Weez

@Flaming_Kaiser Except AI, and even Generative AI are broad descriptors. You're assessing the potential of a whole field of research based solely on your exposure to a small subset: prompt-based diffusion models. You're then claiming that experience is evidence as to why something fundamentally different, in it's infancy, shares the same fate. At the very least, that's endlessly pessimistic.

There are legitimate concerns - advancements in traditional rendering stagnating, quality being paywalled behind hardware capable of neural rendering, differences in the final image when each vendor has its own implementation. That's not to say it isn't worth investigating when traditional rendering is proving itself incapable of the next visual leap alone.

@Scribedby It's legitimately impressive how far Sony's come in just two generations, and I'm stoked to have quality upscaling on my PS, but there's still work to be done. PSSR 2 falls behind even older versions of DLSS in all arenas - stability, image quality and performance. Here's to hoping subsequent generations continue to close the gap.

@zebric21 Overlooking hardware differences between pro and base, AMD hardware is used throughout the console market. I can understand Sony choosing to dedicate resources to tech that'll only benefit their platforms moving forward.

@Dogbreath I wouldn't read too much into the 2 x 5090 claim. Over-allocating resources during tech demo's is the norm, and "development" neural networks are oversized as a result of over-parameterisation.

Before release, unnecessary connections are pruned, knowledge is distilled into a smaller network and weights quantized. It's not uncommon to see a 10x reduction in memory usage. What the final requirements will be - I can't say. Given that it was advertised as for use on a single card within the "5000 series", not on the 5090, I wouldn't be surprised if it comes with quality presets not unlike previous DLSS versions.

@ElkinFencer10 Yeah, I understand the concerns and criticisms - I love the medium and don't want to see it reduced to generative slop, nor do I want to see people lose their jobs. It's an industry I worked in for quite some time. But it's sad to see promising technologies rejected solely based on misunderstanding and classification errors.

An interesting, little discussed use that is starting to see adoption, is Neural Physics. Complex simulations need a lot of iterations to converge on the "right answer". By "guessing" the outcome in the initial iteration, one can solve the system in drastically fewer iterations, allowing for simulations that where otherwise impossible in real-time.

@somnambulance I don't think it will. DLSS saw similar negativity when it first released, and it's now a staple.

@Flaming_Kaiser There seems to be a lot of misunderstanding in regards to how upscalers work. Both existing upscalers and DLSS 5 effectively create details that don't exist based on generative AI - it's why DLSS 4.5 can produce stable, faithful output from as low as 240p input.

Unlike diffusion models like midjourney or stable diffusion, which effectively guide randomness towards coherent patterns using prompt weightings, all generation is directed by the games rendered output - current and historical frames.

Previous upscalers "guessed" missing high frequency details to increase resolution. DLSS 5 goes further, inferring surface dynamics (cloth sheen, subsurface scattering, eye specularity, etc.) and lighting phenomena. It doesn't make "generic" characters, because it doesn't modify their assets. Any perceived inconsistencies are either the result of lighting changes or unintentional artifacts - a product of this being an early developer preview of a first generation tech, achieved by hastily retrofitting it into existing titles. The AI filter look is a matter of poor integration and refinement, not a limitation of the technology.

What is affected in a scene, exposure and color grading preferences and the general "style" - that's all developer configurable. It doesn't undermine art direction, because in future releases, it will be part of the art direction process - much like existing post-processing stacks.

I understand the instinctive backlash - the promotional material was pretty jarring. With that said, much of it is misplaced, and Playstation will inevitably be adopting neural rendering moving forward.

• This is literally the first commercial implementation of holistic neural rendering, done by retrofitting existing games with it. DLSS5 has extensive developer controls and results will improve rapidly. This will become part of the pipeline, with assets and integration tweaked to produce the intended art direction, not undermine it.
• Artistic intent and output differ and modern high fidelity games are already impossible/impractical to make without extensive machine assistance. Photogrammetry for geometry and micro-expressions, procedural generation, projected micro-displacements rather than hand detailing, mocap, and yes, AI texturing). There are limits to what one can do by hand, especially given production deadlines, and so the industry adapts to meet gamers demands.
• Traditional rendering approaches have major fidelity constraints - pixel/fragment density and the limitations of microfacet models. For "photorealism" or Pixar-quality stylized results, you have to super-sample, render considerably above normal resolutions, with more expensive shaders and higher texture resolutions than is feasible. The industry has moved in the opposite direction, with sub-native internal resolutions given the massive computational cost of a lot of shading techniques (e.g real-time GI).
• DLSS5 is an implementation of neural rendering. It will progress, and others will be developed. There's nothing to say that integration won't be considerably less prominent - compensating for little more than the limitations of traditional rendering.
• Much of the uncanniness comes from the juxtaposition of high-fidelity character rendering alongside lower quality animation and dynamics. The look of the skin isn't a limitation of the technology, it's a product of training data and biases.

