Mark Rein Interview @ Shack News

[Sis]

Making the bucket hang more realistically is incredibly unimaginative. What if the bucket was filled with water? When the arrow hits the bucket, it swings realistically, with water spilling over, coming to rest with a slow overflow until the water level is lower than the tilt. You could then walk up to it, take back your arrow, and have the water spill out more until it's again lower than the level of the unplugged hole. You then realize that it's not water but rather some magical substance and you just found the well. You dip the bucket into the well to get some of this potion to fill a few bottles, and when you fill the bucket and bring it back up, the substance again spills out through the hole, realistically spraying the side of the well on the way up. After you've filled up the bottles, you look to your feet and you notice a big puddle from the mess. Not a flat puddle, but one that has seeped into the cracks and followed the seams in the ground. While looking at this, you hear some large creature approaching your position, with large ripples appearing with each footstep, causing the water to move down the seams in a different pattern. Etc, etc, etc...

Physics can easily take the experience to the next level.

Are you suggesting that this degree of simulation is just not possible? That if the designers choose to take shortcuts with the simulation of water spilling out of the bucket, you would notice? "Man, I tilted that bucket at 45 degrees and given the viscosity of the magical substance, 14 ounces should have spilled out! And look, only two puddles formed on the ground. By my calculations, it should have been three."

What I'm suggesting is that a) this would still be a design decision, possibly enabled by enhanced technologies and b) if a designer wanted it in bad enough, they'd put it in. And c) approximate simulations of physics is often good enough, if not better, than real-world simulations.

And finally, any technology, such as CELL or Aegia's board, that allows for a designer's vision to be more accurately reflected in the final product is a Good Thing. We are not in disagreement on this point that better physics technologies is better in general.

 

[pc999]

IMO physics can bring fun in many ways from just blowing stuff, from puzzles (eg using materials to creat a structure/path to somehere), to create traps (eg put some wire and a grenade on a door or use the wire to make some book-shelf fall), open new paths (dont like the small open door just use a jeep to crunsh the big one, or like Crysys if you damage the foliage AI may detect it and track you down...

I wonder if we will be able to see all this kind of things in next gen...?...

 

[seismologist]

What I'm suggesting is that a) this would still be a design decision, possibly enabled by enhanced technologies and b) if a designer wanted it in bad enough, they'd put it in. And c) approximate simulations of physics is often good enough, if not better, than real-world simulations.

The idea is that realistic simulations will look better and behave more realistically. It's like real lighting vs. light maps.

 

[flick556]

In terms of water, current sims cannot handle realistically flowing and splashing water. In fact I think the biggest change the power will bring is physics sims that change and or create geometry. The performance hit for these types of effects has been really high. As for physics being subtle I don't think that is the case. If they where subtle we would not notice when they where missing or inaccurate. Yet physics is still the largest contributor to games not looking and/or behaving realistically. Things that look realistic or natural is more pleasing to the eye than artificial looking things, this has been studied. Without seeing the difference greatly increased physics can make you cannot not fully understand the impact they will have. Currently most geometry in games is static, going forward most of it should be based on simulation. Realistic destruction, animation, and shape should all be based on physics. Their is still allot of ground to cover in this area, next gen consoles are a step in the right direction.

For non realistic games they will still have a increase in cpu power that they can put to good use.

 

[Sis]

The idea is that realistic simulations will look better and behave more realistically. It's like real lighting vs. light maps.

That's like saying the goal of a painting is to be photorealistic. Realistic simulations is not the end, but a means to an end. We seem to be treating it incorrectly.

 

[weaksauce]

I look at Oblivion as an example of "good enough" physics: a wooden bucket was hanging from a rope, over a well. I shot an arrow into the bucket and the arrow stuck into the wood. This caused an imbalance in the weigth distribution, which caused the bucket to hang at an odd angle (once it came to rest from the impact of the arrow). Is there a way that some new technology will make that bucket hang at an even more realistic angle?

I remember playing hl1 and thought it was really cool to move around boxes and push them of edges, like how could it get better?

