• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Deformable Objects Video

thast pretty sweet...hopefully that stuff makes it into games soon...it will make it much more "real". This will carry over into deformable terrain and buildins right?
 
The barrels have high resolution meshes. Nothing massively new going on here. All that you are seeing are vertices moving according to physics calculations. Looks nice but not really practical as yet since object meshes have to be high resolution. Its a similar reason as to why we dont see displacement mapping in todays games even though the feature has been available for years. Higher resoultion meshes mean more strain on a gfx card regardless of wether a PPU is installed or not.

Now deformable objects would be seriously cool. But the extra strain on the gfx card is the main reason we dont see it in games. A single barrel normally would consist of about 50 polygons. For deformation you can multiply that by 10. Unless an engine can adjust polygon counts on the fly (tesselation along the poly's affected), or gfx cards recieve a MAJOR boost to processing power, we wont be seeing this implemented in any big way anytime soon.

The only real games i can see this being implemented easily is in racing sims. The car models in modern racing sims are seriously high res and would deform really well. Also guard rails in the scenery would not need many more poly's because unless you actually park and inspect them you wont notice the incosistencies, therefore keeping fps nice and high.

It'll be a good 2 or 3 years before we see this feature incorporated in games.
 
The guy controlling the "hand of god"... Imagine having that job. I'd love that job.
 
Skirrow said:
The only real games i can see this being implemented easily is in racing sims. The car models in modern racing sims are seriously high res and would deform really well.
Click to expand...

The cars would still probably end up being many times as complex as they are now, as the chassis and various internal items would have to be modelled properly. If the front bumper gets torn off, you can't suddenly be looking at the driver's feet and an empty shell of a car. hmm.... sure would be nice if there was a new Carmageddon game released. :)

Reminds me of a Microsoft demo videos for XNA with the crash tests:
Youtube - MS XNA Crash Test1 Video
Youtube - MS XNA Crash Test2 Video
 
THIS is what I'm waiting for from a physics chip. Give me clothes/grass/fire/etc that deform in real time, THEN I will jump on the physics bandwagon.

I remember that QCON 2 years ago the Ageia rep said the 2nd gen of Physics card would be able to do just that.
 
Skirrow said:
The barrels have high resolution meshes. Nothing massively new going on here. All that you are seeing are vertices moving according to physics calculations. Looks nice but not really practical as yet since object meshes have to be high resolution. Its a similar reason as to why we dont see displacement mapping in todays games even though the feature has been available for years. Higher resoultion meshes mean more strain on a gfx card regardless of wether a PPU is installed or not.

Now deformable objects would be seriously cool. But the extra strain on the gfx card is the main reason we dont see it in games. A single barrel normally would consist of about 50 polygons. For deformation you can multiply that by 10. Unless an engine can adjust polygon counts on the fly (tesselation along the poly's affected), or gfx cards recieve a MAJOR boost to processing power, we wont be seeing this implemented in any big way anytime soon.

The only real games i can see this being implemented easily is in racing sims. The car models in modern racing sims are seriously high res and would deform really well. Also guard rails in the scenery would not need many more poly's because unless you actually park and inspect them you wont notice the incosistencies, therefore keeping fps nice and high.

It'll be a good 2 or 3 years before we see this feature incorporated in games.
Click to expand...

wow this cat knows his shizot.

mal
 
I can't wait to see the future of physics in games. Imagine playing Call of Duty and trudging through mud, having puddles realistically gather on the ground in low spots... or create a stream cutting its way to low ground through the dirt. Imagine being stuck down in a trench while tanks crush the ground over head... or simply leaving massive tracks dug in the earth.

Imagine getting shot through a shirt and actually not taking a hit due to killer hit detection based on physics. Breaking bones and having your body (or any other) dynamically change due to the pain and have to drag your leg while you make your way to a health station or medic.

Or in a game like Oblivion, imagine bashing your opponents armor with a mace or axe and having the deformation calculated on the armor... where after wards the armor needs to be fixed at a black smith due to damage. Swords that crack and break due to hitting solid objects.

