Meant to be Seen

Owen wrote:Unity's graphics capabilities are still pretty poor. Yes, you can use the latest shader techniques, but the rendering pipeline is still an inefficient black box that you can't touch. So you can't make your game run nearly as fast as someing in the Unreal engine, and therefore can't include as much detail or have such large environments.

From what I understand this is no longer the case in Unity, the graphics classes are completely accessible now so graphics coders can get as close to the metal as they want. I don't know how long this has been the case as I am new to Unity and didn't really follow it's development until recently.

I've been working as a Unity3D Game Dev for the past 1.5-2 years

Both engines are great and amazing things can be accomplished in both. But if you are a single person or a small group of people with not too much time, go with Unity3D.

No, nothing has changed there in over two years, except that the free version can now use some pro features. The only way you can get down to the metal is to write direct3d code in a native windows plugin, at which point you are not using Unity's renderer at all.

But I do agree that Unity is the best choice out there if you just want to put something together fast. For prototyping especially it is ideal. Just keep in mind that you will not be able to scale up very far compared to UDK.

Krenzo wrote:You can use C++ DLLs if you wrap it in C#.

Very interesting, so the hydra could work with free Unity ?
Hmm maybe it would be time to make an Unity general thread...

Owen wrote:No, nothing has changed there in over two years, except that the free version can now use some pro features. The only way you can get down to the metal is to write direct3d code in a native windows plugin, at which point you are not using Unity's renderer at all.

But I do agree that Unity is the best choice out there if you just want to put something together fast. For prototyping especially it is ideal. Just keep in mind that you will not be able to scale up very far compared to UDK.

Sure you can, you just need to work a bit harder with the optimizatio, i can agree on that, but you will still save time in the end because Unitys workflow is superior : )

Owen wrote:Just keep in mind that you will not be able to scale up very far compared to UDK.

I am willing to believe you, you toyed a lot with Unity before.
But do you think we can do open-wold games with UDK ? ( I mean, with lagless streaming ?)

Its been done to some degree, such as in the Batman games, but requires careful planning because you can't have very long view distances. It doesn't provide an LOD system like in Skyrim that allows you to see the whole world at once (Arkham City hacked something in, not sure exactly what they did or how hard it would be without engine source). But if you break up the player's vision and create natural bottlenecks then you can link up a huge game world as multiple streaming scenes with no pauisng to load.

And actually I have been working in Unity professionally for over two years as a graphics programmer.

Thanks for you valuable input.
I think things are clear now.
I was also in discussion via Youtube with another Unity devs about that issue.
I was told about fbx sdk and obj parser to do content streaming (so I guess it's no lagless)
Another dev working on a Trespasser remake was forced to switch to Cryengine 3 (and now he's facing another draw backs : importing models,the license of the engine...)

It is possible to generate the geometry in scripts in Unity, loading data from some external format and building the vertex buffers yourself when needed. Making a bunch of async loaded scenes out of level chunks can work too, but asset loading is on the main thread like pretty much everything else. You could do an open world type game this way, but you would also need to handle things like occlusion culling yourself with some kind of pre-baked visibility data, or use a portal/raycast approach to link sections together. Its possible to make moderately large environments using those approaches, but making it perform well is very hard, and you quickly lose the benefits of the unity editor.

Interesting, the team behind Wasteland 2 explains why they chose Unity for their RPG (ease of use by artists and designers, crossplatform (Windows, Mac and Linux), support and expertise available from vendor and in community...)

Owen wrote: Just keep in mind that you will not be able to scale up very far compared to UDK.

What do you exactly mean by not being able to scale up compared to UDK?

I installed the Unity 4.xx last night and started playing around with. I have never used UDK, Unity or any game engine and yet Unity is very easy to understand and play around. The following is the first of over 25 training videos on youtube.

http://www.youtube.com/watch?v=5-X-Ebh1 ... B39F84E292

They begin with the island demo, an older demo you will need to change slightly to get it to pass through debug. Again being completely new I Googled the errors and corrected them. I stop the videos a lot do what they do and more. My wife saw me playing around and was interested. As soon as we build her a new computer she wants learn and build content together. Yea, my wife is that cool. Anyways, if you have any thoughts of trying your hand at editing with a game engine go for it. UDK 4 is supposedly much more easy to edit and debug then previous versions.

http://www.youtube.com/watch?v=MOvfn1p92_8

So I will likely wait till that is released to try my hand at UDK

I have downloaded Unity and tinkered a bit but am stalling until we get the Rift units and can find out what it will take to get each working.

