Incline Unreal Engine 5: final nail of Unity's coffin?

Well yeah, that's the other side of things. The more AAA the game, the bigger the budget, the more constrained by PC cult box-ticking it has to be, the more bug people are involved, the more awful it is. So all that lovely graphics potentially ends up facilitating the mass production of highly polished turds.

 

This was true for earlier versions but in UE4 the signature Unreal Engine plastic look was gone.

 

Unity game engine is for retards and chumps who afford a fried video card/motherboard.

 

Just for people asking when games will look like this demo.. just a quick note: the statue is said to have 8 chunks, each of them using 3 8K textures. That's 24 8K textures; at 4 bytes per pixel, that's 8.5 GB of textures only. Even with compression we're talking of a few GBs for a single statue. Considering the ram limitations of the PS5 or the SSD space, it's very unlikely that in a typical game you'll find assets that are that detailed.

 

Who cares. I still think this looks more charming...



...than this boring UE5 demo, because, well, it actually looks like game graphics.

 

Tbh subsurface scattering is one of those effects that would take a lot of work to make it do what it really does in real life. Have you ever shined a strong light through your hand ? it starts giving off an almost translucent vibe and the stronger the light the more you see your actual blood vessels. It should look like this :


Subsurface in general is responsible for the color of our skin. Skin color is a complex phenomenon and it shifts quite a bit under different lighting. Even games that say they implement subsurface don't really show skins that behave the way you'd expect IRL.

 

FWIW Nvidia's Luna demo from 2005 had some nice approximation of subsurface scattering. Though for some reason, it didn't catch on:



Then again it also has a nice approximation for long hair (even some rudimentary physics, though that was improved in their later demo) which also didn't catch on :-P

 

I can give them that UE5 can make some nice realtime movies. Hollywood and Animations Studios I am sure will be excited.

The only interesting thing about the Demo is when the character zaps around fast, although I am not sure how much that is a lie since in normal games data needs to be streamed from the harddrive and from memory to the GPU, that's a physical limit not a software one.
Preloading all that is nice and dandy but real games aren't as predictable.
However we might have some nice driving or superhero open world games if that really improved.

I also expect a Half-Terabyte or more games soon. All that fancy photogrammetry,textures,materials and zbrush data is going straight to your hdd.

 

Although I wouldn't mind it if the game did not download the assets you don't use anyway. But then again most devs would never bother with this, unless steam etc. started charging for bandwidth used.

Still this would only affect AAA so it is not a huge deal, most AAA is shit anyway.

 

That Jane Nash body movement circa 2008 3DMark Vantage! Oufff! You guys are easily impressed or wayyy too young. First, consoles are not going to be doing this, this gen or next. Consoles are powered by tuned IPC-deficient low power AMD CPUs with buncha shitty Radeon magic bolted in, and that's talking about current gen stuffs. Next gen consoles? AMD CPUs with buncha shitty Radeon magic bolted in v2.0.

Second, it. does. not. look. that. great. Not for 2020, not for 2015 ffs. Blur effect is the new bloom I swear. One blinded you to the poor/downscaled textures and low-poly stuffs in the background, the other LITERALLY blurs the poor/downscaled textures and low-poly stuffs in the background.

And again, that quality Jane Nash animations! Anyway, crank up Crysis-1 on PC to ultra and experience just about this. Only before even Obama was a thing or half the Fortnite zoomer brigade was even born. Tech-demo? More like demo of tech stagnation.


...
..
.

 

Sadly indie games do that too. There's the narrative that game graphics are great nowadays, but this is not the case when everything looks identical.

 

I saw this thing on the making of the recent Mandalorian tv show - the environments are basically UE environments on gigantic screens behind the actors, and they move appropriately to the actors' movements, so both the UE environment on the screen and the actors in front (on smaller real sets), are filmed in realtime. All the SW desert/jungle environments in that series are basically UE.

