Unreal Engine 5 = Game Changer

but don't you want 10,000 spinning cubes in your game?

 

Atomontage is probably just 1bit voxel anyway, you can encode a 2x2 morton brick as a byte, so you can make fast comparison/modification by using bit trick, and each bit is also the (implicit) key to a dictionary that contain the rest of the necessary data (like colors). 64bits can represent 4x4 brick array, that's a lot of voxel to modify in a single operation. Also it probably compress very fast and well, just check if the 64 int is = to 0 (empty) or 1 (full) to skip fast 4x4 brick, OR use simd casting to do the same on subbrick, building lod using the same trick would also be fast, just write everything that's not 0 as 1bits in the lod brick, write a cache friendly mapping and the throughput will go out the roof (a minimal 16ko cache can hold 2048 voxels). Good luck animating that.

 

You're thinking in a completely wrong direction.

Atomontage is not 1 bit voxel, because surfaces have color all the way down to interior.

Additionally, technically you'd want voxels to store its density, if you plan to make smooth surfaces with lower resolution grid (see Planet Explorers for practical examples).

Atomontage works on GPU. Bit tricks do not necessarily make sense on GPU, which is geared towards floats and do not help you much when voxel grid is not axis-aligned. Likewise, working on 64bit numebrs on GPU is not necessarily the best idea, and you gain no advantage from bit operations when surface is deformed.

Morton cubes would not be a very suitable data structure when voxels are destructible.

Also, atomontage supports skeletal animation.

Spoiler

https://twitter.com/atomontage_com/status/1158687767641149442

 

I usually write off most game engine demos, because most demos have been designed more like non-interactive movies just to show carefully prepared visuals. This UE5 demo actually looks like a real interactive game. It is very impressive.

And the PS5 is also quite impressive. It will likely outperform most current PCs on Steam with ease. Additionally, the PS5 will have the added benefit of being a known target platform, while each PC is a unique combination of parts.

I can easily see how UE5 (and PS5) benefits large AAA studios, who have hundreds of artists generating extremely detailed content. But I do wonder how useful this will be for most small indie devs. Most small indie devs won't be able to generate all of the art assets to build an ultra realistic game like the one shown in that demo.

I even curious how many games total will be made available at the detail level shown in that demo. It will take a lot of investment to deliver that level of detail throughout an entire game, and that will even seem like a lot of money to large AAA devs.

 

As an indie, I can buy one high poly model and use it 500 times. no problem!

Maybe unreal needs a rebrand though.

'The Brown Engine' how about that?

 

Yeah, that is valid. I love NVMe style SSD drives and I constantly recommend them to everybody, but a lot of PC users don't have them yet. In fact, most of the PC users on Steam currently have specs well below the PS5 in terms of CPU, GPU, and storage. Building a game for PC usually means trying to be as flexible as possible to support low end, mid range, and high end specs. If every PC gamer already had an NVMe drive, then we would be able to directly stream assets from the storage system during gameplay like that PS5 can do.

 

Mate I think u are looking at it wrong. The features of UE5 benefit indies more than AAAs.

Think about it. Until now, super high quality assets were out of our reach because of the amount of work and funding required to pay/generate it.

Nanite allows us to skip retopo, polycount, normal texture baking
then realtime GI allows us to skip baking

That already is two or three people's work no longer needed for a project.

That means the required resources to achieve the AAA levels of graphics have significantly been reduced, it benefits indies more than the AAA studios.

Also, even if you are not working in photo realistic graphics, imagine pumping out stylized assets with highpoly sculpted meshes. Just sculpt and texture and boom...

Really, this is a game-changer for indies and bad news for AAA. They can no longer maintain AAA status just with money anymore...or at least make it harder XD
The gap that naturally incurs from the project budget is slowly decreasing.

 

You do two pass, once you filtered the useful data you can do extra work, there is magnitude less data after filtering (for example screen visibility is n² instead of n^3). Atomontage is discrete anyway.

Last time I checked they were on cpu anyway, I didn't looked at later solution, but the follow up of voxlap (pn3d http://advsys.net/ken/voxlap/pnd3d.htm ), by Ken Silverman do all on cpu too, it get decent fps. Also I assume axis alignment. Morton is what's used in all the fancy SVO gpu solution, they all use bit trick too. I helped someone designed a RTGI (nigiri) also used svo bit trick using compute.

.... also I did a sloppy proof of concept on blitz3d, but blitz is a performance joke, that doesn't count ...

