Announcement

Collapse
No announcement yet.

New Unreal Egine 4 Tech Demo "Inflitator"

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    #16
    I wanted to see this.
    Video has been removed by the user!

    Comment


      #17
      UT4, destruction everywhere. Great idea.



      Now let's stop and think through the notion of having full PhysX destructible environments in a game that also depends on being reliably networked across as many as 32 players.

      Suppose there's this high hanging wall or rock that you explode into hundreds of chunks of debris. About the same time, you notice there's a deft enemy running underneath the falling debris, who, it appears, will be just able to escape. As you are yourself a skilled UT player and expert blower-upper, with your perfect aim and timing you're able to shock-rifle the biggest of the falling chunks forward, such that your fleet-of-foot opponent will assuredly be crushed under ten tons of tumbling terrain.

      Very cool.



      Now, what is actually involved when it comes to simulating all of those events to (potentially) all 32 players on the server?

      When you explode some piece of something PhysX-style, it spawns thousands of new somethings all at once, perhaps hundreds of which are significant enough that they must be assigned a consistent and replicable size, shape, orientation, and velocity. If the simulation is to be believable, none of these properties can be predetermined, as any player could have triggered this piece of something's destruction at any time, from any point, and from any angle. So in one single fraction-of-a-second tick, every little newly spawned chunk must be described at least to its major corners and crannies and speed and direction ... then as that shower of debris hurtles to the ground, many of the pieces are going to collide with each other and with other bits of the environment. All of those events need to be replicated faithfully to each observing client, or else of course you end up with what appears to the player as clipped or invisible rocks and random damage events — in a word, lag.



      All told, a single player's single interaction with the environment may set into motion a whole sequence of events like this, which could easily require several hundred thousand bits to be pumped across the network within a tiny fraction of a second. To say nothing of what would happen in the likely common scenario of several different players all triggering similar complex chains of events that would also need to be crammed into the internets at the same time.

      In a typical multiplayer game of UT, your connection to the server consumes a little less than 100000 bits per second. For small-scale physics environments, and servers with very low player counts, that's actually ok — but throw PhysX-scale destruction into the mix on a decent sized server, and it's a safe bet you'd need your pipe to be at least 100 times fatter, possibly even more than that, while still able to accommodate a constant stream of data, for several hours solid, of course with no buffering or interruption.

      Then you consider the load on servers, which must process this exponential cacophony of events, burn extra CPU to filter the most relevant interactions to be selectively broadcast to as many as 32 players, and stream what becomes hundreds of megabits of data per second to all the clients ... suddenly an admin hosting just one well-trafficked UT server will be looking forward to an expense comparable to serving Netflix to your entire town.



      These easy-to-wish-for features may be fun to dream about, but out in the real world there's a lot more technical heft behind them than most of us bother to consider.



      ... more buckets of cold water — I am no fun am I =/

      Comment


        #18
        Originally posted by tarnationsauce2 View Post
        I wanted to see this.
        Video has been removed by the user!
        http://www.youtube.com/watch?v=dO2rM-l-vdQ

        Comment


          #19
          nice video....more "UTish" than the samaritan and the elemental ones

          Comment


            #20
            Originally posted by Veggie_D View Post
            UT4, destruction everywhere. Great idea.



            Now let's stop and think through the notion of having full PhysX destructible environments in a game that also depends on being reliably networked across as many as 32 players.

            Suppose there's this high hanging wall or rock that you explode into hundreds of chunks of debris. About the same time, you notice there's a deft enemy running underneath the falling debris, who, it appears, will be just able to escape. As you are yourself a skilled UT player and expert blower-upper, with your perfect aim and timing you're able to shock-rifle the biggest of the falling chunks forward, such that your fleet-of-foot opponent will assuredly be crushed under ten tons of tumbling terrain.

            Very cool.



            Now, what is actually involved when it comes to simulating all of those events to (potentially) all 32 players on the server?

            When you explode some piece of something PhysX-style, it spawns thousands of new somethings all at once, perhaps hundreds of which are significant enough that they must be assigned a consistent and replicable size, shape, orientation, and velocity. If the simulation is to be believable, none of these properties can be predetermined, as any player could have triggered this piece of something's destruction at any time, from any point, and from any angle. So in one single fraction-of-a-second tick, every little newly spawned chunk must be described at least to its major corners and crannies and speed and direction ... then as that shower of debris hurtles to the ground, many of the pieces are going to collide with each other and with other bits of the environment. All of those events need to be replicated faithfully to each observing client, or else of course you end up with what appears to the player as clipped or invisible rocks and random damage events — in a word, lag.



            All told, a single player's single interaction with the environment may set into motion a whole sequence of events like this, which could easily require several hundred thousand bits to be pumped across the network within a tiny fraction of a second. To say nothing of what would happen in the likely common scenario of several different players all triggering similar complex chains of events that would also need to be crammed into the internets at the same time.

            In a typical multiplayer game of UT, your connection to the server consumes a little less than 100000 bits per second. For small-scale physics environments, and servers with very low player counts, that's actually ok — but throw PhysX-scale destruction into the mix on a decent sized server, and it's a safe bet you'd need your pipe to be at least 100 times fatter, possibly even more than that, while still able to accommodate a constant stream of data, for several hours solid, of course with no buffering or interruption.

            Then you consider the load on servers, which must process this exponential cacophony of events, burn extra CPU to filter the most relevant interactions to be selectively broadcast to as many as 32 players, and stream what becomes hundreds of megabits of data per second to all the clients ... suddenly an admin hosting just one well-trafficked UT server will be looking forward to an expense comparable to serving Netflix to your entire town.



