I'm not going to broadly disagree that more work could be done to make some of these ports better. In fact that should be obvious from the differences in part 1 and part 2, yes? They clearly did more work porting part 2 and deserve praise for that. But I think we PC gamers (broadly generalizing... many of us play things on both these days) also need to avoid the temptation to fall into the same sort of "moving target" comparisons that we criticize console gamers for. It's very easy to throw out notions like "oh this DirectX feature or IHV-specific API would make things so much better!" without really having good data for a given game to back that up.
DirectStorage is actually a perfect example here... based purely on the advertising from Microsoft and others people got really excited and somehow got the idea that this was going to revolutionize PC gaming and so on, but I think it's pretty fair to say that for the few people that have used it (such as here), it doesn't exactly show off anything particularly revolutionary that can't be done without it.
Another example is sampler feedback and related streaming stuff; this is the stuff that gets marketed to consumers as to why they should get a new GPU or operating system or so on but in reality we are pretty deep into the point of diminishing returns on many of these API/hardware features (raytracing aside, since so much of that is still hardware black box on PC).
Mesh shaders specifically is really a tougher one to argue because it is a fundamental change to the content pipeline. You can't just "implement mesh shaders" and get a benefit. The important part of such a shift is the content pipeline changes that restructure meshes into appropriate clusters with associated LOD logic and so on. The actual mesh shader "hardware" feature is way less relevant than those changes, even if you implement them without mesh shaders at all (such as previous GPU culling systems did, nanite on systems without mesh shaders, etc). So while it's easy to say "just add mesh shaders and your geometry culling will get so much better", we don't really know here that geometry is a huge bottleneck in the first place, and the ask is potentially a rather huge one depending on their current geometry pipeline.
I'd certainly be curious for someone to do a deeper dive into some of these frames to see where the time is going, but I don't know that I would broadly say this is a "minimal effort port" for something that came from a single-platform engine. Even games that target both Xbox and PS5 are much easier to port because they fundamentally have to deal with more than one platform in the first place. Single platform stuff can just make a lot of simplifying assumptions to the point that "porting" them can sometimes stray into territory more like writing a PS5 emulator than just optimizing some code.
None of this is to say that bringing criticisms is not useful; I think a lot of the improvements in part 2 were probably motivated by the criticisms of part 1 and presumably if PC ports continue to be a thing for the series future engine iterations will keep this in mind even if they come out first on the consoles. But I think it's worth giving some leeway to folks who are porting a single platform game that was never designed with ever running on anything other than a playstation in mind, long past the time where any amount of art or content pipeline iteration is realistic.
Happy to see you respond Andrew. I agree with your appreciation of how optimising for PC and getting the most out of the platform is challenging. I also want to talk about the aspect in your quote of me that you did not mention.
"But you have *other* transformational performance enhancements that can be added but probably are not added because of Budget and time table reasons. "
What my original quote is saying is that there are always things that can be done to make a game perform better, but budget and time tables are getting in the way. Regarding mesh shaders which was a thrown out as an example by me, as you write yourself, the content pipleline change is the aspect that makes it a challenge in terms of time and money. Feasibility as being technically possible is most likely already there. Throwing out more triangles to in a way which maps closer to the underlying hardware is always gonna be nice - but for x amount of money and x amount of time before a release date to get x amount of return in sales? Maybe not. Let us just ignore for a moment the specific example of GPU performance, which I think is symptomatic of what I am talking about, but in general I think Nixxes has not been getting enough time and resources to do its ports since just after its first release of Spider-Man. The quality of releases has been increasingly strained from a reviewers perspective especially in light of how polished Sony games tend to be. Why do I say that?
