Digital Foundry Article Technical Discussion [2023]

[Dictator]

The second is that I think some are upset that I pointed out Digital Foundries technical hypocrisy.

"Technical hyprocrisy". Yeah could you try and prove that? Because that makes no sense and you are projecting your view onto things we are not even saying.
How exactly are we supposed to logically fully critique a game's performance before we even have our hands on it (especially on PC) to say whether it is unjustified or completely justified? You seriously need to think that one through. There is no "bias" here - we are just speculating and articulating as to why BETHESDA maybe locks down the performance: we are not expressing our preference or pleasure for that performance.

Was it also "bias" when I made this video about all the visual compromises in Spider-Man's reflections?

Was it "bias" with me trying to excuse and defend the developer's technical competence? Or was it me trying elucidate logical reasons why they are compromised as such?

 

[cheapchips]

Correct me if wrong, but aren't real life planets connected by space?

Humans can't travel between planets in real life.

If it's realism you're after, during entry decent and landing you'd spend hours while the computer deorbits, a few minute where there's a glow outside the window and then a few minute of sky out of the window while the computer lands.

Not being able to control a spaceship through EDL is such an odd hangup.

 

[Subtlesnake]

Wow, do you think this is the first space game with procedural generation? You guys keep mentioning things about Starfield while conveniently ignoring no man sky who has done basically everything Starfield has on ps4. It has procedurally generated planets with way more than 1000 planets and the environments, weather, etc are generated real time. All the concerns you raised were already done by NMS.

Does NMS feature an equivalent real time GI system and have cities with "tons of NPCs" with simulated lives?

 

[Cyan]

my most expected video in a while. Gotta watch it. VRAM isn't an issue in the case of a A770 but I am curious to know what is the only possible issue of the game.

Glad to have the accessibility option, increasing the cursor size has been a must for me.

 

[Below2D]

The second is that I think some are upset that I pointed out Digital Foundries technical hypocrisy. Unfortunately, this isn’t the first time I’ve seen them pick and choose when to criticize a game for one thing while giving another game a pass for the same thing. It’s just wrong. If you want to criticize, be consistent so that we know where you stand. If you can’t be consistent, just discuss the technical aspects without letting your opinion spill into it. This way, viewers can arrive at their own conclusions without being told what to think.

I'm sorry but that's complete bullshit. 99% of the time, DF actually tries to understand the reasons behind compromises and is seldom overly critical of anything. Hell, people even blame them for not being severe enough with their critique. There was an outcry for Spider-Man 2 looking basically identical to its predecessor and a step below the reveal trailer but DF has nothing but good things to say about it. Same for Halo Infinite that everyone dumped on but again, DF tried to navigate the treacherous waters and find reasons for the visuals rather than roast the game.

I'm not sure what "technical hypocrisy" you're referring to but DF tends to be quite nice (more than I would be) with most games except for when it's egregious and downright unacceptable such as with Cyberpunk on consoles, TLOU Part I on PC, and so on. You're turning this into a personal attack rather than sticking to a technical discussion.

 

[iroboto]

Wow, do you think this is the first space game with procedural generation? You guys keep mentioning things about Starfield while conveniently ignoring no man sky who has done basically everything Starfield has on ps4. It has procedurally generated planets with way more than 1000 planets and the environments, weather, etc are generated real time. All the concerns you raised were already done by NMS.

Procedurally generated facial animation? Horizon 2 does that and does it so much better, it’s not even funny and that runs on ps4. Nothing you’ve mentioned is remotely impressive because the scale at which they’re doing it is smaller than that of a game running on ps4. I’m sorry but, I’m not impressed by things which have been done before and done at a bigger scale on worse hardware. Maybe if they delivered this in a technically competent package and not 1296p at unstable 30fps, I could actually be impressed. Again I’m sorry but nothing this game is doing justifies this level of performance on a machine that is 8x more powerful than its predecessor.

To make matters worse, you tried to shift goal posts to file size? Last I checked, this isn’t a thread discussing file sizes.

Honestly, I wish we had the ability to profile the game so that we can see how they just wasting resources. If there’s one thing that’s a guarantee with Bethesda, it’s that they always inefficiently utilize resources.

I can’t even. I don’t want to waste my time on this, you’re not getting piled on because you’re attacking DF. You’re getting piled on because your arguments are heavily flawed and ignore some major items. And we keep bringing up scale with fidelity and most games have one or the other. This has both.

It’s like saying any web developer can make Twitter. It’s just posting a tweet of 250 characters. But scaling twitter Requires a completely different level of power.