 

THIS IS A GAME CHANGER!!...Incredible progress, I don't think every one is grasping the significance of this tech. With Unity people are still waiting for grass on terrain with HDRP, I mean there really is just no comparison to Unreal Engine.

Let me remind you with Unreal you also get access to Quixel Megascan library, the demo shown would not really have taken long to do as the engine handles all the work for you, No LOD levels required, No Normal Mapping, No Lightmapping. Plus multibounce Realtime GI, not to mention custom Physics Engine.

Unity right now is extremely fragmented with no clear direction in sight (Everything works so so, half complete or buggy) Even ECS/DOTS is confusing and good for parallel computation only, also no clear documentation. Unity should drop all the fragmented pipelines and do a single clean engine, but even than I don't see any cool tech like this shown coming to Unity anytime soon regardless of the direction they take.

 

To really compete with UE5, Unity will need to figure out some way to get all of the preview and experimental features polished to production ready status. For example, DX12 support has been unstable in Unity, and most game devs have simply worked around that problem by building for DX11. However, Unity will actually need completely stable DX12 in order to build a product that can compete directly with UE5.

In addition to that, Unity has been painfully slow to fully implement 64 bit support. Resources in scene files still have a 32 bit limit last time I checked. I did additive scene loading to work around that in one project, but that is obviously not ideal for all projects. And obviously UE5 in not clowning around with 32 bit limits on resource files. In that UE5 demo, they were directly streaming 64 bit resources from the NVMe storage to the APU memory, and that APU can directly 64 bit address that data from the CPU and GPU sides.

The first step toward competing with UE5 involves going full war room against all of the "permanent preview" features that already exist in the Unity engine.

 

Incredible stuff, if this 'just works' it is the real answer for removing a couple of the biggest obstacles in indie development.

 

Well Put..

 

I see what you are saying, but I think AAA will still be far better suited for generating massive amounts of ultra high detail original artwork. Both indies and AAA will have access to some of the tools, and that is great. But indies won't be able to create hours of gameplay based on original artwork at the detail level shown in that UE5 demo.

Imagine if every NPC has the visual level of the main character in that demo. Image if every house, mountain, cave, rock, tree, etc. had that level of detail. Then imagine a team of hundreds of the best AAA artists building those assets at a large AAA studio, and then compare that to how long it would take for a team of 4-6 indie devs to build those same assets.

 

It is very savvy INVESTMENT. In next 10 years global game business will be generating $200-300 billions in revenue every year and Unreal will be getting 5% of it year, after year, after year...

It is getting really hard to say no to Unreal - "free" AAA engine, free tools, free assets, free AAA material library, free web services, great distribution platform with small revenue cut, generous grants, great publishing deals... It all happened in just one year.

Meanwhile last year in Unity feels like

 

I see where you are coming from, but there will always be that "gap" when there is a massive difference in the budget. That...my friend...is just the sad reality of life

I am just saying that we dont have to sell our kidneys.

 

You know what's big boy business? CAD, autodesk still dunks on epic's fortnite money with CAD. AND GUESS WHAT, putting CAD into unity is a multi step process that was praised up until this morning. Because the nanite mean now you can just plug the CAD directly in unreal without multi step process, and it would just work, with instant real time GI lighting, unbothered. ALL that unity effort of complex multi step CAD import is now not looking so great.

This is a blow at so many level, on so many industry, one focused blow.

 

The irony is that Epic would not even be able to keep it half busted in permanent preview, because it is open source. A bunch of coders would simply polish it to production ready if Epic tried to release it in a half busted state.

For Epic to really be like Unity, Epic would need to close the source code first, then release each separate new feature in a half bust permanent preview state for several years. Next, they would need to let developers vote on which busted features needed fixing the most. Then they would need to have a bunch of internal meetings to try to decide which busted features deserved fixing. But before they fix anything, they would need to remove some of the stable working features with no ETA for the replacement features.

I absolutely love Unity, but I will be honest that it has room for improvement right now.

 

This sums it up really well.

 

The frustrations I have with Unity today are , in my opinion, directly connected to its leadership, specifically Unity's CEO who has zero leadership skills and made Unity far worse than it used to be. Under his broken leadership, Unity has become less and less performant, has more "preview" features than "production-ready" features and made very insignificant strides in technological advancements compared to its competitors. Even petty things like "dark skin" is still a restricted feature because god knows why.

I guarantee that if Tom Sweeny was Unity CEO, the state of Unity would have been 100000% better. At the end of the day, a product is as good as the person who leads the team who builds it, and sadly I have zero faith in Unity's CEO.

