JoeJ said:
Maybe you don't need that anyway, but this also means the industry has decided that the old and easy high level APIs are dead. Intel already only emulates those old APIs.

Yup, a natural fallout.

Players want photorealistic worlds with tremendous complexity. Major game engines provide all the heavy lifting without individual developers needing to implement it. Click a checkbox and you get realtime, photorealistic dynamic lighting, shadowing, and effectively limitless view-dependent on screen detail.

The systems aren't designed around “draw a triangle" or “draw a cube”, and haven't been for many years. Instead the systems have changed to “draw this complex scene document that's longer than a bestseller novel." The consequence is that you build a system capable of rendering the complex scene, but instead of throwing gigabytes of data at it, you give it the document with your single triangle.

JoeJ said:
Can it do reflections as good as UEs SDF tracing? I have never seen such thing in a Unity game. I guess Unity also lacks VSM and recent advantages on soft shadows, but idk.

Technically speaking Unity is quite modifiable as engine, with SRP and other additions over the past years (but not just those - it was quite “bendable” even before) - there are SDF generators + tracers well integrated and implemented by 3rd parties. Same goes for various shadows algorithms.

The question is - whether this counts (I mean from users-perspective it definitely does … but it isn't developed by Unity Technologies)? I don't have much experience with UE - so I can't speak about how modifiable it is.

On the other hand I do have quite a lot of experience in Unity (and we had to make it work with custom ray tracing, integrate with in-house engine, etc.). I can't share that work though due to legal reasons (unless I'd talk to my partners about sharing specific things out).

…

Additionally, Unity does also implement HW Ray tracing for quite some time in HDRP for example. Feature-wise they aren't really as behind as people tend to portray them.

JoeJ said:
Yeah, they work on it, but afaik that's a fork which has nothing to do with Unity. And i have not yet seen that either in shipped a game.

It is Unity official version for China - https://docs.unity.cn/cn/tuanjiemanual/Manual/VirtualGeometry.html​ - this is the official docs for it and they also have some more details in https://developer.unity.cn/projects/64883165edbc2a116e4f941e​

I still have no idea why they haven't considered adding that tech also outside of China and what's their reasoning behind that.

There are of course other various implementations on Unity - at various stages of implementation and reaching various performance.

As for shipped games - one of the problems of Unity is that AAA studios just doesn't select it. Yes, we have Cities: Skylines for example (incl. the 2nd one - which has terrible performance) - and it is by all means a high fidelity game … still, most of the games in Unity are smaller.

For whatever reason decisions of Unity corp. doesn't seem to be in line with their target audience … which I consider much bigger problem than missing tech A or tech B (which is, in case of Unity, still supplied through 3rd party assets).

JoeJ said:
Nah. This is still dominated by SSAO, no complex materials just flat diffuse everywhere for most. Like a game from the 2000s but with hi res textures. They fail to take profit from baked lighting. There would be better examples, but i can't come up with one quickly.

I wanted to point that one out because it's kind of abusing an engine that's ages old, while still keeping the results somewhat good enough quality.

It will heavily differ on what kind of visualization/game you look at, and what you aim for. For this kind of games it's absolutely okay to have precomputed lighting, etc.

JoeJ said:
Now, i have played quite a few UE5 games already. And to me it's a noticeable generational leap visually. It's more than what we've got with the PS4 generation back then. Though, i never noticed any dynamic GI in any game. Never i can open a door and see how the light floods the whole room. I know it works, but only from demos on YT, not games. I guess the level designers just miss the opportunity so far, and have not yet learned to take advantage. Or maybe they need to exclude doors from GI, similar as with characters and most dynamic objects. That's something i really wonder about. But otherwise i'm impressed and also happy with perf. on my old HW. (Sadly i'm not impressed from the games themselves til yet, but that's another rant.)

So far I wasn't really that impressed with UE5 games, and that mainly counts for visual/performance rate (when you need to upscale the image - it is by far worse than PS4 era games).