I cant even understand why you are attempting to compare NMS that does not have anywhere close to the level of scale thst Starfield has, and Star Citizen which is a tech demo no where close to complete And full of bugs and using these as your main axioms as to why Starfield is embarrassingly bad and has been done before.

It’s pretty clear it has not. No game today has this level of scale and fidelity. You will find better looking games. You will find games with more scale. You will not find one with more scale and fidelity combined.

 

[Cyan]

video watched. An excellent one. Two things:

- I missed DLSS vs FSR2 vs XeSS comparison
- could it be that the 32GB recommendation means full system RAM -conventional RAM + VRAM-?

 

[Below2D]

video watched. An excellent one. Two things:

- I missed DLSS vs FSR2 vs XeSS comparison

There was none as far as I'm aware. I find the video lacking.

- could it be that the 32GB recommendation means full system RAM -conventional RAM + VRAM-?

Nobody in the history of gaming has ever done that so I hope it doesn't start now lol. A 4GB card with 28GB of memory isn't the same as a 16GB card with 16GB of system memory.

He also glossed over the performance of the 1060 and options to mitigate VRAM usage besides, drop textures to High. Would a 8GB 2080 do fine at say, 1080p+ Ultra Textures?

 

[Flappy Pannus]

Very comprehensive, great video by Don.

It's almost like the texture streaming system is completely broken, I mean why would they recommend 32GB system ram for Ultra but the game never even fully utilizes 16 regardless? Why aren't they specifically recommending a vram amount for texture settings at all? It's bizarre. Extremely well optimized and scalable in every other aspect, so that's frustrating.

Goddamn that Blizzard launcher though.

Also, the game's textures just don't look good or varied enough to justify 16GB of VRAM.

It's an isometric game with limited camera control, so you're not even getting up that close to the textures in the first place to boot. Yeah I can't see the quality warranting that insane vram usage at all.

 

[Below2D]

Also, the game's textures just don't look good or varied enough to justify 16GB of VRAM.

 

[BitByte]

"Technical hyprocrisy". Yeah could you try and prove that? Because that makes no sense and you are projecting your view onto things we are not even saying.
How exactly are we supposed to logically fully critique a game's performance before we even have our hands on it (especially on PC) to say whether it is unjustified or completely justified? You seriously need to think that one through. There is no "bias" here - we are just speculating and articulating as to why BETHESDA maybe locks down the performance: we are not expressing our preference or pleasure for that performance.

Was it also "bias" when I made this video about all the visual compromises in Spider-Man's reflections?

Was it "bias" with me trying to excuse and defend the developer's technical competence? Or was it me trying elucidate logical reasons why they are compromised as such?

I’m not projecting at all. I’m just calling it how I see it. You can’t go on about how FSR is poor and it is, then in the same vain describe 1296p fsr as rather crisp. That is hypocritical. My claim of technical hypocrisy lies with DF's lack of consistency with criticism. DF complains about one game doing something yet doesn't complain about another game doing the same thing. Again, this is the problem with injecting personal opinions into technical discussions. If you just said, it appears Starfield is using fsr from 1296p to 4k, that's a data driven statement. There's nothing subjective about it. Viewers can watch the video and decide how it looks for themselves. When you go ahead and start to describe it subjectively as rather crisp, you run into the risk of being inconsistent with your previous statements. For example when you complained of FSR fizzle/pop, ghosting, other artifacts, etc none of which are crisp. I only chose that example to highlight patterns of inconsistent criticism. Most of the time, it arises from the injection of personal opinion which is rarely consistent and the delving into speculation.

I'm sorry but that's complete bullshit. 99% of the time, DF actually tries to understand the reasons behind compromises and is seldom overly critical of anything. Hell, people even blame them for not being severe enough with their critique. There was an outcry for Spider-Man 2 looking basically identical to its predecessor and a step below the reveal trailer but DF has nothing but good things to say about it. Same for Halo Infinite that everyone dumped on but again, DF tried to navigate the treacherous waters and find reasons for the visuals rather than roast the game.

I'm not sure what "technical hypocrisy" you're referring to but DF tends to be quite nice (more than I would be) with most games except for when it's egregious and downright unacceptable such as with Cyberpunk on consoles, TLOU Part I on PC, and so on. You're turning this into a personal attack rather than sticking to a technical discussion.

Sorry but, I disagree. What I'm discussing is DF's lack of consistency when it comes to criticism not the degree of their criticism. I don't care if they're critical as long as they display the consistency in all subject matters and they don't. I didn't call out anyone from DF so I don't know how it can be construed as a "personal attack". If you criticize something, be consistent and they aren't consistent.

