UE4.5 released...

We'd love Unity Free 5.0 in open beta.

 

Exactly. I think doing an open beta for 4.6 was a great idea and I hope for future releases (5.0) to do something similar.

 

Well the problem with open beta from the start is the signal to noise ratio. You'll find the largest number of people are actually not well equipped to understand if it's their fault or not and pretty much overwhelm QA with noise... meaning the real issues don't get through.

So I understand the need for Alpha, but I believe that beta should always be open to the public. I haven't (yet) seen a real value to a beta program that gets leaked constantly by people who haven't signed NDAs etc.

 

Yes. I would absolutely love to see more open betas, similar to Unity 4.6.

 

Fixed for you bro.

Aside from jabs at the UnrealTurfers, what makes Unreal and Unity different is that they're different tools that are suited to different things. Find which is better at a given task, and use said engine appropriately.

 

I would, personally, want to have access to the Unity 5 beta right now, both to try out the new features, as well as to give feedback. I usually spend some time figuring out if a bug I encountered is really a bug, and I don't submit bugs until I find a clear replication case. I know not everyone is like this so, I guess, it could create noise if everyone starts reporting random stuff without clear explanation or replication.

 

Yes....please do, like shadowK said it would be faster to iterate and find bugs. That's how UE did all this time isn't???

 

Unity is not responsible for what people do with their open betas. People are already warned that it's not a final version and loads of issues/problems could happen with that version.
What Unity should do is like an "experimental section" similar to UE4 on the project settings to enable features for testing. But I'm not sure if that will be feasible in Unity. In unreal source, you got plenty of modules and systems, they are pretty independent and can be replaced, thus adding plugins and modules for testing is a breeze.

 

It probably would be a good idea to label stuff as "experimental" rather than as "beta" given all the definitions now.

 

Specially when the whole idea is to gather user feedbacks and will probably never see the daylight.

 

I like them both. Each is more suited for certain projects over the other. Neither is going to go away because of the other. UE4 is so cheap that the engine vs engine conversations are mute for me. I just like to see and talk about the updates when either of them are updated. Both Unity and UE4 are awesome. I just like Game Development period so I might be bias in my view Cutting my cigarette smoking down pays for my Unity Pro lease and the UE4 monthly cost is nothing

 

If I was building a full on MMO with large terrains, custom networking code and huge infrastructure, I'd use UE4.

For everything else, I'd use Unity.

 

You know what I've found it excels with is Corridor games / FPS shooters, small RPG games / platform games / beat em ups and boxed off sandbox games. Anything that takes advantage of light mass reflection / advanced features..

It doesn't actually do the large world MMO thing or large openworld RPG games very well, CryEngine is much better suited for the task... Although it's improving drastically with every release.

Not that you couldn't do it with a lot of effort, then again you could do it with notepad and OpenGL but I wouldn't ..

 

Why not?

 

UE4 does not do that well at this point in time with large outdoor scenes. The terrain stuff needs work.

 

no

every thing is made for you at UE so u dont have to make anything, if you have been to UE then you know what i am talking about

its stupid

if people cant figure that out, then they are stupid

UE is stupid.

 

Most of the tools are capable, but! Foliage rendering model is pretty poor at the moment overdraw can strip your 780TI to it's knees as soon as you paint a smallish field and my grass is only 32 verts . It lacks an FR model, which makes transparency an issue, but that's not the biggest problems. Even with Cull volumes, occlusion tech's and LOD's the whole engine grinds down the mightiest of graphics card. You really need a farm if you're considering using lightmass or LPV just adds heavily to the problem. Not that lightmass really seems to make a big difference for outdoors anyway plus can't be used with TOD systems.. Oh and multiplayer doesn't support origin re-shift..

Trying to balance terrain resolution and still have tools working correctly has been problematic, I did a comparison and with a 2X2 quad at 2017X2017 resolution tile, it was eating up 8 million polys. Best is 1009X1009 @ 1X1 Quads with equals to around 350,000 polys for the terrain. Then you have issues with tools..

Sure with modern hardware that really doesn't matter too much 11-12 million polys in a view isn't the end of the world. But you push your luck really quick, origin shifting isn't supported in multiplayer. Low res causes paint tools to go doolally.. It does crunch millions of polys well and does have parallel rendering..

Also it's been buggy, most of the people using it won't be making big rpg games or MMO's so they won't notice and / or care. Funnily enough, the thing that seems to make no difference is post effects.. They don't really seem to add much latency at all.

You see Skyrim running of pieces of junk, you can surpass it graphically no problem with UE4 or you can strip it down to the bare which looks worse but still requires top end hardware to even have a chance. Personally I wouldn't like to tackle all this, it would be quite a large effort.. Not saying we can't or we won't, I'd seriously prefer not to..

TL;DR, the performance side needs a lot of work.