I'm pretty hesitant to buy into PPU's until these things can be put into games... bring the level of immersion to epic proportions and I will buy a card, because we all want to be deeper immersed, no?
 
see I think that car racing games the the best option for a "killer app" that Ageia will get anytime soon. Most of the physics involved in car racing is "felt" more than seen. I don't need hugely deformable objects, nor do I need to see the tires deform while playing NFS or similar, but I need to feel real-world physics!

I am still waiting for a HL2 patch that will finally deal with the "jumping into cabinets" thing.
 
Nate_MachV said:
I can't wait to see the future of physics in games. Imagine playing Call of Duty and trudging through mud, having puddles realistically gather on the ground in low spots... or create a stream cutting its way to low ground through the dirt. Imagine being stuck down in a trench while tanks crush the ground over head... or simply leaving massive tracks dug in the earth.

Imagine getting shot through a shirt and actually not taking a hit due to killer hit detection based on physics. Breaking bones and having your body (or any other) dynamically change due to the pain and have to drag your leg while you make your way to a health station or medic.

Or in a game like Oblivion, imagine bashing your opponents armor with a mace or axe and having the deformation calculated on the armor... where after wards the armor needs to be fixed at a black smith due to damage. Swords that crack and break due to hitting solid objects.

I'm pretty hesitant to buy into PPU's until these things can be put into games... bring the level of immersion to epic proportions and I will buy a card, because we all want to be deeper immersed, no?
Click to expand...

Other than the first thing you mentioned (the water collecting in low spots etc.), all this could easily be implemented on the CPU. Those are very basic interactions, and wouldn't need the add-in card from Aegia.

After seeing the demos for the water from Havok, I'm sure your water scenario could work out on GPUs as well.

As for the video, while cool, I really just don't care anymore. These tech videos are getting ridiculous because all I care about is how they're implemented in games. While with new GFX cards we can see obvious improvements in graphical fidelity, the Physx and Havok solutions just need to be implemented so that we can play a game that is enhanced with the hardware accellerated physics, and not imagine how the power of 100,000 rolling boulders would improve things.
 
Yeah, in the case of puddles, this is much more efficiently implemented by analyzing the level topology for small bowls (which can hold water), factoring in the terrain permeability, and then putting a horizontal plane across the region with a water texture/shader.

This sort of stuff seems pretty trivial to do, so I'm surprised we haven't seen it yet in games. If I ever get some time off I might do a proof of concept.
 
Nate_MachV said:
I can't wait to see the future of physics in games. Imagine playing Call of Duty and trudging through mud, having puddles realistically gather on the ground in low spots... or create a stream cutting its way to low ground through the dirt.
Click to expand...

Proper fluid dynamics like that are a LONG way off, with or without dedicated hardware. If we had photo-real moving/flowing/reacting water even within 5-10 years, I'd be surprised. (Am thinking comparable to those scanline flowline vfx render reels.)
 
Skirrow said:
Unless an engine can adjust polygon counts on the fly (tesselation along the poly's affected)..
Click to expand...

Or you had some sort of vectorized object model. The physics card processes the interactions and the shape the resulting objects should be. Then the graphics card can then render those objects based on angles and curves instead of thousands of polygons. I'm not sure how feasible this is but it seems like it might be something that would indeed be enabled by a physics card, unlike really complex meshes everywhere which are difficult to deal with regardless.
 
jimmyb said:
Yeah, in the case of puddles, this is much more efficiently implemented by analyzing the level topology for small bowls (which can hold water), factoring in the terrain permeability, and then putting a horizontal plane across the region with a water texture/shader.
I disagree - this is one of the reasons why actual physics processing would make maps better / more interesting - just because we can simulate efficiently with a global water level and CPU processing doesn't mean we should - when physics processing can take over and actually add to the gameplay experience. The limitations you describe take away from the realism and ability to create interesting levels. I remember being frustrated when I was making maps for Desert Combat (BF1942) that water occurred only at one level - no lakes higher up in the mountains, or waterfalls, etc.

Getting a PPU into the gamestate gives us more flexibility than we have now, and makes realistic simuations of the flow of water or steam or smoke more likely to show up in games sooner.