 

[BlueTsunami]

That's like saying the goal of a painting is to be photorealistic. Realistic simulations is not the end, but a means to an end. We seem to be treating it incorrectly.

Just put it like this then, instead of constricting developers to one path Physics adds another path they can take. Also, weaksauce above me has a good point.

In all honesty I don't think its a good thing to be contempt with what you've got (which you seem to be), developers should be striving for more. I wonder where games would be today if the majority of developers just shrugged and said its already good enough.

 

[Sis]

Just put it like this then, instead of constricting developers to one path Physics adds another path they can take. Also, weaksauce above me has a good point.

In all honesty I don't think its a good thing to be contempt with what you've got (which you seem to be), developers should be striving for more. I wonder where games would be today if the majority of developers just shrugged and said its already good enough.

I'm not saying that at all (and if you read all my posts on this, I've made that perfectly clear). I just think we should have some perspective on what physics-enabling technologies are all about. The same is true for AI: having true artificial intelligence in enemy AI might actually prove frustrating for the gamer, since the characters would behave in perhaps totally unpredictable ways. In the end, the game world is a simulation not of the real world, though it may behave like it, but a simulation of the designer's world. Anything that helps them simulate their world better--and it very well could be better physics implementations--is a Good Thing.

 

[BlueTsunami]

I'm not saying that at all (and if you read all my posts on this, I've made that perfectly clear). I just think we should have some perspective on what physics-enabling technologies are all about. The same is true for AI: having true artificial intelligence in enemy AI might actually prove frustrating for the gamer, since the characters would behave in perhaps totally unpredictable ways. In the end, the game world is a simulation not of the real world, though it may behave like it, but a simulation of the designer's world. Anything that helps them simulate their world better--and it very well could be better physics implementations--is a Good Thing.

I agree with you on the AI front. All in all it comes down to how the developers utilize these technologies. Be it Phyiscs, AI or anything else.

Also, Physics doesn't need to apply to everything. In all honesty, theres a bunch of genres that will get no benefits from heavy physics. Though, on the flip side you've got games that can be greatly enhanced by physics.

I feel that its a road a developer can take or choose to disregard (which is good). You've then got all the shades in between (using minimal physics or a using a large amount). It all depends on the developers and how they impliment it (like with AI).

One company and game your going to see being pointed to alot is Evolution with Motorstorm. Another would be Half Life 2. Unfortuneatley those are the two that are really known for their physics and come to mind. I point that out because there aren't really any games that have used physics in games to its full capacity. Until that day, I think its premature to say anything about physics in general (good or bad).

This generation (with PS3, Xbox360 and PPU's in PCs) I think we'll be seeing what physics can add to games. I'm personally excited over this technology.

 

[mckmas8808]

Well I don't agree with the A.I. part. Hell make it as smart as possible. Obviously you can have a difficulty level within the game though. I would love a game that has it's enemies change tactics everytime I approach them.

 

[scooby_dooby]

Well I don't agree with the A.I. part. Hell make it as smart as possible. Obviously you can have a difficulty level within the game though. I would love a game that has it's enemies change tactics everytime I approach them.

Brothers in Arms 2

 

[Sis]

Well I don't agree with the A.I. part. Hell make it as smart as possible. Obviously you can have a difficulty level within the game though. I would love a game that has it's enemies change tactics everytime I approach them.

You would find it enjoyable every single time you play they react unpredictable such that you never get a feel for what they are doing?

In fact, this is a well known problem in AI. You may think you want super smart AI, but in reality you want unpredictable within reason, or perhaps predictable unpreditability. Read up on some of the Halo AI implementation to understand what I mean. It's not about making the AI smart or dumb, but about making it behave well within the game world, in a predictable fashion.

 

[NavNucST3]

You would find it enjoyable every single time you play they react unpredictable such that you never get a feel for what they are doing?

In fact, this is a well known problem in AI. You may think you want super smart AI, but in reality you want unpredictable within reason, or perhaps predictable unpreditability. Read up on some of the Halo AI implementation to understand what I mean. It's not about making the AI smart or dumb, but about making it behave well within the game world, in a predictable fashion.