I know everyone is "promising" full support but I'd hate to have to wait six months working on a project until support gets patched in.

Is there any word on Unreal Engine 4's UDK likely release date?

Diorama wrote:Is there any word on Unreal Engine 4's UDK likely release date?

I have been looking around for answer to that too, and as best I can see, they aren't talking about it. I imagine it will be a long while (over a year), since they will focus first on getting it out and working smoothly to the big-wig developers for the new consoles coming out this year.

Modab wrote:

Diorama wrote:Is there any word on Unreal Engine 4's UDK likely release date?

I have been looking around for answer to that too, and as best I can see, they aren't talking about it. I imagine it will be a long while (over a year), since they will focus first on getting it out and working smoothly to the big-wig developers for the new consoles coming out this year.

I too poked around after seeing that amazing demo, but found the same thing, no word of when this will actually release. I find it funny that Epic is pushing Microsoft and Sony to put powerful graphics processor giving the "gtx670 or gtx680" specifically as suitable Yet these cards are already a year old and Unreal 4 is still not released.

Pingles wrote:I have downloaded Unity and tinkered a bit but am stalling until we get the Rift units and can find out what it will take to get each working.
I know everyone is "promising" full support but I'd hate to have to wait six months working on a project until support gets patched in.

Well i'm not too worried here. Seeing the last demo videos, the Rift looks already well implemented in Unity and UDK...
We should have a well documented procedure, at worst we'll always have the mbts3d and Unity communities to give a hand...

Well as for unity i think it will be easy as just dragging pre-made oculus ready first person character controller into your scene, then press play and away you go

hammerbot wrote:Well as for unity i think it will be easy as just dragging pre-made oculus ready first person character controller into your scene, then press play and away you go

Yeah, as for the TrackIR or Hydra integration. With just a few variables to fiddle with (PID, convergence...).

Owen wrote:Its been done to some degree, such as in the Batman games, but requires careful planning because you can't have very long view distances. It doesn't provide an LOD system like in Skyrim that allows you to see the whole world at once (Arkham City hacked something in, not sure exactly what they did or how hard it would be without engine source). But if you break up the player's vision and create natural bottlenecks then you can link up a huge game world as multiple streaming scenes with no pauisng to load.

And actually I have been working in Unity professionally for over two years as a graphics programmer.

arkham city seems to use it's horseshoe shaped main area as a wide looping corridor then.

Reading this :

jis wrote:I think you should fear these:
- frame-delay latency: some engines, and mostly super optimized engines like Unreal or maybe the CryEngine, add an additionnal frame of latency to be able to parallelize stuff more. The parallelized rendering commands buffer will be sent at the next frame.
- normal input->render latency: no direct input bypass to let the render use the newest input data instead of using the one coming from the beginning of the frame. I think most of the engines will have this issue. Carmack implemented the bypass specially for the Doom3 BFG I heard.

Has somebody info about this concerning Unity and UDK ?

Cool tech demos I hope haven't been posted.

[youtube]http://www.youtube.com/watch?v=QDBoXwkTZcE[/youtube]


[youtube]http://www.youtube.com/watch?v=acR4n6lJEdQ[/youtube]

Anyone working with Unity Pro and a DIY Rift?

I've got my YEI 3 Space Sensor working but it looks like you need an Image Effect in Unity(Unity Pro only) to do the barrel distortion for the optics and stereoscopic 3D. Anyone already write an Image Effect for Unity(Yes, I know about the Doom3 and Biclops shaders, that's not what I'm talking about) before I take it up?

Does anyone know how will the Rift be implemented into Unity? Is it gonna be an asset just like the FPS Controller -with PnP/ drag and drop functionality- or is there gonna be coding and scripting involved?

I've been trying to look for it but all I can find are the announcements that Unity will support the Rift but no specifics...

ftarnogol wrote:Does anyone know how will the Rift be implemented into Unity? Is it gonna be an asset just like the FPS Controller -with PnP/ drag and drop functionality- or is there gonna be coding and scripting involved?

My guess is that there will be a new RiftCamera entity or options on the default Camera entity as well as new FPSRiftController and ThirdPersonRiftController entities.