I'm interested to see where it goes in time. There are major hurdles to improving fidelity, and this is one way over them.

@Nnfyrbsnss Yeah, maybe I have to see it in motion, but to me, it's about as "gorgeous" as a Jackson Pollock. Lots of visual noise, and random world objects are given the high-contrast treatment. Looks visually confusing more than anything, which is a weird choice for the genre.

@RoomWithaMoose They all feature millenial culture. Later games, the reboot most egregiously, feature millenial writing (in the derogatory sense)

That's what happens when you take a millennial sized dump on a franchise.

9.5 - complaints only with some optimisation issues and the Katana engine showing its age. Excellent game.

@crossbit I felt much the same - used the sword a lot in 2 and losing stances felt off to begin with. Once I started using it for positioning and tool deployment, treating it more like a separate stance, it plays incredibly well. If you're looking for ideas, I found this quite useful.

@KundaliniRising333 The open area in the demo is apparently quite representative of performance. The full game is downloading as we speak, so I'll be able to comment more later.

As for whether it's worth it, I'm admittedly biased. I slept on Nioh, tried 2 and didn't see what the fuss was about until I saw high-level play and poked around trying to figure out how to replicate it. Combat is very different to other soulslikes, with the depth of any 3 of them combined, and is at its best when you play aggressively as opposed to opportunistically. Nioh 2 now has one of my favourite combat systems of all time, with 3 looking to improve on it.

Just look at what people are already doing with it

@KundaliniRising333 4080 and 14700k at QHD/120fps. Having tried it on PS, it plays better on PC so long as the framerate's locked, but it's more demanding than it should be. Try the demo, and play a good bit of the open world - performance seems to vary drastically based on CPU.

@Rich33 Yeah definitely, and not all that surprising for a TN port if we're being honest - Nioh 2 was worse at launch but ultimately shaped up pretty well. I'll give the demo a try on the PS tonight and pick it up there if everything is alright, double dipping when they get the PC version sorted out.

Despite their technical shortcomings, I adore Team Ninja combat.

@Rich33 I tried it on PC, and can confirm there are some issues - notably, a stutter when playing with an unlocked framerate. I'm playing locked 120fps and its smooth, but it gets fairly demanding in terms of CPU usage once you reach the open world. Older hardware might struggle a bit at that point. Noticed a bit of a slowdown during one of the particle heavy crucibles as well. I'll be doing a PS5 playthrough this weekend, so can't speak to its performance just yet.

Really can't wait for this one.

Calling Nioh 3 a souls-like is reductive. It has some of the meta elements (corpse runs, checkpoints, stamina-influenced combat), but the combat is closer to a character action game. If you play it like a souls-like, you're diminishing your own enjoyment.

There are legitimate concerns with how things are - access journalism, embargos, the four-point scale and how a number of reviewers are incapable of compartmentalising personal sentiments unrelated to gameplay when scoring a game.

This addresses none of them, and is a terrible idea. I'm not sure what the answer is; While following reviewers who's tastes you agree with works on an individual level, it doesn't help the industry when outlet scores guide executive decisions.

@nessisonett Kratos is an original character, designed by Charlie Wen. Voice and performance actors where chosen for characteristics they wanted in the character. TC Carson, Brandon Molale and Joseph Gatt suited the original games. Christopher Judge and Shad Gaspard where chosen for the more grounded Kratos in the Norse era. No actor owns the character, or has any bearing on his explicitly Spartan ancestry.

@TheGreyandTheRed Agreed, he was really good in SoA. Kid actors are always a wildcard, and this isn't an easy casting.

@Frmknst I'm legitimately interested in why you think 2 is amongst the best of games?

Personally, I'm of the mind that games that attempt to replicate film are at a disadvantage. A real-time camera system can't hope to match the expressive potential of a well designed shot-list. Environments have to be practical and intuitive, placing significant constraints on design and lighting relative to film sets. Sound design has to be systemic - even at its most curated, responding to triggers in a linear game, it's still limited compared to film, where sound design and scoring is done over finalised scenes.

Games have their own strengths - interactivity, agency and increased embodiment as a result. To me, it was an interesting exercise, and a remarkable achievement in terms of presentation, but largely failed to exploit the format beyond the length/depth games allow for. That is, I'm glad it exists, and that people are experimenting with the medium, but don't see it as an exceptional use of it, if that makes sense.