The cool thing about this video (if I understand what's happening) is that the GPU only has to quantify the light & colors while the PPU determines what's bending / bouncing and the CPU monitors gamestate. Not something you generally see in an actual game.

The idea is to get away from "painting" the walls with bullet tags and actually being able to shoot simulated bullets and watch wooden walls splinter and burn, stone walls spall and crumble, and tanks actually be able to run over trees / fences / buildings - stuff which cannot happen now unless the programmers previously designed an animation for a particular sequence.
Click to expand...
Click to expand...
 
The key word in PPU technology being "Dynamic". Much like the advances in lighting we have achieved ala HDR. It's time to like was said before, begin constructing actual objects that have weight, react to the enviroment correctly and realistically... instead of being a static object with scripted attributes.

What would happen if I rolled this boulder down this hill towards that hut made of sticks?
the boulder would dig a path through the grass rolling, based on the shape and weight would pick up speed... either hit the hut and crush it into individual branches cracking some under the weight or missing it completly

What would happen if I redirected this stream by piling wood and rocks funneling it towards that village?
over time water would rise due to rain fall and the stream itself, causing flooding to the village and creating a whole other group of events to unfold.

The possiblilties while intensive by current technologies, I'm sure with the addition of either a Ageia PPU, making SLI/CF mainstream or the soon to be reality of Quad Core processors.

Yes it's a big step, and no I do not expect to see these implications for a while. But there is a world of possibilties. Maybe once we hit that brick wall of Photo Realism in games (which seems to be looming ever so closely) developers can devote their teams to implicating real world physics, and have them be more than a gimmick breathing new life into gaming.
 
Very nice! I just can't wait til your playing like an FPS and someone gets shot with like an RPG and you see their leg/arms/head/chest go flying up in the air with all of this blood and guts! Yeah i know i'm wierd! :p
 
DarkSideA8 said:
I disagree - this is one of the reasons why actual physics processing would make maps better / more interesting - just because we can simulate efficiently with a global water level and CPU processing doesn't mean we should - when physics processing can take over and actually add to the gameplay experience. The limitations you describe take away from the realism and ability to create interesting levels. I remember being frustrated when I was making maps for Desert Combat (BF1942) that water occurred only at one level - no lakes higher up in the mountains, or waterfalls, etc.
Click to expand...
The method I described allows for puddles to occur at varying altitudes, in conditions where you expect puddling to occur. It could also very easily be modified so that the puddles occur at different rates, their depth varies with time, etc.

I agree with your general point though, in that it is better to do things more generally, but we are a ways off realistic realtime fluid simulation, so this would be a fairly effective technique for the next 10 years or so.
 
Skirrow said:
The barrels have high resolution meshes. Nothing massively new going on here. All that you are seeing are vertices moving according to physics calculations. Looks nice but not really practical as yet since object meshes have to be high resolution. Its a similar reason as to why we dont see displacement mapping in todays games even though the feature has been available for years. Higher resoultion meshes mean more strain on a gfx card regardless of wether a PPU is installed or not.

Now deformable objects would be seriously cool. But the extra strain on the gfx card is the main reason we dont see it in games. A single barrel normally would consist of about 50 polygons. For deformation you can multiply that by 10. Unless an engine can adjust polygon counts on the fly (tesselation along the poly's affected), or gfx cards recieve a MAJOR boost to processing power, we wont be seeing this implemented in any big way anytime soon.

The only real games i can see this being implemented easily is in racing sims. The car models in modern racing sims are seriously high res and would deform really well. Also guard rails in the scenery would not need many more poly's because unless you actually park and inspect them you wont notice the incosistencies, therefore keeping fps nice and high.