It's a new thing apparently and it's probably going to be a thing in a lot of movies now. Makes it easier for the actors to feel like they're in the weird and wonderful space, and act naturally - much better than green screen. Also, lighting and reflections are handled automatically (e.g. the scene from the screen will reflect on someone's armor, for example). Very clever.

Don't know whether they're UE4 or UE5 environments, but I'd guess, given the fact that the rockiness aspect looks very like this UE5 demo, that they're more this sort of tech, maybe a prototype specially made for purpose using some monumental hardware?

Ah, here it is:-
https://www.youtube.com/watch?v=gUnxzVOs3rk

 

They aren't going to film that plaster white cylinder with contrast ratio in single to double digits. Lmao.

Though they definitely wanted the audience to get that feeling, from the snippet I saw - showcased some crazy editing skillz. Dunno what use for such falsity, though Epic is no stranger to smelling their own farts either. Cougghh.

 

I am not 100% convinced.

For starters, I am not sure how they are going to animate characters with billions of polygons and neither do they it seems. If you pay attention, the character is low poly and stands out in a bad way in close-up scenes, probably also because she is slightly stylized. By animation I don't mean rocks falling or stone moving. I mean deforming joints and so on. So it might be great for Environment Art but maybe not for characters.

I am sure the game engine is capable of rendering billions upon billions of polygons but just because it can do it doesn't mean it's feasible, at least not right now. High poly files are very large. I am a 3D modeler and my decimated Zbrush exports of 3-4 million triangles are in the 500 MB range. Multiply that by the thousands of objects in your game and you will get OBSCENE file sizes for just one game AND THAT IS JUST THE MODELS! Those 8k textures will also be quite large. Then there is also sound files and so on. I am sure they will optimize the file sizes but expect the games to be huge, maybe even terabytes. This is why I think that "No more Normal Maps" is technically true but practically not. Normal Maps can still be used for miniscule details to reduce the file sizes. Additionally, in more linear games, I suspect low poly models with normal maps will still be used for objects in the background to further reduce the size. No need for 1 billion triangles for an asset you will never see up close. So for the early cycle of next gen I suspect we will see a combination of Low Poly with Normal Maps and straight out of Zbrush models for Hero Assets. At least this might be the case until we have a major advancement in storage capabilities and by that I mean also affordable. I know Sony is making strides in this regard but we'll have to see how this goes.

Finally, I haven't seen any new gameplay capabilities. I am not sure all developers will want to completely restructure their workflow just for graphics no matter how good the games look.

Just the opinion of an intermediate 3D modeler.

P.S Think of the bandwidth required to download these games in an era where distribution is moving more and more towards digital.

 

Might have something to do with Unity still being single-threaded.

Fixing that has been a multi-year project for them.

 

The real advancement here isn't the eye-candy. That was just PS5 propaganda because they needed a way to say that the graphics will be okay even though it is less powerful than the next Xbox.

The advancements as advertised are:
-no need to bake lighting: faster iteration time, better looking environments for less work, dynamic lights look nice when you want to sell a game

-art pipeline improvements: use the same model in-engine as your editing program, the machine will do half of the technical artist's job

-cheaper level art: use our ultra-high resolution texture library for free, tell your artists to fuck off when they say they need to go outside to capture reference material

-cheaper multiplayer: use our online services for free so you don't have to build your own before you know if you are Anthem 2 or not

-only pay us if you make a million dollars: since San Fran indie devs are terrible at math we have eliminated the confusion about whether Unity is a cheaper choice, this will also make it cheaper for us to hire you to work on Fortnite later once you have paid for your own training on our tools


In conclusion, this is going to put Unity in a tough spot. And any studios with a few Unity releases. But unless each of your releases is guaranteed to gross $1m, paying that Unity monthly bill is going to feel really stupid with each passing month.

My hope is that other smaller engines will take this information and focus on becoming the best 2D engine or even genre-specific like RPG Maker. For them, that false hope of ever catching up for 3D should be done with now.