@BionicDodo That's the critic score, audience scores where lower at 7.4. It saw some attention, owing to it launching on GamePass, but quickly dropped off, with only about a third of players completing it. Steam performance was dismal. I'm with Sammy on this one.

It's a lovingly made niche game - some enjoyed the interactive experience, many wanted more gameplay in their game. Personally, I'd be interested to see if they can keep the strong visual direction while working in their action pedigree.

@PuppetMaster Psychological horror is a genre that many games fit into. The Evil Within 2 includes psychological horror, clumsy combat, symbolism, an otherworld and grotesque monsters - but it's not Silent Hill. It lacks the tone, dream-like quality, sparse exposition, subtlety and specific lore that made Silent Hill unique - similar to Silent Hill: F.

Silent Hill had a distinct identity. Before Short Message, everything tied back to the location (2, Downpour, Shattered Memories), the cult (1, 3, Homecoming), or its rituals (SH4). There was a central mythos. It also had an established tone, and the further games strayed from this (i.e. early western-developed entries), the worse they were received. Post-Short Message, it's the "Silent Hill Phenomenon" - any place with collective trauma gets fog and monsters. SH:F muddies this further - from mental illness and hallucinogenic capsules to Shinto gods in a conflict that may predate the phenomenon, to the original "powerful land" trope, and lazy allusions to vessels and White Claudia. Allegories are now spelled out in objective messages. Visual metaphors are explained through verbose monologues as they play out. My issue isn't F's premise - it's with shoehorning its disparate tone and unrelated ideas into an established franchise, undermining both. Just make a new IP. Konami won't though - corporate concern for the franchise is exemplified by Silent Hill pachinko machines.

"Actual gameplay" refers to the game loop. Stamina is a vague notion - battle royale and horror games both use it. How it interacts with mechanics, in part, determines the character's capability and the combat framing. Stamina is present in earlier entries, though more subtly and always to punish. In F, alongside perfect dodges and other ARPG staples, stamina establishes a combat flow more consistent with that of an action game, allowing aggressive play and extended attack chains with a little skill. Enemy frequency further reinforces this. Konami makes no pretences here - they explicitly stated that they sought to introduce "thrilling" action combat to appeal to younger audiences, selecting Neobards partly due to their history of producing action games.

As I said, it's a good enough game, but good enough to stand on its own - not to further dilute an already mishandled property solely for the sake of marketing convenience.

@PuppetMaster I don't think they should remove the fog; It just draws superficial comparisons, and if removed, I feel there is very little in tone or execution to tie it to the main series. That's my point.

1-4 are rooted in individual psychology - just not always the protagonists. SH1 is a manifestation of Alessa's personal hell. 4 is a projection of the antagonists psyche. Tangible though, possible due to the town and cult. F does bear superficial similarities to the series, but departs significantly, lacks all subtlety and has no direct connection to the property. Silent Hill is now just fog and monsters.

Stamina had little impact on actual gameplay in the originals, and perfect dodges (or stamina meters) didn't push it into action game territory. SH3 also had magical girl transformations and "sexy beams" as unlockable mechanics - maybe not the best indicator of intended tone. Combat simply wasn't foremost, and on higher difficulties never felt empowering. Being good at souls-likes, I was unstoppable from the outset in F. In subsequent playthroughs, I did no-damage boss runs on hard. The frequency of combat coupled with the action-oriented gameplay undermines the atmosphere the setting succeeds in establishing. I mean, play/watch a Kyubi boss fight and tell me your first thought is "Silent Hill".

The series has tried different things, but while maintaining a common style. Ambiguity, the eerie uncanniness, sparse dialogue and weird delivery, atmospheric story telling, trusting the audience to understand it. F changes nearly everything, all at once. Why force it into the franchise, other than marketing?

@PuppetMaster Silent Hill F is an interesting one. It's a good enough game; I just think if you removed the fog, and released it under another name, no one would have thought to call it a Silent Hill clone. And while I disliked the wolf-arm, that has very little to do with it.

Gone is the trauma rooted in an individual in favour of collective trauma. Gone is the desperate or clumsy combat in favour of perfect dodges and stamina meters. Gone are the motifs of decay, replaced with growth. Gone is the Lynchian storytelling and execution, replaced with wordy, explicit J-Horror.

The last point is likely why it'll never feel like a "proper" Silent Hill game. That IS Silent Hill, at least to me.

> like generative AI was used rather than an artist

This is the part I just don't see. What makes it look like AI? I don't think she looks more uncanny than any other recent female character. Just more attractive.