 

Hi there! Just a 2 cents.

I too have been thrown on the floor by the presentation. And Unity I'm sure you are too. I have only good things to say about you/Unity. It could be so great if you could find a way to add some features like that in the next versions; like say in Unity 2021-22 to get Micropolygon technology (voxel like tech) than changes the old polygon problems, and make us have the possibility to have 'virtually' limitess amout of polygon...right now most developers have to use tricks to hack it and make it work (I mean for people making large games with heavy assets), must use mesh combining or other tricks; they said : ''no more LODS, no more 'polygon limits', it can render Bilions (with a b) of polygons, something that only Offline renders can do (Even they struggle to do that/fast I mean...it's not realtime), this realtime microtesselation that would put any engine to its knees...you made Megacity Demo to show the power of ECS/DOTS...now must go for technology that will make AAA games possible for smaller indies - like this UE5 engine will do. The removal of the costs to pay for releasing a game is like the double whammy (1$ million window befoer neededing to pay cost of using UE5 engine; most developpers do not reach this, I mean indies)..it mostly is a huge advantage for indie developers...AAAs would benefit too; I think that AAAs will become AAAA (4 AAAAs), not just AAA/3 As anymore; that'S impressive yet scary in the same; it means AAA games, Becoming AAAA, will be just nigh-high impossible to reach; but At Least, we indie devs will be a ble to reach A or AA...so that'S would be a good thing. Unity look into that it could be a very important changer of things because there may be some devs who feel 'starting' to make games now seems more attractive by these things shown from the UE5 presentation. The Heretic demo is close, not so far behind, but, this micropolygone technology is Really something (I used this when I would render in renderman (REYES, later RIS) offline rendering for CGI VFX...the fact you can 'stream' billions of polygons and render Quixel Assets megascans like its nothing (and keeping all full resolution) means that UE5, in the future, will become an engine Used by CGI Films VFX (it's alreasy happening, but iwll happen More and will slowly Replace (well not replace but supplement and become, in some points, 'main render engine') supplement Offline renders that take an eternity to render; because of realtime). Realtime should be Everything in unity (I know that is wishful thinking/dreaming in HDRcolor).

From UVs, Rigging, you name it, the more realtime automatized it is the more we speed up this whole creation process and we can 'fine tune it' to keep an artistic/control of it as to avoid 'canned feel/non-human made (computer made)'. Asset creation is so long we need to make it faster; as some others said; there is also the thing that because it is faster/better; it measn that the Bar is going to Increase...which makes it harder..but at least indies will reach A or AA faster now...but AAA...will Also increase to AAAA. AAA will take advantage of this techonology Too. And they will Increase tbe Bar/Standard of quality of games...and it's a double-edged sword..it improves things/visually but makes it harder to attain that Ultra Quality that AAAAs will get to (after getting UE5); thus, people/games may be 'less' impressed by A or AA games made by indies using UE5; because now they see AAAA gamse are made with it. It almost always Goes Up for everything in terms tech improvement. All the industry will be helped. The counter effect is that the Standard will be nigh high...this is obviously tech demo like Heretic, Book of the Dead and Adam.. they are not necessarily what (most) games will look 'in game'; this is ultra-quality that is very hard/if impossible for indiees to reach; I mean AAA assets like Quixel scan help...but can you fill this much data/make a game...only 'Nanite' micropolytech could do this..you Unity needs something like this. Also that GI/Global Illumination is so good that it makes RTX almost futile now (it is not) but like why go for RTX GI if it so expensive in frames per second when this 'Lumen' technology gives Realtime GI that is a 'in between' Baked GI and True Realtime GI; Unity needs a Hybrid GI that will do llike Lumen. This would sove the 'need for RTX Raytrace lighting', it is a solution but a costly one (in terms of CPU/GPU), Lumen is less expensive and can render Full 'unlimited bouncing' GI...(!). I would take that for my game (I am doing a FPS shooter game..I need that...but right now I fake it using tricks/Cinema film tricks; would love true GI like that; otherwise must do baking (baking is very long/tedious) or use GI scripts like Sonic Ether GI (a true realtime GI, which is great but vEry costly in frame rate))) so I just fake bouncing of lights using 'under floor lights/bounce cards' like in movies' studios...If Unity you can increase the amount of Window/monetary cost, like something along to what UE does now with 1 million window for royalty that is major thing. IT would be incentivizing people to decide to stick around, and though revenues are important for you, the More People Make Games, the more Unity wins...it's winwin, UE did so in that optic, they said: ''We want to offer indie devs more lax..to not freakout lacking funds to make the game and release it/if make less it is not as hard because they never reached 1 million in revenue''. Also, Unity, I agree with other, does not Need to chase all of that...but it would be a Good/Smart move...to. you can stay with smaller market mobiles and small indie...but to go at A/AA/AAA this a move to consider; because some devs are making A or AA games As Indies and would love to do it on Unity 2021-22 (but UE5 is tantalizing/temptation with so Many Perks/Advantages..you would too/go for it if you were a struggling solo/micro-team indie dev). Thanks for reading.
Just a 2 cents.

