Dvoxel' idea essentially turrns every single 'atom' making up an object or a character into a hit zone that can be affected on the fly.

So really, it's only an evolution of the hit zone concept.

Wouldn't it be more work to designate the damage potential for every 'dvoxel' in a volume, than to just designate it for a few hit zones? For example, in a typical FPS, dvoxels in a characters' head would have to have a lower 'death-threshold' than dvoxels in the characters' hand. In classical damage models, you would probably just have a head-volume and a hand-volume. With dvoxels, you'll easily end up with many voxels (given the size of your voxels), even for volumes as small as a hand.
Right?

[Kodiak]

No, the idea would be to be able to assign a material to the dvoxel.

You build your model, and as you go along creating meshes, etc, you assign the material to the various bits, for example: bone, muscle, fat, tendon, soft / rigid internal organ, etc. The challenge is that now you have to build up the entire anatomy (or engineering for objects) of objects on screen. You could either assign this to the model/group/layer the same way you assigned or mapped textures. We spend all kinds of effort drawing bump, displacement, specular, reflection maps and the like; or you could define the component's material during the 3D modelling phase, this would simply be one more step in the creation process. The information encoded with the material would then allow the engine to figure out the rest on its own.

The modelling could definitely be simplified (i.e. you don't include every bone in the inner ear, every tooth in the mouth, etc.), but much of the ground work is already being done in high end animation anyway. Look at muscule animations done for movies these days. I think King Kong made extensive use of the process. For humans, there are plenty of references out there already. A quick look through the anatomy database at Turbosquid alone shows that this level of detail is frequently dealt with by modellers.

It would get interesting modelling totally imaginary creatures. Maybe biologists, surgeons, veterinarians, mechanical engineers and the like would start getting involved in game production. Could you imagine designing the anatomy for a dragon?

You are right with regards to the number of dvoxels even in a small object such as a hand...especially given the requirements for highly detailed graphics...maybe this is a 16-core, 10 TB of RAM and ten linked GPU issue!

[ronnoc10]

Well, the Xbox launch game Deathrow simulated a simplified skeletal system in all of it's characters, and Star Wars: The Force Unleashed simulates a range of materials, and your explanation reminds me of it's early tech demos, so I wouldn't be surprised it we see something like that soon.

[Protector one]

Originally Posted by Kodiak

No, the idea would be to be able to assign a material to the dvoxel.

You build your model, and as you go along creating meshes, etc, you assign the material to the various bits, for example: bone, muscle, fat, tendon, soft / rigid internal organ, etc.

Well, this would only partially solve the problem of relative damage, I think. Damage to bone in the hand should have a different effect than damage to a leg bone. This applies to different muscles as well.
Or do you suggest that subsets of dvoxels are to be used? For example: [voxel1,bone,femur], [voxel2,bone,skull], [voxel3,muscle,left_forearm_tendon], etc.?

[Kodiak]

I think it could be kept relatively simple without breaking down to specific bone types.

The reason hand bones are more fragile then, say, a femur is mainly because of the thickness and arrangement of the bone itself. A smaller quantity of bone, particularly in a thin sheet, such as in the skull, is therefore easier to penetrate or break than a thick, tubular arrangement.

This would lead to interesting game play even for FPS: why bother putting 10 lbs of lead into the beastie's thick armour-scaled hide when you can put a round through it's soft eyeball, thin eye socket bone and straight into the brain?

Of course, this can be convinvingly substituted by further subdividing existing hit zones, and probably for less effort...

As an aside, I happened to stumble across an awesome video at: http://http://pc.ign.com/dor/objects...ry_080708.html which looks as if some people are seriously attempting to add more life-like reactions to their creatures...the werewolves' 'rage level'...also check out the fear reaction to being shot at.

Very cool!