It'll be a good 2 or 3 years before we see this feature incorporated in games.
Click to expand...
You have never played red faction have you? Okay i see wheere you are coming from but it is possible without high rez mesh. Now don't get me wrong it's not the same thing as destroying metal bars with bening plyers in a game...But it's close. http://en.wikipedia.org/wiki/Red_Faction


GeoMod



One of the main "selling points" of Red Faction was the game engine's GeoMod technology, which allowed players to significantly alter and otherwise affect the surrounding terrain and buildings, either with explosives or otherwise. Past games had allowed limited alteration of the environment through scripted means, or special triggered areas which would add or remove certain predefined level elements. Red Faction was the first game to allow players relatively limitless level-altering possibilities. In fact, many of the areas in the first few levels, such as a hidden underground lake, serve little purpose other than showcasing the capabilities of the GeoMod technology.

GeoMod allows the player to alter (and destroy) the environment so significantly because of the way it designates altered (or "GeoModded") areas. Whereas other game engines would have to modify the shape of the altered object to create a similar effect, GeoMod creates special objects which represent empty space. When, for example, an explosion blows a hole in a rock wall, the GeoMod technology creates an "empty space" object which is the approximate size and shape of the explosion. The training level uses a much larger "empty space" object than the rocket launcher is able to produce in other circumstances, simply because tunnelling is required to complete part of the level.

The use of these "empty space" objects is not limited to in-game environment alteration. Some of the objects in the game, such as the wastebaskets found in certain office areas, are stored as one solid shape with another "empty space" shape defined within it.

Geomod has become popular in multiplayer as players tend to build "tunnels" with explosive weapons. This is done by repeatedly shooting a wall with the rocket launcher or using the remote detonators (or occasionally the fusion rocket launcher and grenades). Some players build tunnels for an advantage because they could hide in them and use a sniper to pick off enemies. In CTF games, players build tunnels as a shortcut into an enemy base to gain an advantage in capturing their flags.

One of the main criticisms of the game was inconsistency in the use of GeoMod, mostly owing to the design limitations of linear first-person shooter. Where the game needed to halt the player's progress, locked doors and appropriate nearby walls were invulnerable to attack; players were not allowed to circumvent the obstruction using their own initiative, except in a few cases presumed to be glitches. Other, less justifiable inconsistencies in the use of GeoMod could be found in the game; some players noted how it was possible to blow holes in the rocky caverns of the mines, yet at the same time the partitions in office cubicles were completely indestructible. Despite these and other GeoMod issues in the singleplayer game, critics were essentially unanimous in their praise of the system in multiplayer matches.

In the extras menu there was also a small level called "Glass House" which allowed you to create endless tunnels and caverns and really experiment with the Geo-Mod technology more than you could in any other part of the game.
Click to expand...

The premiere feature of the game was its "GeoMod" technology which allowed the player to destroy the geography in the game. For example, instead of opening a door, a player could blast through the rock surrounding the door.
Click to expand...
 
Red Faction destructable geometry was all preplanned and only made in layers it wasn't as random as people might think. Also the dents on the barrells shown are really accurate I think it would be pointless to use lower res models because then you just get weird shapes out of the cylinders and that isn't to realistic now is it?
 
The destructible geometry system used in Red Faction makes it necessary to determine the deformations beforehand and design them specifically, whereas in this demo the deformations are calculated in real-time. Due to the fact that all the deformations are calculated specifically depending on the actual interactions between objects, it's impossible to predetermine the exact deformations required for every possible interaction between objects, which is why the Red Faction implementation of destructible geometry is not suitable for use in applications which aim for the most precise representation of actual physics as possible.
 
The mesh doesn't need to be high-res until you want to *display* the deformation. You calculate the deformation based on the equations which define the shape of the object. The mesh is determined after that based on other things(desired polygon count, for example).
 
I remember this one off road racing game called 1nsane that had deformable vehicles. Not sure how realistic it was though. ;o)
 
Lord of Shadows said:
I remember this one off road racing game called 1nsane that had deformable vehicles. Not sure how realistic it was though. ;o)
Click to expand...

1nsane uses scripted damage. I own it, and still play it. It is also the only offroading game of it's kind, and has an amazing ability to simulate rock crawling.
 
psychoace said:
Red Faction destructable geometry was all preplanned and only made in layers it wasn't as random as people might think. Also the dents on the barrells shown are really accurate I think it would be pointless to use lower res models because then you just get weird shapes out of the cylinders and that isn't to realistic now is it?
Click to expand...