> primary driving factor behind the prompts was that they wanted something certain vocal male groups on the internet would find hot

So what? A design is inherently negative because a subset of people regard it positively?

> “fix” by giving her plumper lips and mascara

Lara has had plump lips and mascara from the outset. Making a billionaire heiress, described in her initial design brief as a "babe with brains", who became iconic largely as a result of her appearance, conventionally attractive doesn't strike me as a design failure.

@Ogbert What does that even mean?

@Nei Terrible analogy; Feeding an early draft to a LLM and asking for N possible variations on a set piece or bit of dialogue, from which a writer may or may not take inspiration when iterating, is closer. The process is about generating ideas to refine.

In its current form, GenAI is effectively throwing sh*t at the wall. Diffusion models don't have intent which makes them vastly inferior to a human creative for intentional design; For the same reason, they often produce combinations a human likely wouldn't think of. Not all GenAI is prompt based, either. You can quickly "polish" hand-drawn thumbnails to rapidly explore 10's to 100's of ideas, before committing to more mature, hand drawn designs that incorporate elements you liked from earlier exploration. This is the usual process, you're just automating additional detail into thumbnail sketches - detail which may or may not prompt new ideas.

TLDR; You're assuming AI output need be authoritative; It's just another idea to be "processed in meaningful, surprising and unpredictable ways by human thought and creativity". I swear people see promotional art and think that's what concept artists do most of the time. It's not. Countless variations are produced and discussed (with peers, leads, art/project directors) as to arrive at a design which moves down the pipeline. I'm not discussing or advocating for shipping AI assets, largely because Larian explicitly said it's not doing this.

@Nei Concept Art is largely about ideation and in-team communication. It's permutative and iterative. You create new ideas by combing and transforming existing ideas, repeatedly refining them to fit the vision. It isn't done in a vacuum; Those ideas come from the real world and fiction - films, books, and yes, other artists work. Throwing GenAI into the mix doesnt change this - it's just another source of potential ideas to start with, and a means of rapidly testing certain ideas early on in the process.

@Dogbreath I hear you - arguing suspension of disbelief in a game where you go guns akimbo against Incan velociraptors is probably cherry picking...

My point was that not much was required to begin with, and the two games have very different tones which require different design choices.

@Ogbert I mean, it could be as simple as "because the originals did".

Overthinking it, character design has considerations beyond well documented, generalised audience responses (the implicit appeal of aesthetics, the halo effect, etc.). A mundane/grounded character demonstrating improbable capability while facing impossible odds in an quasi-mythical setting creates juxtaposition, which would have to be addressed narratively. This is a remake; Derailing the original to do so would be a misstep IMO. I thought the "less glamourous" Lara in 2013 was a good choice for the same reason - being more of an origin/underdog story focused on survival.

On a tangent, I do a lot of climbing and the overwhelming majority of women I've climbed with wore makeup. Often quite a bit. SPF rated and/or water-resistant makeup is a thing.

@Ogbert What makes it "generic" though - I thought it was fairly faithful to the original designs, which are iconic?

Their art director deserves 3 seperate raises.

I'd love a good Tomb Raider game. I'm just not sure what the odds of a good Tomb Raider game are, given the leaks.

@get2sammyb There certainly is a lot of negativity out there, but I think hand-waving genuine complaints as simple negativity isn't fair either. Just look at the indie space - plenty of positivity. Same with retro-gaming, which has seen massive growth. To me, that indicates that the industry is failing to meet the needs of (a subset) of its audience.

@BAMozzy The polish of first party titles always impresses, but I think a number of sequels have taken a step back in other regards. Take Spiderman. The sequel cost more than 3 times as much to develop. The visuals saw improvement, and gameplay got some nice enhancements. But much of that budget went to the 300 minutes of cut scenes. The writing took a step back - pacing was poor, with Miles, Peter and many plot-lines being underdeveloped, while the villains where paper thin and the dialogue was... serviceable. Not what I'd call return on investment.

To me, personally, it feels like Sony has the wrong priorities. First party titles simply cost too much, and take too long to develop as to allow the experimentation or risks taken in previous generations. Cinematic, character driven story telling is great, but not every game has to try and replicate the success of Last of Us; You need a premise and cast suited to it, and writers with the chops for it.

Unless Sony invests heavily in memorable new IP, I won't be picking up a PS6. I have cause to keep a powerful PC, so the only meaningful reason to own a PS, beyond nostalgia (been along for the ride since the PSX), is the exclusives.