I can’t even. I don’t want to waste my time on this, you’re not getting piled on because you’re attacking DF. You’re getting piled on because your arguments are heavily flawed and ignore some major items. And we keep bringing up scale with fidelity and most games have one or the other. This has both.

It’s like saying any web developer can make Twitter. It’s just posting a tweet of 250 characters. But scaling twitter Requires a completely different level of power.

I cant even understand why you are attempting to compare NMS that does not have anywhere close to the level of scale thst Starfield has, and Star Citizen which is a tech demo no where close to complete And full of bugs and using these as your main axioms as to why Starfield is embarrassingly bad and has been done before.

It’s pretty clear it has not. No game today has this level of scale and fidelity. You will find better looking games. You will find games with more scale. You will not find one with more scale and fidelity combined.

So essentially what you're saying is that the result of combining last gen graphics + last gen animation + last gen scale justifies the utilization of resources on a machine that 8x more powerful than last gen? If that's the case, i don't believe you for a second.

 

[Subtlesnake]

So essentially what you're saying is that the result of combining last gen graphics with last gen scale justifies the utilization of resources on a machine that 8x more powerful than last gen? If that's the case, i don't believe you for a second.

If you think Starfield has PS4-era graphics then it's probably not worth discussing further.

 

[BitByte]

If you think Starfield has PS4-era graphics then it's probably not worth discussing further.

I'll leave it to you to explain how this is "next-gen" graphics

 

[BitByte]

Call of Duty Infinite Warfare

Starfield:

8x difference in graphics..... As an aside, I'd love to hear where people think the light is coming from in the starfield shot....

 

[iroboto]

Call of Duty Infinite Warfare


Starfield:


8x difference in graphics..... As an aside, I'd love to hear where people think the light is coming from in the starfield shot....

baked
Last Gen got away with their graphics stack by baking everything.

So essentially what you're saying is that the result of combining last gen graphics + last gen animation + last gen scale justifies the utilization of resources on a machine that 8x more powerful than last gen? If that's the case, i don't believe you for a second.

If you can’t bake, all the power is out towards computation.

If that doesn’t make sense to you, I don’t know what will. 8x compute power does
Not result in 8x better graphics. That’s probably the most non sensical thing written.

The resolution is already 4x the workload of 1080p.

you can’t seriously see the difference between Star Wars Outlaws and this are entirely different in how they approach rendering ? And why it has to be?

 

[BitByte]

baked
Last Gen got away with their graphics stack by baking everything.

If you can’t bake, all the power is out towards computation.

Starfield claimed real-time GI yet the lighting is not characteristic of realtime GI. I already posted a screenshot showing lighting errors that are not consistent with real-time global illumination. I can't wait for them to go into their deep dive on how they're doing it.

If that doesn’t make sense to you, I don’t know what will. 8x compute power does
Not result in 8x better graphics. That’s probably the most non sensical thing written.

The sarcasm seems to be lost on you. As an aside, if the outcome is "similar", what benefit is there to the end user to do "real-time" gi on a machine that can barely do it? I mean other than wasting computational resources?

The resolution is already 4x the workload of 1080p.

False. 1296p + the cost of FSR is not 4x the cost of 1080p.

you can’t seriously see the difference between Star Wars Outlaws and this are entirely different in how they approach rendering ? And why it has to be?

Starwars Outlaws? I never even mentioned that game. I don't know what you're talking about so if you could provide additional clarification, that would be appreciated. With regards to the different approach to rendering, if it fails to provide an outcome that is noticeable better than the previous approach, it's most certainly not worth the cost.

 

[Seanspeed]

DF......is seldom overly critical of anything.

I agree with basically everything else you said, but just wanted to take an opportunity here and say I do think DF is probably super sensitive to certain things that the large majority of others probably aren't in reality.

A recent example for me is the Dead Space remake. It exhibits some hitches, but I think the way it does so, which is almost always out of combat and usually only like a single hitch entering a new room and all that is really *super* forgiveable for anybody except the most sensitive of players. Yes, it means you dont get a perfectly smooth experience, and I agree that devs should try and fix/alleviate such an issue where possible, but I also think so few people in real life will actually care.

And I think there's more than a few people who actually watch the videos and take DF's perception of the problem as gospel and then adopt it as a real problem for them as well, even though if it wasn't pointed out to them in the first place they'd probably never actually care much at all, even if they did notice it here and there.