As for the games - I have to second that I'm not impressed by the games itself. They can be visually beautiful - but that will only hold the game for so long. If they're boring, with shallow characters and bad story - the game is just bad (but yeah, this rant can be done elsewhere).

…

This being said - my most “regular” game these days is AoE2.

JoeJ said:
Personally i do not wonder if some devs switch from Unity to UE for better visuals. I wonder much more why Unity buys a offline VFX company, but refuses to bring their gfx up to date. It took them a decade only to get their TAA right.

The TAA bugged me for so long (among other things). They did (and still do) very … weird things as a company.

You have 3rd party unbiased renderers that integrate well with hw rt they provide. But they're unable to provide theirs? Even after buying whole company focused on such thing?

The worst thing on Unity (and again, I can't really compare to UE and comparing to in-house tech is unfair) is that their performance gets version-by-version more terrible. Every single change in C# code or assets ends up in progress bar where you wait for multiple seconds. That wasn't case 5 or 6 years back, now it is.

This one thing is objectively much worse than any of the visual things above - and it doesn't matter how much visual extensions they bring in - as long as they won't fix interactivity, it won't matter. Doing larger-scale projects in Unity is annoying and you're constantly fighting the performance of editor (I know that far too well). Sigh…

Vilem Otte said:
Every single change in C# code or assets ends up in progress bar where you wait for multiple seconds. That wasn't case 5 or 6 years back, now it is. This one thing is objectively much worse than any of the visual things above - and it doesn't matter how much visual extensions they bring in - as long as they won't fix interactivity, it won't matter. Doing larger-scale projects in Unity is annoying and you're constantly fighting the performance of editor (I know that far too well). Sigh…

This is my biggest problem with Unity and it happens even in small projects (<1 GB). It's a huge productivity killer. I sometimes wonder how much developer time (and money) is lost waiting for Unity to “Hold on…”. I think the editor is just very poorly designed and implemented. Even opening Unity Hub takes 30+ seconds, much less the time it takes to actually open a project. God forbid you need to re-import a large project. You might as well come back the next day (I literally once had a 18-hour first opening of a 35 GB project). Yuck.

Aressera said:
This is my biggest problem with Unity and it happens even in small projects (<1 GB). It's a huge productivity killer. I sometimes wonder how much developer time (and money) is lost waiting for Unity to “Hold on…”. I think the editor is just very poorly designed and implemented. Even opening Unity Hub takes 30+ seconds, much less the time it takes to actually open a project. God forbid you need to re-import a large project. You might as well come back the next day (I literally once had a 18-hour first opening of a 35 GB project). Yuck.

It's a massive problem and I have no idea what they made with their editor to become this bad. I remember that with 5.6 or so, it wasn't nearly that bad.

When they changed their versioning to years, it started getting worse, but after 2019 versions it became unbearable.

…

I haven't tried their Unity 6 so far (I'd like to at some point during this or next month).

…

I can't really compare that to other engines - because I never made anything more major than test projects in UE, Godot or others. Apart from in-house tech, but that's not really comparable (massive parallelization of that one, and most things are done asynchronously - so you don't even know that scripts are being compiled - no "Hold on" or such things … but vastly inferior in feature set to Unity, it's a specific tech).

frob said:
Major game engines provide all the heavy lifting without individual developers needing to implement it. Click a checkbox and you get realtime, photorealistic dynamic lighting, shadowing, and effectively limitless view-dependent on screen detail.

Nice try. But i don't buy it. They still look gamey, uncanny and hacky. The promise is there, the checkbox is there, but the outcome is what it always was: Smoke and mirrors.

Which is not a problem. Our selling point is not that the current game is bigger and more realistic. It is that the next game after that will be even bigger, looking even better, etc.
If you just claim ‘we have achieved photorealism right now’ you only get two things: Either doubt on your marketing claim, or they buy it, but then conclude the journey is over and we're stuck at this point.

What i mean is: We should not brag on our technological achievements to be the one and only solution for long standing problems. We better just point out we have progress along the way, and ideally we never get there so the road never ends.

