Digital Foundry Article Technical Discussion [2024]

IMO and i may be dead wrong on this its gonna be much harder for non UE5 games to compete with UE5 games when it comes to graphics. Just kooing at Senua and Lords of the fallen (wich i played recently) my god this is completely new quality in games. Level in details is just absurd nothing we are used to. Even SH2 (havent played it yet myself) developed by blooberg, their previous game (forgot title, horror where two worlds are rendered at the same time) jesus what a leap in quality. And this is coming from small studio that never produced anything that looked so good. When more big AAA games using Ue5 will be realsed with massive budgets i think it will completly overshadow any other games build on custom engines. Would be great to know what kind of budget games like Indy, SH2, Senua and Wukong had.
 
Again, you're missing the point. The point is that there are levels to how accurate baked lighting is versus ray-traced lighting. Sometimes, the differences are subtle such as with this case.

oCEosi4.jpeg

tyCqVKG.jpeg


Whereas at other times, the difference looks literally almost like night and day such as with this case:

e8IEFUl.jpeg

wB5AN5q.jpeg


No, sometimes, the differences are minor unless you look for them. Low-quality geometry cannot be hidden OR FAKED. Anyone can spot its ugly head whenever it appears. It's possible to have good quality lighting even without ray tracing. Some people are good enough to approximate it and give excellent results.

That's assuming there is light bleed occurring in the scene, which isn't always the case.
Pictures does not tell the full story, just like with TAA, motion gives a lot away and I can cleary see the missing shadows in yor examples 🤷‍♂️
 
I played at max settings on a 4090+13900K. Alex also touched upon the LOD issue in the video. It occurs even at max settings and you have to manually tweak the config files (or was it a command?) to increase it.
I watched that video. It's nowhere near as bad with full RT on. I've also only noticed on foliage which restricts that to the first level in the game and the trees on the Vatican level. The local lighting is still using shadow cascades and so that's not using RT shadows. I'm on the last level now in the snow and haven't noticed it yet.
 
That's assuming there is light bleed occurring in the scene, which isn't always the case.
The problems with GI light probes is that there is no variation in illumination on the same asset. One entire mountain face, for example, would get the same illumination instead of it varying based on the cracks and crevices. You see this a lot in Horizon game. The entire asset loses all direct lighting and can only be visible by evaluating a light probe that's literally a constant color. It also doesn't take into account the inverse square distance law or normalization between adding up all the lighting contributions (i.e. making them all add up to 1). This is especially the case for trying to evaluate a subsurface skin shader, for example, which models light bouncing into layers of skin and leaving the surface at a different location. That totally breaks down using baked lighting because in the real world there is no constant ambient term. Another thing is the screen-space solution for ambient occlusion to try getting some sort of directional occlusion going but is too crude and over darkens where the light is supposed to bounce around.

I can't think of a game that uses baked lighting where I say it's good enough compared to path-traced lighting.
 
The problems with GI light probes is that there is no variation in illumination on the same asset. One entire mountain face, for example, would get the same illumination instead of it varying based on the cracks and crevices. You see this a lot in Horizon game. The entire asset loses all direct lighting and can only be visible by evaluating a light probe that's literally a constant color. It also doesn't take into account the inverse square distance law or normalization between adding up all the lighting contributions (i.e. making them all add up to 1). This is especially the case for trying to evaluate a subsurface skin shader, for example, which models light bouncing into layers of skin and leaving the surface at a different location. That totally breaks down using baked lighting because in the real world there is no constant ambient term. Another thing is the screen-space solution for ambient occlusion to try getting some sort of directional occlusion going but is too crude and over darkens where the light is supposed to bounce around.

I can't think of a game that uses baked lighting where I say it's good enough compared to path-traced lighting.
Forbidden West is a good looking game, but let’s be honest, it has some of the worst lighting for an open-world AAA game. That thing at times looks horrible. I know you’re using it as an example to drive a point, but Cyberpunk without RT still has far superior lighting.

We don’t have many path traced games, but I think Wukong in general looks good enough even without RT. It’s much smaller in scale outside of Chapter 6 though. Open-world games should embrace path tracing. Far too many instances of botched lighting in them and I can’t blame them. Properly lighting all those environments would probably take months of additional work.

Senua’s Saga lights and shadows seem painstakingly doctored, so I’m not sure how much PT would enhance it. Might even kill the mood a lot of the time. Real life isn’t that dark lol.

Pictures does not tell the full story, just like with TAA, motion gives a lot away and I can cleary see the missing shadows in yor examples 🤷‍♂️
In motion, it’s even less obvious because you have to stop and look at the scene. Developers have gotten excellent at baking lighting and can sometimes get close to ray traced lighting.

Furthermore, accurate doesn’t necessarily equate to aesthetically pleasing. Some scenes have deliberately inaccurate lighting to serve the mood. There’s no such thing as inaccurate geometry or textures. We just call them bad.