I seem to remember John Carmack suggesting a similar concern with destructible environments several years ago. Basically where you have a really pretty looking wall, with high resolution textures, normal maps, bump maps, etc and then a really blocky/ugly looking hole in it. The deformation definitely needs to be as high quality as the rest of the static environment, otherwise it just ends up clashing terribly.
 
Enthusiasts spend hundreds of dollars on GPU's, CPU's/MOBOs, RAM and monitors mostly to avoid frame chatter during gameplay. Programmers have taken advantage of hardware improvements to make their games more visually appealing - better lighting, textures, effects and higher polygon characters and objects. In some cases, AI scripting has also greatly improved (e.g. FarCry). The purpose has always been "added realism."

These programming improvements have been so successful that each generation of games actually taxes the new hardware generation's ability to crunch numbers effectively, meaning that framerates continue to be a problem for gamers.

Despite visual improvements, the game worlds typically remain static. Far Cry had a couple of areas wher if you tossed a grenade just right you could get a nifty "hole" in the ground. It didn't do anything to affect gameplay. Other games have a rope you can swing on, a rock you can move, a mesh you can destroy/deform, etc. - all with pre-programmed animations. These things only have minor, preplanned effects on gameplay. Adding physics to the equation also slows things down further.

Although the story telling and environments of today's games are far better than we experienced in the early days of Castle Wolfenstein and Doom (although they were amazing for their time), most games generally play the same (admittedly we're beyond cubes). If you hide behind a mesh of any type you're safe from the enemies that charge toward you, until they get a direct line of sight. The exceptions to this rule are minimal. Those that do have meshes that change currently rely on CPU number crunching to change the gamestate.

All of that said, I don't understand why some "enthusiasts" object to PPU's. The PPU is the first new piece of hardware designed (as the GPU was) to offload work from the CPU to improve overall game performance. It should allow programmers to let their meshes deform, break, or change in response to an actual random game event, rather than only through a pre-programmed occurrence (ala rocket launcher to the "weird" wall in Duke Nukem).

The biggest gripe I read on these boards is, '[tuL33t4U / wrote] "my current favorite game, 'Pancho's Obscure Elephant Adventure' doesn't need physics acceleration, therefore I think its stoopid for Ageia to make a PPU"

That said, why not embrace the technology? In some games, smoke looks little removed the old "spraypaint" effect seen in "Paint", wooden fences stop tanks, and few objects have realistic simulated weight.

If this demo is any indication of the future of gaming - and something we might see throughout the entire gamespace - I, for one, say 'full speed ahead, and damn the torpedoes!"
 
psychoace said:
Red Faction destructable geometry was all preplanned...
Click to expand...
Zero82z said:
The destructible geometry system used in Red Faction makes it necessary to determine the deformations beforehand and design them specifically....
Click to expand...
...Did either of you actually read that Wikipedia extract?

Red Faction does not use the traditional method of subdividing sections of the environment into destructible pieces. It just places an "empty space" object at the site of the explosion. Nothing's actually been destroyed - it may look like a hole and act like a hole, but the wall's still there underneath it. The only thing that's determined beforehand is which areas are destructible and which aren't.
 
LuminaryJanitor said:
...Did either of you actually read that Wikipedia extract?

Red Faction does not use the traditional method of subdividing sections of the environment into destructible pieces. It just places an "empty space" object at the site of the explosion. Nothing's actually been destroyed - it may look like a hole and act like a hole, but the wall's still there underneath it. The only thing that's determined beforehand is which areas are destructible and which aren't.
Click to expand...
By deformations, I was talking about those empty space objects, which serve the purpose of deformations. The point I was trying to make is that the shape and dimensions of the actual empty space objects must be determined beforehand, and it would be impossible to create an object for every possible scenario. Therefore, that implementation is inherently flawed in terms of being able to achieve absolute realism.
 
wera said:
You have never played red faction have you? Okay i see wheere you are coming from but it is possible without high rez mesh. Now don't get me wrong it's not the same thing as destroying metal bars with bening plyers in a game...But it's close. http://en.wikipedia.org/wiki/Red_Faction
Click to expand...

