Announcement

Collapse
No announcement yet.

UE4 will use Lightmass for its lighting system!

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

    #76
    and one of my major concerns there is the lack of transparency (alpha) shown in pretty much any of the existing UE4 demos, voxel based lighting or no.
    We have quite a bit of transparency in the Infiltrator demo, although it's more of the volumetric kind, like smoke. At the moment we only have a good lighting solution for volumetric type effects, we would like to do something for surfaces like glass and water as well.

    Comment


      #77
      One question from me, reflections so far seem like a env probe + SSR combo; so are the render target texture ones (like if we ever needed a mirror) still supported?
      Atm we have only scene captures, which render the scene from a certain perspective. It would be easy though to extend that back into planar reflections. The performance is pretty bad though, as with any technique that requires re-rendering and lighting the entire scene. We expect to rely on SSR much more for dynamic reflections.

      Comment


        #78
        Originally posted by daniwrig View Post
        The primary reason it didn't work out was lack of scalability. Rendering techniques that we develop need to work for a large variety of games, not just the highest high end machines.
        Thanks a lot for the info and for talking to us in the first place.

        All I can ask is, will it likely still need the highest of high end in 2-3 years? It would be nice to have the option even if it's a placeholder for something better you're working on. I would imagine transitioning from one dynamic system to another probably wouldn't be too painful, then again...

        Comment


          #79
          Originally posted by ambershee View Post
          Sure - on hardware that likely cost in excess of $3000.
          Well, I know you're right, and it'd be embarassing if engine runs well only on super-duper-high-end PCs. I'm totally agree here, I was just curious about SVOGI remaining a 'high-end-owners-only' feature.

          Originally posted by SeanO'Connor View Post
          Well that is what I was getting at. It seems like they had it working just fine for PCs--I can understand that consoles would need it removed, even though that is somewhat ridiculous for next gen machines that should be able to handle it.
          <...> I dare not question Epic developers, but so long as such effects could be turned off, I see no harm in leaving them in.
          +1, same here. I am also concerned, whether that beautiful Voxel Lighting (VL) was removed as well as Voxel GI, or these are two different techs thereby dynamic Voxel GI can't be included, and Voxel Lighting actually can. Honestly, I'm not a tech person, so I don't know much about VL, may be it's much faster than Voxel GI, or it isn't?

          Originally posted by daniwrig View Post
          The primary reason it didn't work out was lack of scalability. Rendering techniques that we develop need to work for a large variety of games, not just the highest high end machines.
          Thx for clearing that up! And one more question here:

          Are there any Area Lights & Area Shadows in the current build of UE4? I thought all these Area lighting, shadowing and specular features were available only via Voxel Lighting pass, so are those things cut for now, too?
          I've seen CryEngine 3 GDC presentation on its Area Lights, though there was nothing said about actual Area Shadows, besides the stuff showed off was pretty slow, it ran in editor on poor 30 fps.

          Comment


            #80
            Thanks for all the insight.

            Originally posted by daniwrig View Post
            The primary reason it didn't work out was lack of scalability. Rendering techniques that we develop need to work for a large variety of games, not just the highest high end machines.
            Not being a programmer I actually thought voxel approach advantage is scalability.

            I guess you can't disclose what other GI techniques are in consideration now?

            Comment


              #81
              whoa daniwrig thanks for so many answers! you're certainly 'calming the waters' around here
              the only question that remains from me now (in terms of rendering and graphics), is how much of all of these imrovements are DX11-only (I can guess a few but I can never be sure), and which of these improvements will affect DX9.

              Comment


                #82
                UE4 will supports DirectcX9/10 oO? I hear that on GDC interview an boss from Epic talks about UE4 is DX11-only, thats sounds promising for me (hope to have real next-gen in UDK 4.x, not an "UE3 with new skin for Editor + some minor graphics features for high-end PC").

                But my next-gen hopes realy weak - large companies prefer cheap and fast upgrading same old engine again and again - some new limited features and thats all

                Comment


                  #83
                  Originally posted by Lex4art View Post
                  UE4 will supports DirectcX9/10 oO? I hear that on GDC interview an boss from Epic talks about UE4 is DX11-only, thats sounds promising for me (hope to have real next-gen in UDK 4.x, not an "UE3 with new skin for Editor + some minor graphics features for high-end PC").