            These easy-to-wish-for features may be fun to dream about, but out in the real world there's a lot more technical heft behind them than most of us bother to consider.



            ... more buckets of cold water — I am no fun am I =/

            Yeah I can see what your saying about the Lag aspect but then when you look at other games that have done it perfectly fine like Battlefield Bad Company 2, that was able to do massive ammounts of destruction online so I dont see this being as big of an issue as it might first seem, so long as you have card that can handle graphical nature of destruction then this shouldn't cause lag. I know that the new Unreal Engine 4 "does" support destruction, its in the technical notes on the engine itself, so theorectly, this can be done... Its worth thinking about anyway, people are always complaining about how Unreal Tournament is boring, needs to change, what can it do differently? well, destructable enviroments is definitely something that could potentially change how unreal is played forever in my opinion.

            I think most people will agree its a nice idea, but weather it can be put into an unreal tournament game is another matter, I'd like to think its possible.

            Comment


              #21
              I gots wood...

              That looks smexy!!

              Comment


                #22
                Thanks! Yeah looks amazing especially for realtime.

                Comment


                  #23
                  Destructible environment, that could be a challenge for games like UT which usually heavily depends on level structure and map flow.
                  Gameplay-wise these have to concerned:
                  - Indoor or outdoor? It can be very different.
                  - Vertical gameplay being ruined by broken floor
                  - Dynamic AI pathing
                  - Dynamic pickups. It will look ridiculous if a health-pack floats in the air on the "floor" which is no more there
                  - Is this going to affect "Fast-Paced"
                  ...
                  Basically the trick is find out what part of a level can be broken, and how that really affect UT's gameplay. That's hard. I think, but if they work out a nice solution, it could be the change many ppl would like to see.
                  Maybe something like "the environment gets rebuilt/healed by itself after a few minutes/seconds", since it is called Unreal Tournament, that would sound unreal enough to me.

                  Comment


                    #24
                    Maybe something more like canned particle effects that are precomputed but just up the visceral nature of the game a little. Probably a solution in there somewhere to add to the wow factor without killing the game.

                    Comment


                      #25
                      I can already imagine my PC hyperventilating and begging me to stop using UE4

                      Comment


                        #26
                        I remember alot of demos looking that good and then the game turning out to be not even half as good looking. Remember it's an advertising demo. Advertising demo for what you ask? Money.

                        Comment


                          #27
                          It's just a tech demo, that is all. A brief and informative read on the matter:
                          http://www.gamespot.com/events/gdc-2...ml?sid=6406178

                          On a side note: complete destructibility in the MP side of UT is a bad idea, it would totally shift (read: ruin) the metagame.

                          Comment


                            #28
                            There's an astute point people are homing in on ...

                            Compare the promotional strategies employed by Epic circa 2006-2007, for Unreal Engine 3, to what you're seeing today for UE4.

                            Back then, Epic's technical demonstrations consisted of: (1) this as-yet unheard of title for Xbox360 called "Gears of War"; followed by (2) preview after preview, and eventual release, however shaky, of the widely anticipated sequel to the venerable "Unreal Tournament" — both real-world, bleeding-edge, de facto reference implementations, if you will, for what studios could expect to achieve after buying a license for this top-shelf commercial game engine. An engine which, remember, at that time was still unfamiliar to a good lot of people.

                            And then there's today. Virtually everyone, across industries far and wide, is well aware of the Unreal Engine brand. Further, the bulk of Epic's remaining talent and intellectual assets are, now more than ever before, concentrated on the Engine, and not with any particular game franchise. So naturally, what is Epic putting on display to make their already-famous engine famous? Certainly not the showcases of upcoming titles like Gears, and definitely not Unreal Tournament. Why would they?

                            I think it's well understood, even by us rabble-folk, that Epic has long been in a position where they stand to profit far more by selling and supporting the Engine, versus bothering with the creative hassle of churning out real-world "AAA" titles that must be developed, polished (ahem) and marketed (cough).

                            Marketing the moneymaker, then, boils down to a simple proposition for Epic: what is the minimum investment required to make client studio heads pee their pants and throw hundreds of millions of dollars at us? Do we still need to prove the Engine's worth by dumping tens of millions of dollars into projects that will take years to complete and then might not even turn a profit? Or at this point, would a few hundred thousand spent on some apropos-of-nothing cinematic set piece achieve the same effect?

                            Epic's mantras for the modern era are clear: ultra-conservative, low risk, same-or-way-better reward. To the Unreal Tournament lovers still among us, yes, Epic's souls may be sold out, but businesses that stick around don't run on soul. It's all dollars, cents, and public relations.

                            So let us marvel at their latest eye candy, but let us not run our wishful imaginations amok — UT's coffin has been nailed.

                            Comment


                              #29
                              On the day the first UE4 game is released, or the day before, we should all login to UT3 and give it a decent farewell. Whether it was worse than the other UT's or not, i bet most of us still had a bit of fun there and it was the sequel of the legendaries UT2004 and UT99. You shall never be forgotten UT3, you shall be immortalized

                              Comment


                                #30
                                Mega Detail Man!!!

                                I have to admite that UE4 demos really look beyond what is expected for real time game play!...

                                ...the question is "is it a movie or a PC realtime animation!?"

                                ...and what kind of PC hardware does it run on???

                                ...But if that is a glipse of Unreal3! you got to like it!!! (rumors and speculations, but still mighty cool!!!)

                                Comment

                                Working...
                                X