Yeah i've played the Red Faction games. They used boolean subtraction to take chunks out of walls. Thats a completely different approach. The same approach wouldn't work when simulating dents in metals and such since its not an efficient or realistic way of achieving the effect. Hence why all metals in Redfaction were not affected by explosions, only rocks.

Boolean subtraction works something like this.

Say you have an object (the boolean primary), e.g a cube, and you want to simulate a rocket hitting it. You would assign a boolean subtractor to the rocket model. (The boolean secondary such as a low poly sphere, with a noise modifier, an algorythm that randomises the position of vertices of an object)

When the rocket model hits the cube, boolean subtraction uses the shape of the sphere to 'cut out' a hole in the cube that follows the contour of the sphere. What this does is create a lot of extra polygons on the original model.

The problem is this wouldnt work with the barrells in the video. For one, it would totally screw up the texture co-ordinates, since there would be extra polygons to cope with. Redfaction only uses a seemless rock texture to fill these extra poly's so you cant notice it (though i suppose you could 'project' the texture along the barrels affected axis, i forget the proper term since its years since i did a 3D project).

However, with physics deformation, you are only deforming the original geometry. The only things that are changing are the vertice co-ordinates. What this means is that you can accurately simulate how once vertices affects the other. Like when you crush a coke can in the middle the ends bend toward the middle too. The pulling forces simulated by the physics engine.

That being said. The deformation in the video is impressive and i'd love to see it in games.
Combined with boolean subtraction, you could have some seriously realistic destructive enviroments.

Imagine firing a rocket at a steel reinforced pillar on a watch tower and the concrete is broken away. You then watch the steel underneath buckle under the weight. Or damaging a bridge with C4 and watching the beams twist and warp when a convoy of tanks goes across finally collapsing.

(btw, i studied animation for 3 years, using 3DSmax and Havok, wanting a career change. Didnt work out :(, cost me just under £2000 for max 3, then they announced a cheap student edition for max4 grrrrrrrrr :mad: )
 
Off Topic: The problems developers have is that how destructive can you make an enviroment without ruining the game. Make it too destructable and it'll be like having a chainsaw in a hedge maze. Just cutting a straight line to the end. That would take the fun out of it

On Topic: You could get around this by having steel walls you can only dent (i.e a bomb shelter type place, or battleship interior) which deformation would enhance. This would give the false impression that there is the possibility of escaping the confines of the level. You could also use deformation to smash open doors realistically, rather than having pre-modelled damage. Though you'd have to be careful as too much damage could jam the door. Which is something else developers would have to overcome.

Imagine throwing a grenade down a corridor only to find it has damaged a door and jammed it in a doorway thanks to the realistic deformation.
 
the destruction of the game world depends on how much heavy ammo and weapons there are.

Like a Tank have 40 shelss spit in 2 types. General purpouse and Anti armor.
RPG7 4 reloads.
Javanyl 1 tube per infantery. And one fire unit per sqaud.
Frag Nades aren't destructive. more frag producing. However they come in some types.
TermicGranade. To destroy equipmenst. Like a downed Balckhawk in enemey teritory.
Few mates in a squad would have one.

To go mayhem with geomod youn need a cheat. For enough ammo for firering tunnels or completly destroy a full wall.

In real Gameplay it's more, use your heavy weapons wisely because there is a Ammo restriction.

Oh and I palyed RedFraction to.
Deformal Physics vs Geomod is a world of difference.
Firstly a PPU could hardware accelerate a more realistic and large scale goemod engine. But Deformal Objects is a other Physics feature wich could be used besides Geomod ish implementation.

The Break vs Bend materials.
 
Skirrow said:
Imagine throwing a grenade down a corridor only to find it has damaged a door and jammed it in a doorway thanks to the realistic deformation.
Click to expand...

That would suck :)
 
Skirrow said:
Imagine throwing a grenade down a corridor only to find it has damaged a door and jammed it in a doorway thanks to the realistic deformation.
Click to expand...