Apple had an issue with their random play for iTunes, they made it so random that people felt it wasn't random at all, so they had to dumb it down so that people felt that it was MORE random. So, perception is everything.

On a side note, physics is near and dear to my heart, so much so that it is even in my handle. Having said that, I find it interesting that people find physics so amazing, while I am happy for more people to join in the physics community, I don't think people actually want "realistic" physics. I think they want to be fooled into thinking they are looking at realistic physics; I feel the same about AI.
My background is more in the theoretical nuclear and quantum mechanics realm of physics so my "reality" is probably much different than those of the classical physics types, which I believe to be more prevalent and relevant in games.

 

[_phil_]

Y Read up on some of the Halo AI implementation to understand what I mean.

agreed ,it's about smart overlaping beetween code and level design.
http://halo.bungie.org/misc/gdc.2002.haloai/talk.html?page=1

 

[Shifty Geezer]

There's a fine line between playable AI and too much AI, depending on the genre. Take for example a side-scrolling shooter like R-Type. It's a good game in part because you know exactly where the enemy are. The AI is limited to the direction the enemies shoot, which varies the players motions so it's not the same game every time. If the enemy were ever intelligent you'd be dead in a second. Pitting one player versus hundreds of enemies, intelligence you is the last thing you want them to have! Whereas in a game like Morrowind/Oblivion, increased intelligence is important in several ways. Where you're not likely to be heavily outnumbered in an encounter, smart enemies are important to challenge the player.

It's actually a difficult area to pin down. I tend to find a lot of fun for me from games comes in finding ways to exploit the gameplay and weaknesses in the AI. I'm a problem solver, and like finding the best solution to a problem. As an example, RTS's tend to have a weakness. In Dune II there were several, not least constructing turrets in the middle of the enemy base where the enemy vehicles would attack it and destroy any of their own buildings in the way. In AOE you'd build an army of archers with a front line defense of melee and march them in formation. In these games, if the computer was truly intelligent there'd be very much a chance of stalemate, and certainly I'd get bored of not being able to build up because the computer has enough sense to keep whittling me down.

I think ultimately AI requirements might be divided into moderate AI for most games and Human Players for those who want a true challenge. If you want your enemy to be smart enough to know and do all the things you can, play online. Elsewise I think there needs to be a different 'computer game' experience, you versus the computer, for those that are wanting a different sort of challenge.

 

[blakjedi]

My background is more in the theoretical nuclear and quantum mechanics realm of physics so my "reality" is probably much different than those of the classical physics types, which I believe to be more prevalent and relevant in games.

Great post, heh :smile: . Speaking of theoretical physics no one wants a "magic bullet" to hit them in either a game or real life.

 

[Ben-Nice]

Well I don't agree with the A.I. part. Hell make it as smart as possible. Obviously you can have a difficulty level within the game though. I would love a game that has it's enemies change tactics everytime I approach them.

I don't think that is really possible but the new SIN on the Source Engine will take down statistics while your playing and adjust accordingly so there isn't really a set "difficulty" level. Like if your getting crazy headshots they will dive more and even wear helmets and stuff. No way to make an enemy change tactics everytime but you could have them base tactics on the enviroment itself. I mean in real life you would base your tactics on the enviroment, they do that to an extent right now in games like hide behind walls flip tables etc.

 

[scooby_dooby]

agreed ,it's about smart overlaping beetween code and level design.
http://halo.bungie.org/misc/gdc.2002.haloai/talk.html?page=1

wow, killer link, that was a great read. I thought this was pretty funny:
"nobody wants to have a reflex competition with an electron"

 

[mckmas8808]

I don't think that is really possible but the new SIN on the Source Engine will take down statistics while your playing and adjust accordingly so there isn't really a set "difficulty" level. Like if your getting crazy headshots they will dive more and even wear helmets and stuff. No way to make an enemy change tactics everytime but you could have them base tactics on the enviroment itself. I mean in real life you would base your tactics on the enviroment, they do that to an extent right now in games like hide behind walls flip tables etc.

Well I didn't mean litterly every single time. I meant what somebody said above, predictable unpredictablity.