Instead the systems have changed to “draw this complex scene document that's longer than a bestseller novel." The consequence is that you build a system capable of rendering the complex scene, but instead of throwing gigabytes of data at it, you give it the document with your single triangle.

What's that? a promise to turn a single triangle into rich, never ending scenes? How? By duplicating the same triangle a million times like Epic proposes? Or by halucinating an entire world out of that single triangle using some neural net trained on Roger Dean paintings, like NV suggests?

Well, we will see. But we are not there yet. I'll believe your claim if i can enter every single house and skyscraper apartment in the next GTA game, but each is unique and tells a story about its residents, each having a simulated life, daily routine, and blah blah blah… You know, the usual promises we make without ever getting there.

Which is fine. But what's not so fine is this: The promises (or self-confirmation) you just made talk about one thing: State of the art.
But contrary to what you say there are no multiple engines doing that magic anymore. The industry converges towards just one single engine for PC/console games, and just one other single engine for mobile games.
As a result, everybody has the same state of the art. The one achieved by Epic.
Now i ask you: If every game has the same level of jaw dropping visuals, sound, animation, etc., then how should i be impressed from one game over the others? They are all the same, so i'm not impressed from any one.
And worse: How can you claim to be a creative industry, if you all outsource the actual development - which is a central part of the video games medium - to the same centralized spot?
Where is your contribution, your creativity, your innovation, your part on this?

So that's what i think why we have this thread here. People don't want everything coming from just one or two giant tech companies, making marketing promises and business. They want individuality and variety.
A creative industry can't centralize, even if there are economical arguments to do so. That's a self contradiction.

Just look at the music business again: First we've had Pink Floyd, where band members could play instruments and wrote all songs themselves.
Then we had Backstreet Boys, where band mates could not play instrument or compose, but only dance and style their hair. Their songs where still good, but outsourced and written by experts. It's just not as great as Pink Floyd was, due to a lack of honest creativity in favor of optimized business.
And now they have absolutely nothing to show anymore. Their industry is dead.

That's exactly where we are heading too. The ship is sinking.
Contrary to the music industry, we have a never ending source of inspiration: Potential technological progress. Everybody can take from this pool for free. But nobody does anymore. They all outsource it to Epic, giving the advantage away for no return.

Vilem Otte said:
Technically speaking Unity is quite modifiable as engine, with SRP and other additions over the past years (but not just those - it was quite “bendable” even before) - there are SDF generators + tracers well integrated and implemented by 3rd parties. Same goes for various shadows algorithms.

So, basically you can extend Unity using 3P libs (which may see further development or not) or your own developments (which is more work), to eventually reach visual parity with UE, which has all those features out of the box and comes with source code by default?

Well, yes - that's nice. But i still don't see why you laugh at those who prefer the latter option. You have failed to explain. But never mind ; )

JoeJ said:
So, basically you can extend Unity using 3P libs (which may see further development or not) or your own developments (which is more work), to eventually reach visual parity with UE, which has all those features out of the box and comes with source code by default?

Well, yes - that's nice. But i still don't see why you laugh at those who prefer the latter option. You have failed to explain. But never mind ; )

The main reasoning here is - they have worked in Unity previously, released titles in it, etc. Switching from one engine to another (and rebuilding not just asset pipelines, but also throwing away anything you could use in future) in a next title (which I believe was laid out to use Unity first - but that's purely based on timeline and version of UE available at the time I suppose they started on it, so I can't call the shot here) … just because of few “visual effects” is likely going to be much more pricey than 3rd party solution specifically for those few effects.

Other option would be to completely change your team - but when whole team had experience only in Unity, switching tech is not the wisest thing to do. Not over the visual effects.

We might differ in opinions here. But I call that very bad business decision.

Vilem Otte said:
The main reasoning here is - they have worked in Unity previously, released titles in it, etc. Switching from one engine to another

Ah ok, so what you mean is like ‘They made a game in Unity; almost done; then they switched everything to UE just to jump on the photorealism hype train.’
Then i can understand the laugh.