Well, if the door could be realistically deformed to such a degree you could just throw another grenade down the hall and blow what's left of it to peices. Or kick it in.

Or perhaps you should just pay closer attention as to where you're leaving your grenades.
 
Skirrow said:
The barrels have high resolution meshes. Nothing massively new going on here. All that you are seeing are vertices moving according to physics calculations. Looks nice but not really practical as yet since object meshes have to be high resolution. Its a similar reason as to why we dont see displacement mapping in todays games even though the feature has been available for years. Higher resoultion meshes mean more strain on a gfx card regardless of wether a PPU is installed or not.

Now deformable objects would be seriously cool. But the extra strain on the gfx card is the main reason we dont see it in games. A single barrel normally would consist of about 50 polygons. For deformation you can multiply that by 10. Unless an engine can adjust polygon counts on the fly (tesselation along the poly's affected), or gfx cards recieve a MAJOR boost to processing power, we wont be seeing this implemented in any big way anytime soon.

The only real games i can see this being implemented easily is in racing sims. The car models in modern racing sims are seriously high res and would deform really well. Also guard rails in the scenery would not need many more poly's because unless you actually park and inspect them you wont notice the incosistencies, therefore keeping fps nice and high.

It'll be a good 2 or 3 years before we see this feature incorporated in games.
Click to expand...

translation: blah blah extreme pessimism blah blah

Nothing new? The physics calculations are what's NEW. Nobody's done it to that extrme before, that's what.
 
jimmyb said:
The method I described allows for puddles to occur at varying altitudes, in conditions where you expect puddling to occur. It could also very easily be modified so that the puddles occur at different rates, their depth varies with time, etc.

I agree with your general point though, in that it is better to do things more generally, but we are a ways off realistic realtime fluid simulation, so this would be a fairly effective technique for the next 10 years or so.
Click to expand...

That kind of seems counter-intuitive. Sure we can make things look a certain way, but the whole point of this post and the whole point of physics simulation is to make things to ACT not look.

I think we might even need to take a step off the beaten path in order to approach the kind of simulation that becoming more in demand.

Things are made of tiny particles in our world right? Bonded together with various degrees of electromagnetic force? The best system to accurately simulate "real world" types of matter would probably be Voxels. Now we're going to have to start from the ground up but I think it is worth it.

FEAST YOU EYES.
http://www.realmatter.com/
I think this is based on voxels, if not it at least shows that some of the things your guys are talking about is possible. It's quite exciting actually.
No shotguns, but plenty of Zombie anger management.
 
Skirrow said:
The barrels have high resolution meshes. Nothing massively new going on here. All that you are seeing are vertices moving according to physics calculations. Looks nice but not really practical as yet since object meshes have to be high resolution. Its a similar reason as to why we dont see displacement mapping in todays games even though the feature has been available for years. Higher resoultion meshes mean more strain on a gfx card regardless of wether a PPU is installed or not.

Now deformable objects would be seriously cool. But the extra strain on the gfx card is the main reason we dont see it in games. A single barrel normally would consist of about 50 polygons. For deformation you can multiply that by 10. Unless an engine can adjust polygon counts on the fly (tesselation along the poly's affected), or gfx cards recieve a MAJOR boost to processing power, we wont be seeing this implemented in any big way anytime soon.

The only real games i can see this being implemented easily is in racing sims. The car models in modern racing sims are seriously high res and would deform really well. Also guard rails in the scenery would not need many more poly's because unless you actually park and inspect them you wont notice the incosistencies, therefore keeping fps nice and high.

It'll be a good 2 or 3 years before we see this feature incorporated in games.
Click to expand...

I totally agree, if you don't have crap load of polygons/vertices on each deformable mesh, than the part that is deformed will look like a pushed in triangle or any other polygonal shape, like so many games in the past, which was more than likely preset damage.
but i do think that with the latest hardware, it is very possible to run this at decent speeds, in the case of high detail, depending on how much detail, and how many high polygon deformable meshes are in the game at one time.
.
 
You must log in or register to reply here.
Back
Top