There has been no official details of Unity specifics.

Yeah - they said we would get an early access to the SDK, I hope this will happen at some point soon. But I can wait until a week before I get my Rift ... just want to have it integrated by then so I can try it out right after playing HL2 or Quake with it

German wrote:

ftarnogol wrote:Does anyone know how will the Rift be implemented into Unity? Is it gonna be an asset just like the FPS Controller -with PnP/ drag and drop functionality- or is there gonna be coding and scripting involved?

My guess is that there will be a new RiftCamera entity or options on the default Camera entity as well as new FPSRiftController and ThirdPersonRiftController entities.

There has been no official details of Unity specifics.

Fingers crossed for you being right!

Something I haven't seen mentioned is networking and multiplayer.

UDK - good.
Unity3D - might as well not exist.

Some might say I'm being overly harsh on Unity's networking. I would say "Okay, so it may be useful for a prototype to demo a concept, but only if it's incredibly basic". It's the only really sub-par feature in Unity that I've found. It really needs to be pulled out and completely redone imo.

If you want to do multiplayer in Unity you're going to want to ignore Unity's built in networking and use a 3rd party library or write your own multiplayer networking library.

ChrisJD wrote:Something I haven't seen mentioned is networking and multiplayer.

Well seen !
I didn't get into Unity networking yet (I keep the hardest for the end), but it looks like you are right, the built in networking is apparently badly documented, badly implemented, and buggy as hell.
Did anyone try a 3rd party library like Ultimate Networking Project ?

I'm mulling this exact question currently, and this thread has only added to my doubts about both of these engines. It seems the best option for someone who only knows a bit of Python and aspires to create an open-world MMO geared toward the Rift would be best served by finding $1500 laying in the street, doing some prototyping in Boo under Unity Pro with something like Photon for networking, and spending the previous 10 years coding a custom engine in C++. Anyone want to lend me your time machine?

drifter wrote:Reading this :

jis wrote:I think you should fear these:
- frame-delay latency: some engines, and mostly super optimized engines like Unreal or maybe the CryEngine, add an additionnal frame of latency to be able to parallelize stuff more. The parallelized rendering commands buffer will be sent at the next frame.
- normal input->render latency: no direct input bypass to let the render use the newest input data instead of using the one coming from the beginning of the frame. I think most of the engines will have this issue. Carmack implemented the bypass specially for the Doom3 BFG I heard.

Has somebody info about this concerning Unity and UDK ?

Well, that's a little troubling. Uh, bump?

For Unity you can try out Photon. They have a good sample in Assetstore. My own multiplayer experiece with Unity was 3 years ago when someone made a good package with Smartfox that allowed me to have multiplayer straight out of the box. So i converted my single player demo into multiplayer in just 5 minutes. Of course i did not have any good avatars in it. The avatars were only a floating balls.

But it worked and it was easy to set up. I have not tried with Photon but i believe that they should have "out of the box" solution aswell.

nateight wrote:I'm mulling this exact question currently, and this thread has only added to my doubts about both of these engines. It seems the best option for someone who only knows a bit of Python and aspires to create an open-world MMO geared toward the Rift would be best served by finding $1500 laying in the street, doing some prototyping in Boo under Unity Pro with something like Photon for networking, and spending the previous 10 years coding a custom engine in C++. Anyone want to lend me your time machine?

You have access to the full .NET 2.0 library. You could always roll your own network stack in regular Unity. However, it's going to cost a lot more than $1500 for "an open-world MMO".

Unity's built-in networking is not unusable. It would do a passable job for a pitch prototype or something.

German wrote:it's going to cost a lot more than $1500 for "an open-world MMO".

You're not wrong about that; the $1500 was just for the Unity Pro license that opens up the parts of Unity necessary to start building the thing. I remain hopeful about the funding that would ultimately be necessary, though - if you manage to get a workable demo constructed, Kickstartering up the bankroll to afford a few strategically placed server clusters starts to become realistic, and at that point it might even be feasible to use AWS or Azure and keep the total cost merely large rather than astronomical. But if both Unity Pro and UDK demand a design philosophy dominated by smallish rooms connected by s-curve hallways to manage the occlusion culling, they're entirely out - I may as well crack open a C++ book without even installing either one.