I wonder how @Chariots is doing? They are doing similar stuff to us.

Well they are taking steps to resolve all this, but right now.. I wouldn't go there.

@aaronhm77

I wish they did, make life a lot easier for me. How deep did you get into a project? I've not really noticed any difference between what Unity give you and what UE does.

Last time I checked there was no buttons that created Artwork, AI, Performance tweaks, storyline or anything that merits an actual game for me.. What I have seen is tons of GRFX OMG examples..

 

What people should be taking from this is if they want a top notch game, then yes... you need game developer balls. You need to put the work in, no matter what you use.

 

I would say the complete opposite is true. I would use Unity if I was trying to build a large seamless world.

Unreal excels at small to medium, static, indoor scenes. Lightmass and reflection probes are key to making a good looking (and fast running) environment and both aren't viable in a large, dynamic world. Reflection probes have a cap of ~350 and don't update asynchronously as of 4.4. (hiccups) You might be able to bake just a skylight and have a single dynamic directional light for night/day (with color correction to sell the effect) but even then I think lightmass would take weeks for a mid sized world.

That doesn't even tackle the foliage issues that ShadowK mentioned. I'm amazed at how poorly Speedtree performs in UE4.

I'm afraid that Enlighten's bake times might make large worlds an issue for Unity 5 too.

 

I've said it before and I'll say it again. You need to put the work in with unity or ue4. Ultimately, they're just game engines, they're not your actual game, so you will need to do a lot of custom work to squeeze the maximum out of them.

In my case, for my current game I scrapped Unity's lights and shadows and built my own from the ground up, using as much of Unity as I could as a base. For the 2.5D perspective this gained me at least double the visual fidelity and 2x the performance - so an incredibly big win for the lower end hardware.

This isn't absurd - it's precisely what big boys often do, and indies fail to do. You have to do it to reap the rewards and it's why I'd prefer Unity to carry on being flexible but make it easier for us to dig in and change things. Currently, Unity makes it harder to do this than UE4 does, and I'd like to see that change with more options, even if it means more work.

A massive open world game will fail hard on UE4 and Unity out of the box. You'll have to sort out a lot of tools and figure out where your bottlenecks will be for both engines, then put tonnes of work in addressing those issues.

I understand Unity is aware of this and wants to help things along in this manner as a longer term goal, so it'll be interesting to see what happens there, but that's not really there at the time of writing.

 

You know what, I put a lot of thought into it and no it wouldn't.. Simply due to the middleware, AAA wouldn't put a lot of though into it. They'd do what Unity does and buy middleware, Umbra and Enlighten are perfect for these sorts of games, but with the UE4 IPP it's going to cost a fortune and all of a sudden Unreal doesn't seem like a good option anymore..

Umbra alone can cull shadowcasters, one of the biggest performance sucks ever invented. It would resolve a massive part of it, also why would we go to the trouble of introducing a Multi-Pass deferred rendering solution so we can use transparency and effective shading models? There is a point at which the duty falls back on the engine developer, or what the hell is the point in using the engine? If you spearheaded my team of 10, could you build a solution comparable with Enlighten and Umbra and still release an openworld RPG within a 4 year time frame? I doubt it I must admit.

Simply put bar world composition, Unity is far more equipped to deal with this sort of stuff. Even if it's out the box graphically far behind and lacks some fancy tools like world composition and a material editor / cinematics and advanced VFX. Ok quite a few tools , but still to make a game y'know it's better to have one working than it is to have one shiny.

For Unity my wish list is, speed up Enlighten, Particles (needs to be supa shiny pwease *Precccious)*) / post (TXAA ) / cinematics and node based material editor Job done.

 

The reasons I like Unity are it's rendering flexibilities. I know, out of the box Unity doesn't provide you the shiny features, but it provides you lot of flexibilities on that side. We were able to re-create all lighting and shadowing stuff with ease. Doing everything in one single pass, lighting and shadowing faster than build-in. Well shadowing requires an extra pass but the decoding happens in the main pass. Rendering-wise, with Unity you got less features and more flexibility out of the box. With UE you got a lot more features/tools and bit less flexibility. It doesn't mean things can't be acomplished the way you want in UE4, you also have source code.

 

Yeah, for our next title, we threw out all of unity's lighting and redid most of the shadow code, the benefit is a tailored great looking game, optimised to the game rather than the game optimised to the unity defaults.

 

Hippocoder – I'm curious, how does one exactly go around learning how to write a shadow algorithm? Is there enough info in unity's docs to figure out where to start? Do you browse the internet for papers detailing certain optimized algorithms for certain situations and then implement them in Unity? This seems to be a really cool aspect of game development that I'd like to get into, but I really don't understand the workflow for achieving these effects.