Dont get me wrong, I love DF, watch their stuff nearly religiously and think they do a lot of good for not just the community but also the industry itself in raising standards, but I think it's worth pointing out that they are like the '1%' folks in terms of their discerning eyes and sensitivities in the gaming community and everything they point out/complain about will not necessarily be a big deal for the vast majority of gamers, including a good portion of enthusiasts.

I do get what you meant by your comment, though. They dont resort to hyperbole. They try and be reasonable about things, which is great and I would not watch so regularly if they weren't. But still - they are the 1% and their views will not always represent the experience of most others.

 

[iroboto]

Starfield claimed real-time GI yet the lighting is not characteristic of realtime GI. I already posted a screenshot showing lighting errors that are not consistent with real-time global illumination. I can't wait for them to go into their deep dive on how they're doing it.

The sarcasm seems to be lost on you. As an aside, if the outcome is "similar", what benefit is there to the end user to do "real-time" gi on a machine that can barely do it? I mean other than wasting computational resources?

False. 1296p + the cost of FSR is not 4x the cost of 1080p.

Starwars Outlaws? I never even mentioned that game. I don't know what you're talking about so if you could provide additional clarification, that would be appreciated. With regards to the different approach to rendering, if it fails to provide an outcome that is noticeable better than the previous approach, it's most certainly not worth the cost.

you’re evidence is a single point without the ability to actually test what is happening, but OK. That’s far from thorough testing as to why the shadows are like that.

You pretty much wrote it in your last line. You have no intention on talking about the technical work of what’s being done, you’re only interested in the output. Sad really, because that’s the difference of why we can run Starfield on next Gen and not last Gen consoles.

You’re right. The upscaling does put them at equal ground. But outside of that one point you don’t have strong ground here.

What’s the point of doing real time computation if the machine can’t do it or if it doesn’t look any better?

Bake 1000 worlds of lighting indoors
And outdoors for very possible variation and sun location and shadows, how is that going to work? How much space does that require? How much labour for each change?

 

[dobwal]

I’m not projecting at all. I’m just calling it how I see it. You can’t go on about how FSR is poor and it is, then in the same vain describe 1296p fsr as rather crisp. That is hypocritical.

Tell me how what Alex says about FSR 2.0 in this video, is inconsistent with what he says about FSR in Starfield? He even uses the term "crisp" twice in reference to FSR and this video is over a year old. Most of the poor showing and flaws of FSR are displayed by the performance mode which has an input of 1080p. 1296p as input falls within the balance setting for FSR 2 not performance. Furthermore FSR is an ever evolving technology so why do you expect the critique to be consistent. An honest critique of an evolving technology is strictly dependent on the performance of the technology at the time of the critique. FSR 2.0 has seen 5 versions since last May and we are currently on 2.2.1.

 

[Flappy Pannus]

I agree with basically everything else you said, but just wanted to take an opportunity here and say I do think DF is probably super sensitive to certain things that the large majority of others probably aren't in reality.

A recent example for me is the Dead Space remake. It exhibits some hitches, but I think the way it does so, which is almost always out of combat and usually only like a single hitch entering a new room and all that is really *super* forgiveable for anybody except the most sensitive of players.

I think you're probably correct that the majority of the public may not be as sensitive as Alex is to hitching, but their harshest critique of the stuttering is directed at the PC version, which absolutely does not have a "single hitch" entering a new room (which frankly even still would not be great - you're constantly passing between rooms!). Like, they have the data - the frametimes are right there, the multiple, bunched hitches are clearly visible. That, combined with the occasional shader stutters can lead to many instances of repeated hitches in excess of 100ms - that's very noticeable and can be felt. They're more numerous and egregious than in most games, save perhaps the latest Jedi. With the PS5 version their reporting on the hitching gave it the appropriate amount of coverage - "You can sometimes see a dropped frame here or there". If the PC version's stuttering was at that level it would still be reported sure, but not harped on to near the degree the PC version is.

If you have a 13900k or Zen3d, yeah - it's probably not as bad, but they test with more midrange CPU's for a reason. Lord knows I certainly felt them when playing the trial on my 12400f system. Stuff like missing compiled shaders in the first few minutes of the game are pretty bad too, not to mention stuff like the broken DLSS mipmaps are still not fixed.

You may not be as sensitive, and maybe the majority of the public isn't either, but I can make that determination myself with the provided data they give. The very nature of a technical review of a videogame by a channel that specializes in this is very niche, as is posting on a forum like this - and PC gamers in general too, at least ones buying $500+ GPU's. For that target market, I'm not sure they're overly sensitive.