Nixxes had to turn this port around of TLOUP2 in 6 months. In total, Nixxes have worked on 8-9 different game releases on 5-6 different engines in the last 3 years (Spider-Man PC, Horizon Zero Dawn patches of Virtuous Port, Miles Morales, Ratchet and Clank, Horizon Forbidden West, Ghost of Tsushima, Horizon Zero Dawn Remastered PC, Horizon Zero Dawn Remastered PS5, Spider-Man 2, The Last of us Part 2). I think the core thing limiting Sony game quality and stability is not technical feasibility of something (can we make X work at all?), rather, can we make x work at all with the amount of budget and time Sony gives us? I think when you look at the sheer volume and diversity of titles in that time scale, you can start imagining why cracks appear in Sony products at launch. It started to get really obvious to me with Ratchet and Clank that Sony is obviously not giving them enough time to release as surprising things were being left in products for launch. The pattern has become: game comes out with issues, initiate of flurry of patches near launch, look to have a greater amount of issues that are fixed or remedied within the 3 month window after launch usually. This is how Sony does it on PC as of late, on console, it is not like that at all. Sony games on console are typically seeded 2 weeks before launch in a gold position having undergone an extensive polishing phase. Based on what I can see in comparison to other game releases I cover, Nixxes is not being given the comensurate polish time before release for the task at hand.
As I said when Ratchet and Clank launched, I think Sony needs to give Nixxes more time before release in general. Both for a polishing and for fundamental aspects of the ports that come out. With the last batch of Sony releases on PC I have had to do a lot of gremlin chasing at launch which filter into my reviews "report on x bug the user might see" instead of spending time talking about optimised settings. With other games that publish in more polished states, I do not have to go bug hunting to try and figure something out for large parts of my review process, so I can spend more time on the normal aspects of a review. A great example is Microsoft's Senuas Saga, Kingdom Come Deliverence 2, Ubisoft Massive's Avatar, etc etc...
Sony's strategic decisions are having a negative effect on their ports on PC. Regarding The Last of Us Part 2, this strategic "moneyXtime" aspect is directly referenced by Naughty Dog and Nixxes in the interview I had with them. Some examples,
1. Where ND they say the originally wanted to Nixxes to port The Last of Us Part 1, but due to timetables, they got Iron Galaxy to do it.
2. Same with Part 2, ND wanted Nixxes on it, but instead they used Iron Galaxy and only transitioned to Nixxes after about 11 moths of of work.
3. When Nixxes mentions how they had to work with Iron Galaxies base port, instead of rolling their own.
With number 3 there, Nixxes has to deal with the sunk cost decisions referenced in Number 1 and 2. All of those things are considerations of "Time and Money" imposed by Sony, and not technical things. The end situation is where Nixxes had to doctor up fixes to issues from a problematic foundation from Iron Galaxy - kind of like they did with the original patches to Horizon Zero Dawn as ported by Virtuous. With TLOUP2 we are not looking at Nixxes' work in isolation, rather them applying fixes and adjustments to someone elses work. This is the opposite of what they did with Horizon Forbidden West, where they started the engine port themselves ignoring Virtuous' port. I really think we would be looking at a different product if Nixxes had done all of the TLOUP1 and 2 porting from the outset instead of what we got in the end.
That above is me just spittballing about money and product planning strategy and how I feel it is affecting its ports. In terms of concrete things in TLOUP2, you can see some of the effects of all that in the game.
1. When reviewing The Last of Us Part 2 pre-launch, we saw a number of visual issues in the game that had been in every single Naughty Dog release on PC, we had to tell Nixxes about them in a listed form. (some of these have since been patched)
2. TLOUP2 has nearly the same graphical options as TLOUP1.
3. TLOUP2 had the same forced sharpening that was present in Iron Galaxy ports (since patched)
4. GPU performance visavis console is largely the same
A long response to you here Andrew. lol
In general my disappointment referenced in the video title is much not at all with what Nixxes is capable of. Given enough time, they always do great things. Rather my disappointment with TLOUP2 is an effect of Sony's decisions on the quality of their PC releases and TLOUP2 has many hallmarks of that at release.