That’s why to me, what Senua does with the textures, characters, and geometry is a lot more impressive than what Indy does with the lighting. Hell, given how moody and dark Senua’s Saga is, fully accurate lighting might have hurt the art direction and tone of the game.

Seeing stuff like this is a lot more egregious than light leakage or missing bounce lights.

bms7niG.jpeg

jngVq3s.jpeg
 
Last edited:
DigitalFoundry's video is about games released in 2024. Which other non UE5 game pushes as much geometry as Indiana Jones and was released this year?

This is an example of pushing the limits with tradional rasterizing:

When someone states it has 'state of the art geometry' its stops being about 2024 and incudes all games from all years.
 
Same with RT.
Look at the generational uplift.

People just assmue today that their entry level GPU can do every at max settings.
People have bad memories, this is "The Witcher 3" at launch:



Only thing that have changed since 2015 is that the whining minority has becaome far to vocal for their numbers and they forgot about the past.

Was no better in 2004 either:

I blame DLSS and FSR to be honest for people's perceptions of what their GPU actually is.
 
When someone states it has 'state of the art geometry' its stops being about 2024 and incudes all games from all years.
And which other games from not 2024 have a better geometry fidelity?
This is Sukhothai:


Nothing on this planet looks better than Cyberpunk with Pathtracing. So does this makes other games or engines not "state of the art" in certain aspects?
 
We don’t have many path traced games, but I think Wukong in general looks good enough even without RT.
I'll load it up and turn off RT and see for myself.

UPDATE: I can see why you would think it's good enough as it's a much closer approximation than last-gen's GI light probes (i.e. HFW, etc.). However, it still comes up short in accuracy compared to it's PT competition. For example, when you place the main character in shadow, his hair, skin and clothing materials break down. He's also not getting lit by the same environment sun. This technique will not be good enough come next-gen if the developer landscape starts putting out nothing but PT games by 2028.
 
Last edited:
Nothing on this planet looks better than Cyberpunk with Pathtracing. So does this makes other games or engines not "state of the art" in certain aspects?
Nearly all recently released games look better than Cyberpunk with path tracing. Just look at the screenshot above and count the polygon edges of the cups in that trashbin. The game has last gen assets because it's a last gen game made with Xbox One and PS4 in mind.
 
Same with RT.
Look at the generational uplift.

People just assmue today that their entry level GPU can do every at max settings.
People have bad memories, this is "The Witcher 3" at launch:

View attachment 12751

Only thing that have changed since 2015 is that the whining minority has becaome far to vocal for their numbers and they forgot about the past.

Was no better in 2004 either:

View attachment 12754

You exaggerated a little. The Witcher 3 in 4K in 2015? Doom III at 1600x1200 with 4x FSAA?

TW3 is difficult to benchmark at launch, considering that the game had a radical change in performance, according to new patches. I even remember Digital Foundry making videos for each new patch, because its radically changed performance.

Not to mention that the game had bugs with Hairworks, even on Nvidia hardware.

56MACcqGSnMWqt8XBYtZ6U-1200-80.gif
MzI4Ny5wbmc
 

Very interesting new antilag tech in the PS5 SDK. Developed in conjunction with Infinity Ward no less. Or is it a feature of the engine?

Edit: "Developed in conjunction with Sony, an extra API call is made on PS5 Pro this way, to predict the upcoming frame-time."

Still not sure what is what.
 

Very interesting new antilag tech in the PS5 SDK. Developed in conjunction with Infinity Ward no less. Or is it a feature of the engine?

Edit: "Developed in conjunction with Sony, an extra API call is made on PS5 Pro this way, to predict the upcoming frame-time."

Still not sure what is what.
I haven't had time to watch the video, but if this is developed in conjunction with Sony, and assumingly a system level feature devs can tap into... I wonder why Sony didn't bother to market this ability at all? Do they test it in the video and does it show any meaningful difference?
 
I haven't had time to watch the video, but if this is developed in conjunction with Sony, and assumingly a system level feature devs can tap into... I wonder why Sony didn't bother to market this ability at all? Do they test it in the video and does it show any meaningful difference?
Infinity Ward claims that it has lower input lag than running the game unlocked. But rewatching that part of the video, it sounds more like a feature of the engine than a SDK thingy.
 
Indiana Jones renders different scenes with state of the art geometry fidelity. That doesnt mean it is on par with UE5.
very impressed with this game. I got past
China
-crazy stuff there :) - and before that, when
in the Himalayas, when Gina throws the grenade
I didn't know whether I was playing a movie or a game. o_O

Currently playing Indy at 1440p no scaling, all maxed out except textures. The lighting of the game gives me that Metro Exodues vibe, but maybe even better. That's my favourite thing about decent RT, that whether it compares favourably vs non RT, the lighting feels logic.
 
Last edited:
Back
Top