 

There's fat zero docs about it. You need to first add shadows to a basic cg shader with vert / frag. Once you work that out, you follow the breadcrumbs from shadow collector / caster passes ie SHADOW_COLLECTOR_FRAGMENT(i) etc - generally just look for these in the cginc files. It's way too much to go into here.

Basically Unity renders info from the light source and this is projected/fiddled with and ends up as LIGHT_ATTENUATION(i) in your fragment shader. To see what all that does requires a lot of walking backward through cginc files to see what Unity does. Unity's shadows are really quite simple if you're targeting one platform, but there's a bunch of defs which define other things too - it's very spaghetti, covering various light types, hardware features etc, so it's understandable that Unity wants to streamline stuff.

What you can do though is cut all that out so you pretty much get depth and use that to project your shadow, and this turned out to be so much faster for my game (and higher quality since unity's fading wasn't necessary for our game and nor were cascades, so it ended up being better looking too).

I did a bunch of my own simpler routines and it works fine on all tested hardware I needed it to work on. As for lights I just do my own from scratch and ignore Unity's lights, mostly again for speed and quality. This is higher speed and more quality for *my particular game*. This doesn't make Unity's one bad, it makes Unity's one general purpose.

It's really not for the faint hearted, but for someone who's already pretty much mastered creating cg frag/vert shaders from scratch, none of that surface shader stuff will help you here.

The only reason I got that far and redid it all was because I really enjoyed doing it.

You can get an idea how to do your own shadow mapping by looking here: http://forum.unity3d.com/threads/ba...ios-a-la-bladeslinger-source-included.182682/

While I haven't looked at the source for that, I'm pretty sure he rolls his own shadow map so its a great starting point for others. Be aware its a lot more limited than Unity's shadows and your quality may suffer but I suspect it's good learning.

 

I have a question guys since it's not too off topic. I make simple mobile mini games as I'm in highschool and have tight schedules and all that jazz. How easy would it be for me to make a simple mobile game in Unreal Engine compared to Unity?

Btw my games are super simple. There isn't even a character to control lol. Basically you tap the screen when two things align (check my sig for the games). Would it be worthwhile remaking the game for Unreal engine? Reason I ask this is cos apparently Unreal compiles iOS apps for you (which I really need as I don't have a Mac).

 

Using Unity shadow macros wont improve performance (wont get it worse either). But it's useful if you do custom CG shaders to get buildin shadows on your shaders.
We are actually not using any of the shadows macros that Unity provides (besides explicit declaring shadowmap textures to get better hardware filtering) as our lighting and shadows are custom too. What we do is rendering from all lights (using cameras and shader replacements), and use the depth directly on custom shaders. Instead of adding macros to the shaders that will flag them to be rendered on the deferred shadowbuffer (this adds an extra pass to whatever is visible in your main camera) and is the one we are avoiding. We can also control how big shadowmaps textures are in a per-light basis. We've also customized our shaders to decode up to three lights (lighting and shadowing) in the main pass (which save extra pass for it). What's nice is that some scene lights camera doesn't render the scene, just characters (or stuff that moves).
So basically, what we have is a flashlight that renders everything within a short distance (opaque geometry) into a depth map, some scene camera lights that renders only characters (we enable/disable them based on main camera distance one at time) and shaders that does decoding of those two depthmaps + the lighting (most of the lighting is computed in a per vertex basis, only when bump is used we do it in the fragment shader).
That's how we run our game @25-30fps on iPhone4s/iPad2, with all post-process filters on (we got a bunch of them btw).

 

Really? Surely you would need Xcode to then deploy the app to an ios device?

 

That sounds need to be honest. Right now I don't have a mac and use VirtualBox to get past that. It works, but it's incredible slow to run Mac OS that way, it's like back to 1990's :S

 

Woops never mind it seems like it doesn't do that or something. Idek I'm unsure of where I saw it but I've seen the windows version deploy a finished app and people talk about it...

I guess I'll stick with Unity

BTW since you work for UT what's the possibilities of a payment to get rid of the start screen logo without paying for Pro? Would help smaller devs tbh

 

IKR! Jeez I wish it was true. Poor misguided me looool

Plan on building a PC at the end of the year though so hopefully it can run Virtual Mac pretty well. What are your specs?

 

And now something on topic from me. Unreal Engine is very hard to get into after using Unity to be honest. The way objects work in Unity is miles better than Unreal.

The updates and pricing of Unity could be much better though and that's where it shines in my opinion.

 

Specs pc: ( damn old processor, about 6 years old )

QuadCore Q9850 or something If I remember right
6 GB RAM
GFX AMD R9 270X

But hey it builds my XCode at least without a Mac, however, took about 30min of compiling a standard scene with 1 cube, 1 plane and 1 camera. Took me 3 hours to update Mac OSX, download XCode, Update XCode and open a project.
However I didn't test putting the App on the store. So I don't know if Apple can see if it's build inside a VirtualBox. If they can't, there should be no reason for not doing it.