 

What other smaller engines? AFAIK the market for licensed 3rd party 3D game engines has been pretty much cornered by Unity and Unreal for more than a decade now. Even Monolith/Touchdown, which used to sell LithTech to smaller studios, stopped licensing their engine since the late 2000s and Torque, the closest equivalent for indie games, was abandoned and eventually got open sourced years ago (though didn't saw much use despite being released under a very permissive license). There is also GameGuru but that one is focusing on ease of game development more than graphical fidelity, so it isn't exactly competing for the same audience.

Am i forgetting some engine?

 

There's the Ogre engine which sees some use (Rebel Galaxy, Kenshi) and the Godot engine which probably sounds more novel than it actually is. But yeah, Unreal and Unity are here to stay.

I'm pretty sure there was another engine though... Crie me Cry something. Who cares.

 

Those are free open source engines though, they're largely made by people whose main purpose is making these engines (as opposed to working on games or products) and have them under a free/open source license, so if anything a proprietary engine like Unreal getting some novel tech is going to make them want to have that or similar tech too.

And yeah, i forgot about CryEngine... and probably i'm not the only one :-P.

 

The interest in a lot of smaller engines was driven by their lower cost. Back in the day you couldn't even get started and look at the source code of Unreal until you forked over half a million dollars.

That argument is out the window now. Unreal tools and source code are free to tinker with until you make a million dollars. Along the way you learn industry-standard tools for employability.

Unless your dream is to make an engine there isn't a strong reason to learn on something other than Unreal as a general purpose engine. Hence the need for smaller engines to become known as "XCOM maker", "RTS maker", etc.

 

Sure, though FWIW these much cheaper engines didn't really lack in terms of functionality (like LithTech which started at $10K for the NOLF2 version, or for indies strapped for cash, Torque 3D was just $100 and was the engine behind Tribes, so it was proven tech - but other cheap engines existed as well).

I do not think it was ever an argument really and besides price is far from the only reason an engine is chosen, there are other reasons like licensing, ease of use, available resources, etc.

Eh, not as important as you might think. Even ignoring UE's own changes between versions (especially the UE3 to UE4 was quite a drastic change), there are many other engines out there and chances are you'll need to learn them anyway. Rarely is engine knowledge considered a requirement, usually you'll be expected to learn the engine and tools on site (an exception might be smaller indie studios that literally cannot afford that time... but chances are they wont be able to afford yourself for long anyway :-P).

I'm not sure how these follow, if you have a dream to make an engine, sure make an engine but not making an engine doesn't mean there are no reasons to not use Unreal: you may dislike its licensing terms (so, e.g., you'd prefer the more permissive Godot instead), the system requirements that Unreal imposes are higher than what you'd like for your game, you simply find Unreal's tools cumbersome to use (AFAIK Unity has much better CSG tools nowadays, which are great for Quake-like geometry), you prefer to work with C# instead of C++ or there are some features you'd like that Unreal misses (e.g. some gameplay feature that can be found in the Unity asset store). Or you simply want to have control over your technology, so even though you still have making a game as a goal, you also decide to make an engine for it - this is a common reason for why some development studios make their own engines instead of licensing others'.

And i'm not sure how you go from recommending that people learn Unreal as general purpose engine to other engines becoming known as "XCOM maker", "RTS maker", etc? Why not have those engines try to compete with Unreal so that there are more choices or try to focus on other aspects of development (e.g. ease of use, like the GameGuru one) while still remaining general purpose?

 

one major reason to not use Unreal is the compilation times
quickly iterating on a design is difficult due to this

 

Interested amateur questions: doesn't Unity still have an advantage in terms of its asset library? And isn't asset creation the most time-consuming aspect of making games?

 

That's what blueprints are for

 

Goodluck convincing a programmer to use blueprints.
Getting rid of unrealscript and not replacing it with a viable alternative was a mistake.