                  But my next-gen hopes realy weak - large companies prefer cheap and fast upgrading same old engine again and again - some new limited features and thats all
                  you do realize that DX11 requires a DX11 hardware GPU and at least windows 7? we're just not there yet.
                  on top of that, UE4 is meant to be as scalable as UE3 (UE4 will be ported to WebGL), and I don't see it dropping their iOS market. so if you have an engine that supports low-end graphics (mobile), why drop the still-mainstream DX9? it'd be a huge door slam to developers, as it'd close a huge potential market

                  Comment


                    #84
                    Originally posted by daniwrig View Post
                    There have been quite a few improvements to dynamic lighting techniques or methods that work well with dynamic lighting in UE4, over UE3:
                    Deferred shading everywhere! This is a huge one.
                    Tiled deferred shading - provides further benefits when many lights are on screen. Over a thousand on-screen in Infiltrator.
                    Subsurface shadowing of SSS materials
                    Screen space reflections
                    Image based lens flares (picks up any bright spot, doesn't need to be preplaced)
                    SSAO has been improved
                    CSM is quite a bit faster due to culling
                    Overall shadow depth rendering is faster, uses position only stream, better organization of vertices for locality
                    Cubemap ambient term, provides specular in unshadowed areas

                    But yes, dynamic GI is for sure needed and we'll get to it.
                    Thanks for stepping in and answering our questions! Well that sounds like some awesome improvements to the lighting system. One thing that has always annoyed me in the UE4 demos are the lens flares. Is it possible to disable them while the game is running, so the players can turn them off ingame?

                    Comment


                      #85
                      was a good read these last couple pages, thanks alot for all the info, it def cleared some things up

                      go epic!

                      Comment


                        #86
                        Chosker - well, you think like man who cares only about money, e.g. quality of project is not a tartget - only maximum profit on maximum multiplatform is main target. I think different - i don't need maximum profit ( +55% of Steam users already have DX11 and thats enough to have good market; also with time whis market will only grow, giving additional players on DX11 project). I need tool to put my dreams and ideas in 3D space with maximum possible quality and thats why DX9-10-old consoles compability is make huge problems with high quality content in "new" engine (like i have in CE3.x for example). Here is difference between us - we need different engines

                        Comment


                          #87
                          Originally posted by daniwrig View Post
                          Atm we have only scene captures, which render the scene from a certain perspective. It would be easy though to extend that back into planar reflections. The performance is pretty bad though, as with any technique that requires re-rendering and lighting the entire scene. We expect to rely on SSR much more for dynamic reflections.
                          Thanks for answering.
                          Just one more, will there be some tech paper like in the case of Samaritan? One of the features was new material layering system, I'm interested to see how different it is compared to the UE3 one.

                          Comment


                            #88
                            Lex4art: yes it's a little about money, but not from me but from Epic. Epic is a videogame company and is of course ruled by profits, and to think otherwise would only be fooling oneself.
                            now would you think Epic would willingly give up making his next flagship game Fortnite available to 45% of their potential customers?

                            as for myself, I can only say that using one engine or the next one is almost irrelevant if such 'quality' that you like to talk about, isn't reflected in the art itself. if the art/effects are good then the game's visual quality will be good. Look at Samaritan for example, it pushed UE3's graphical limits and even if you got to that level of visual quality you're limited by the hardware: Samaritan wouldn't really be playable on an average gamer's PC (because it was a tech demo with no gameplay/AI), and you can't make a game for just the probably less than ~5% of gamers that have a $3000+ rig

                            Comment


                              #89
                              Chosker - imho Samaritan demo shows that adding new features to old engine is realy uneffective/perfomance-costly (this also perfectly clear in DayTime demo that ships with UDK - small 30x30m scene have realy bad fps);btw year later Samaritan demo works on one GTX680 - its ~800$ rig, not 3000$. Also, maxed out graphics - yep, for high-end PC-s; but who says that DX11-middle end and low-end hardware cant run project? Yes, users recive cheaper picture but that was their own choise then they buying cheaper hardware - and its still DX11, just with more agressive LODs ratio/less shaders&shadows samples/lower MIPs unstead of MIP0. Heavy-multiplatform engine (DX9/old consoles/DX11/tablets Open GL ES/etc) will requre way more evil restrictions to content creation side & engine features; it will be full of compromises, will have large and complicated codebase (its 3x longer to implement something new to this moster!) and so on - you can see consiquences of this in current games like BF3/Crysis3/etc: content is cheap (only looks good from 5+m distance), effects quality are "childish" and only post-effects realy maxed out ).

                              Comment


                                #90
                                @daniwrig,

                                Thank you for all the responses and openness. Really appreciate it.

                                Comment

                                Working...
                                X