All this was done for research, nothing else fancy.

 

Are you using the Free version? If yes, you couldn't get cheaper pricing.
And even with Pro, it is considerably cheaper (at a one off fee) than 5% of your Gross earnings (game sales, DLC, iAP, Ads, Kickstarter money, investment, everything!).

Updates probably take a bit longer because our developers like to groom and sculpt the engine's new content more into release-ready products, as opposed to features that have zero internal testing and are unstable. Then, once released, these features can be iterated upon at a controlled, and tested, manner. But, we also do have Sustained Engineering who ship a patch fix build almost every week to take care of last-minute unexpected bugs that may pop-up!

 

Doesn't windows version of unity also make ios builds now

 

W00t?????!!!!

 

It only makes Xcode projects. You still need a mac to deploy

 

What he's saying is UE4 can deploy to your devices directly to playtest however you still need XCode to do an AppStore submission but for development and testing purpose (which happens almost daily) it's quite nice to be able to deploy to your device directly from windows. You don't need a mac everyday to send your game to the AppStore.

 

It is the AMD Phenom II X4 9850. It has hardware virtualization so you might want to verify VirtualBox has it enabled for the virtual machine and verify that your BIOS has it enabled too.

 

I'm not sure, but thanks for the info. I'll try that out, because Mac OS run incredible slow right now, a bit too slow to actually have it as a solution for publishing iOS apps.

 

For iOS platform, UnrealBuildTool use RPC protocol transfer needed source file and compile command to Mac，then UnrealRemoteTool will parse the compile command and launch clang to compile.
Prerequisite:
1. A Macintosh running Mac OS 10.9.2 or higher, virtual machine is OK.
2.With XCode 5.1 and Command Line Tool installed.
3.Change the Mac OS host name to a1011(UnrealBuildTool will searching for this host).
4.Running UnrealRemoteTool on Mac OS.

Seems you still need a Mac in UE4, they just send a RPC call to the mac from windows to do the build...

Starting to wonder if UE4 sales people are started threads in the Unity forums, as it seems odd that UE4 users would be posting here so frequently considering UE4 has there own forums!! Either way it's a bit sad I think...

 

With UE4 you can create Blueprint only games on Windows and deploy to iPhone from Windows.
And to send it to Apple store a very slow VM is no problem, building on Mac is the problem on slow VMs. And the VM is only needed after you finished your game and want to send it to Apple.

 

Their release notes are impressively big.

Considering the scale of UE4 and tools it offers, amount of fixes and improvements, as well as release rate, are light years ahead compared to Unity's.

The reason why people here posting so often about UE4 is simple. I'll took myself as an example.

I've spent several month studying Unity, trying to learn how to make some mature (in terms of internal mechanics, programming) game with it, I've faced many problems, some bugs, situations when something wasn't working and nobody gives a damn, some restricted features I can't use so I have to invent workarounds, but I thought "well okay, problems can be solved after all, and it is still decent engine, I can do some serious stuff with it". Then UE4 was released and I saw how ACTUALLY game engine can look like, what tools ACTUALLY game developer can use, and how ACTUALLY engine development can be paced. And they say "here, take it and use any feature without restrictions, you'll pay for it when, if at all, you achieve some success". And I feel like I have totally wasted my time studying Unity. And this feeling is only growing with time and each UE4 update (and lack of Unity updates).

 

You do know that Unity has been doing weekly updates? Or did that not facture in?
Yes you are right. Unity can't be used to produce games at all. Nope. It can't.
ue4 will rule the world since because it is so awesome and nothing is better.
ah ya ok.....

 

What's going on with this Unreal Motion Graphics thing?

Two big players becoming closer together is never a good thing. Competition is what keeps the market strong

 

UMG is Unreal's widget based HUD / UI system, it works .. What do you need to know?

 

is there any chance you'd know which one is better for generating levels on the fly by instantiating models and making meshes in code vertex by vertex to make up objects in the level and the level itself? Pretty much all of my games now, no matter what style, use prefabs to build a level and only ever use 1 scene. Is Unity or Unreal better at creating stuff on the fly? I'm probably going to stick with Unity either way until Godot gets big, but I am curious!

 

UE4 is very good at procedural generation at runtime, also it crunches polys like no tomorrow.. A lot of the previous scene limitations do not apply. Via blueprints it's also very simple to do, so yeah UE4 is very very good at it..

 

Thanks. I'll consider it in the future, but I'm very excited for Unity's...

PBR shaders
improved networking
improved cil

+other stuff planned for the 5.x cycle. It's also costing me slightly less to keep up to date. $600 to upgrade early for unity every year or 2, but ~$20 monthly for UE. I prefer to own instead of renting. I hear you can buy 1 month and cancel to go out of date and just buy 1 additional month in the future to get a handful of updates... but there's not way that's intended.