 

WHY CAN'T WE GET 'dark Skin' What is Unities justification? It is childish behavior to be so stingy..

 

Let's not blame anybody specifically.

 

Do keep in mind that "going to a real life location and scanning it with your phone" is currently an option. You also can utilize drone footage for larger scale reconstruction.

So rather than building from scratch you should be able to "borrow" real life locations.

 

Epic is Proprietary/Restricted Code access and not opensource.

If you start borrowing pieces, you'll likely get a message from their lawyer team.

 

Dude, Epic was almost going bankrupt and got saved by a chinese conglomerate.
Then they managed to make one of the biggest gaming hits ever created.
Now they can pay a crazy amount per user acquisition and they are giving free stuff around.
Unity is pretty much the reason why UE got to this new model (and not only UE), basically Unity was one of the motors of indie gaming worldwide.
Hell, look back at 2018 when unity showed HDRP and DOTS, we were all thinking that Unity was going to be the default engine of the future (yeah, we didn't think that it will be 2 years later and they are still working on an input system for a GAME ENGINE).

I won't say that Sweeny is bad at all, with all that money they could just shove it on publicity, instead they went straight to the user, so kudos for that (though they having all that money and becoming a pseudo monopoly does gives me shivers), but I don't think anyone could foresee the amount of money a game would generate for them. It's easy to speak with the newspaper on hand.

 

Proof?

It is not a publicly traded company so there's no financial report. There's deleted angry rant on reddit which prophecies doom of Epic Games, but that's it.

 

The strange thing is that realtime global illumination and avoiding LOD baking is exactly the sort of thing I would have expected Unity to focus on, being apparently aimed at indies. These are literally the two things that for me as an indie cause the most pain and take the most investment. And this is the kind of direction you go when you look at things from first principles rather than iterating on assumptions - i.e. why should I have to bake lighting at all? Why should I need to make LODs at all?

We've yet to see what this is going to be like to use, but the focus and intention of Epic in this case is incredibly well aligned with what I'm looking for.

 

To what end? Bigger game worlds? Moving faster through them? Higher-res assets?

 

Here is a reaction video and Tim Sweeney and Epic CTO talking about the Tech a little more. I can't imagine Unity CEO answering questions in a proactive manner like this , EVER!

 

Sure, all of those things are possible with ultra fast storage.

 

They were struggling to get fortnite out, they cancelled unreal tournament, they changed the engine monetization model, they gave out 40% of their company.
I'm not sure, but at the time there were many news sites saying how tencent "saved" them.

 

Well I will say this stuff tickles my imagination at least. I'm really curious to see this stuff in the real world development environment.

Import a model from zbrush at full poly count and cut out all the obnoxious prep steps of normal maps and LOD and all that? Sure, it sounds incredibly fun to be able to do things that way. It also sounds like they're selling us snake oil all the sudden because there's so many technical questions in doing something so radically different, with zero technical details yet.

Also realistically this stuff all sounds like it is locked to PS5 and Xbox Whatever It Is Now for several years at least. We have dinky, slow SSDs on PCs and no mobile game is rocking high end anything. I'm still curious from a technical level how the nanite rendering works though, and what whether any features really are such game changers.

 

I am not talking about open source in terms of people being able to use the code for other stuff. The licensing restricts that. I am talking about coders being able to get into the engine and fix stuff directly instead of trying to convince a company to fix it someday. The vast majority of people won't do that, but there are some who will.

With Unity, there are some bugs and limitations that have been there for years. Game developers repeatedly run into the same snags. If Unity allowed code access to everybody, then some of these bugs could have been quickly fixed by coders in the community and then submitted back to Unity to benefit everybody.

 

Fair enough. This is certainly a thing, it is possible to submit a fix to Unreal engine.

 

i am reasonably new to unity and definitely not a graphics person, but can someone explain what is difference between this and something called 'book of the dead' or 'heretic' by unity.