MMO and Rift should not be in the same sentence, paragraph, or even message body. MMOs are designed to run with lots of latency. All of the talk around here about immersive VR focuses on reducing latency. There's also the fact that you need several million dollars to fund an MMO. MMOs are not the correct genre to target with VR.

Does anyone have much experience with UDK? I'm playing around with UDK, and I am not a fan of Unreal Script. Its syntax is a little weird, but the biggest problem is there's no debugger. It's only available to full Unreal Engine licensees. That's a huge red flag there. Unity seems much better in that you get the familiar syntax of C# with a fully functioning IDE and debugger. I wish Unity came with better samples to play around with. The sample UDK levels are gorgeous, and then Unity has samples that look like they're made for mobile apps.

Krenzo wrote:MMO and Rift should not be in the same sentence, paragraph, or even message body. MMOs are designed to run with lots of latency. All of the talk around here about immersive VR focuses on reducing latency. There's also the fact that you need several million dollars to fund an MMO. MMOs are not the correct genre to target with VR..

The best stuff is always the stuff that folks say "can't" or "shouldn't" be done.

While I'm not in the position to develop a new MMO I'd be willing to try a few existing MMOs if they offered Rift support.

I'd love to fly above the WoW worlds on my Griffin using the Rift.

Pingles wrote:The best stuff is always the stuff that folks say "can't" or "shouldn't" be done.

What about staring directly at the sun or going swimming after eating?

Krenzo wrote: What about staring directly at the sun or going swimming after eating?

Nothing wrong with staring at the sun, given proper protection and http://www.snopes.com/oldwives/hourwait.asp

Krenzo wrote:MMO and Rift should not be in the same sentence, paragraph, or even message body. MMOs are designed to run with lots of latency. All of the talk around here about immersive VR focuses on reducing latency. There's also the fact that you need several million dollars to fund an MMO. MMOs are not the correct genre to target with VR.

But there is a key difference in where the latency lies. For VR, the important latency is between your head motions and your view adjusting- which is completely independent of the latency with the mmo server, or the impact of your actions on the other players. It may not be suitable for some specific cases, like twitch-based combat, but in general an MMO should be able to work fine.

Mystify wrote:But there is a key difference in where the latency lies. For VR, the important latency is between your head motions and your view adjusting- which is completely independent of the latency with the mmo server, or the impact of your actions on the other players. It may not be suitable for some specific cases, like twitch-based combat, but in general an MMO should be able to work fine.

Yes, but isn't one of the key components of VR a heightened sense of interaction? Is your idea of VR just to look at a cartoony, low res depiction of Azeroth and say "oh wow, I feel like I'm really there"? Do you really want to swing your virtual sword that you feel is really there only to have to wait 500-1000 ms to see that a monster has flinched in response to your attack? Just because latency exists in a different area doesn't mean it doesn't affect your level of immersion.

Krenzo wrote:

Mystify wrote:But there is a key difference in where the latency lies. For VR, the important latency is between your head motions and your view adjusting- which is completely independent of the latency with the mmo server, or the impact of your actions on the other players. It may not be suitable for some specific cases, like twitch-based combat, but in general an MMO should be able to work fine.

Yes, but isn't one of the key components of VR a heightened sense of interaction? Is your idea of VR just to look at a cartoony, low res depiction of Azeroth and say "oh wow, I feel like I'm really there"? Do you really want to swing your virtual sword that you feel is really there only to have to wait 500-1000 ms later to see that a monster has flinched in response to your attack? Just because latency exists in a different area doesn't mean it doesn't affect your level of immersion.

That latency only exists if you wait for the server to tell you the monster reacted to your blow. That is easily something you can predict client-side.
And it does matter where the latency is. Head-tracking latency is something that is very sensitive and must be minimized. Body tracking latency, in contrast, can get by with a much higher latency.
Plus, there are all sorts of tricks you can utilize if latency is important. Say, we want to have player vs. player swordfights, and you need to have low latency between them . Well, everybody in the world does not need to have low latency to your swordfight, just the participants. They could establish a direct connected designed for low latency, and communicate their motions directly - the same types of latency people deal with in twitch-based shooters all the time. They can also send it to the server to verify that everything is kosher, but the actual latency between 2 people interacting can be minimized. Client-side simulation of interactions with the world can minimize the latency there.

MMOs don't do these things because they don't need to, but that doesn't mean it can't be done.