It looks same - maybe even better, and accomplish many years prior?

 

This is very conflicting for sure. In UE4 the artistic workflow was already leagues above Unity, and this cements that even further. Now it will be easier than ever to create photo-realistic graphics. However, I don't really see the point in such detailed graphics, at least for indie devs. Not only because of the money/effort required to achieve them, and the hardware demands which haven't been made clear yet, but because they detract from the whole point of video games: gameplay. If I want such stunning and realistic graphics, I'll go see a movie. Better yet I'll go outside. But when I play a video game, I want an experience that is fun, engaging and unique. That's the reason I and many others I've talked to have lost interest in modern games, because they focus so much on visual fidelity while being less fun than old games. And there are still many 10+ year old games that are visually impressive enough for many people anyway.

Although it will take a while for Unity to even remotely catch up to Unreal's artistic workflow, I think in the mean time it should focus on retaining the edge it has in its gameplay/programming workflow. After working in UE4 for years I found C# to be a huge relief from the painful dichotomy of Unreal's C++ and Blueprints. This can go even further with HPC# and ECS, as long as it becomes production ready and streamlined soon enough. And of course making the engine more stable and upgradeable will be essential to allow integration of third-party packages, which is also something Unity has a big advantage in. However if UE5 introduces a more convenient language similar to Unrealscript and allows easier customization than UE4 it will be hard to turn away. But if Unity gets its act together I think it will still be preferred by indie devs.

 

Is unrealscript similiar to unity's cprogramming? If so it will be really useful

 

Not to get carried away with comparisons, but frankly which one of these actually gives a proper sense of distance?

Spoiler

Don't get me wrong, the Heretic is pretty but Unity has a way to go in achieving really comfortable visuals in my opinion. In some ways it's incredibly realistic and in others just comes across as lossy and over-crunched.

I say this as someone who would like to achieve a balance where I have less detailed artwork and high fidelity lighting and graphics presentation.

 

Yes Unrealscript was nice to use in my experience. Pretty much C++ without all the annoyances. Only problem was it was hard to debug. If not for that I'd even prefer it to C#.

 

Neither. It is a still, to get sense of distance, character need to start moving. However, first screen resembles a diorama with a greater degree.

 

Thank my friend tried unreal last week for a day and said it uses something called 'actor' and 'pawn' but it was not viable to make full game using blueprint even if exposed. Is it worth learning 'unrealscript' ?

 

Scale is probably a better word for it, though that does include distance. I agree about the diorama effect, and I think there's a clean break between foreground and background lighting as well that probably contributes.

In any case, if Unity maintained a reasonably good GI along with such things as automatic LODs, the fact that I know it well would probably be enough to keep me here regardless of comparisons. Because workflow is what is going to make or break my game development. Epic are focusing on the right things here, at least as far as indies are concerned, and Unity need to do the same.

 

Unrealscript is from UDK, not UE4. Actor and pawn are classes from the gameplay framework, which you should definitely use only in C++. And that is exactly what made me move from Unreal, the gameplay framework (which is inheritance based) is such a headache to work with if you aren't making a cookie cutter game. And with how bloated the classes are you pretty much have to rewrite them yourself, which often breaks every other dependency so you have to do those from scratch too.

 

Let's see what's going on in the UE5 forum thread
https://forums.unrealengine.com/com...sion/1760015-ue5-is-coming-~-what-do-you-want







The Unreal Engine community either doesn't seem that excited, or I didn't find the proper forum thread. There are a gazillion more replies here regarding UE5 than I was able to find at the Unreal forums.

EDIT: Thanks @Metron for the link to a more exciting thread at the Unreal forums:
https://forums.unrealengine.com/unr...l-engine-5-with-free-wallpapers-gifs-and-more

 

To them, it's just a renderer update. They have tons of other to get excited over. Free this, free that along with many sexy features. Numbness from the excitement overdose.

One thing I'm dying for is the new scripting language that they are working on. It is the day I'll go back and I'll keep unity for just clicker style games.

 

Another thing is there are tons of non-programmers there. My experience with unreal community forums wasn't very positive.

 

Fanzy stuff indeed. But it's not relevant to me... After all, my projects just need good performance in outdated hardware.

 

Well...

 

If this works and gives same or better performance than manual created topologies and LODs then it will be a game changer. I hope it will be! So it will not be another Todd Howard saying it just works

edit: not to mentioned a realtime GI that just works. Without lightmaps. I wonder how it performs in VR