Previous generations had a number of legitimately memorable games. Games I still think about and return to. I can't imagine myself doing the same for the PS5 line-up 10 years down the line.

@EfYI While I mostly agree with the sentiment, it's for different reasons. Those time-honoured studios are nothing but a name. The quality of the products associated with them comes from the people behind the name. Some of those people left content with a long career. Many where systematically removed from the industry. Sometimes that was the result of external stakeholder pressure or the excessive commoditisation of the process leaving creatives frustrated. Often times it was from within the studios they helped establish - petty politics, infighting and new "talent" feeling entitled to a legacy they could never sustain.

The technical excellence one sees is the product of evolving pipelines, disproportionately more processing power and team sizes ballooning by a factor of 10-100. One can easily mass-hire for graphical fidelity. Legitimately imaginative worlds, iconic characters, memorable stories and innovative game systems... Not so much. At this point, seeing these studios shuttered leaves me conflicted. It's the end of an era, but also, in many cases, a mercy killing. No amount of funding can change the fact that, internally, they bear no resemblance to what they once where.

@judgmentarrows I'd rather they not - what you described is already a sizable number of AAA titles. For games leaning heavily into grit/realism (Last of Us), mundane designs are justifiable. For those not, how are "boring looking" character design not regarded as a negative - they always where, and would be for any other facet of art direction in that context?

I can understand not liking stripperiffic designs - that's certainly a matter of taste. I can't say I understand the aversion to female characters appearing feminine though.

As much as I'd like to see new IPs, or even main-series entries from TN, I'm looking forward to this. Been ages since I played Fatal Frame.

Yotei certainly is popular. As someone who loved the art direction of Tsushima, liked Jin's characterisation and the mythic tales, but found the combat lacklustre and open world to drag (a lot in late-game), is it worth trying?

@nessisonett Most titles, including massive first party titles, leave lines on the cutting room floor. It's not solely a matter of budget, it's a matter of perceived value of content and the logistics of the design, casting, approval, iteration and integration cycle not being feasible (most VOs are recorded late in production). I'm simply stating that there's a subset of lines, idealistic hypotheticals aside, that would have never made their way to an actor.

I'm generally not an advocate for "AI". I'm opposed to its use outside of simplifying and accelerating tedious tasks that undermine creative workflows (skinning, UVs, foley pre-processing); But even automating those could reduce job human work hours or job security. As do a litany of non-AI choices made over the past decades - licensing engines, use of asset libraries, short-term outsourcing. Hell, even procedural generation. Where does one draw the line? If the bar is set at "reducing job security", practically every AAA studio is guilty as sin and has been for a while.

@nessisonett Yeah and no. If they are lines which budgetary or logistical constraints would have precluded, which are pretty common, they would have never found their way to an actor to begin with.

Nah, plenty of backlog to work through and frankly, after the first, I have little interest in the second.

The writing used to be a big part of the charm of Obsidian games; After Outer Worlds and Avowed, it's now a reason to avoid them.

@tselliot Fast-paced, challenging and complex/varied gameplay. Many are equally confused by the appeal of cinematic heavy games in a medium defined by its interactivity. To each to their own.

@Questionable_Duck Steam doesn't have graduated scores - you either recommend or do not recommend a game. That score means 94% of players didn't think it bad, not that the average player thought it a 9.4

I really liked the dark 80's OVA-inspired aesthetic of NG1 & 2, and that seems largely absent in this one. Not sure I'm a fan of Ryu playing second fiddle either, or the girls not making it in as playable characters. Who am I kidding? It's a NG game. I'll put at least a hundred hours into it; Nothing quite compares, action wise, if you take the time to get good.

@Brundleflies21 You missed the extended cutscenes, drip fed amidst the genocide yet chock full of self-righteous indignation, that alone nearly bankrupt studios. Jokes aside, some are really good, but I am growing tired of every game feeling the need to embrace "gritty realism" and hours worth of cutscenes in what is meant to be an interactive entertainment medium.

@nessisonett It's a tricky one - one I'm not sure anyone "comes back stronger" from. 4 studios very publicly cut ties based on spurious claims, removed his contributions and in the case of EA, vowed to never work with him again. There've been no public retractions or apologies by said studios. It's a stigma he will likely have to carry, largely due to the response of the studios (and press) - presuming guilt.

You're right, stepping back is the professional response. Messy public witch-hunts based on unverified claims, not so much.

@nessisonett I remember reading that he and Mitsoda had concerns about the design of the game from the outset, so maybe there was no saving it? Everything hinges on executives actually listening to veteran developers concerns.

As for his firing, it should have never happened. There's a reason why presumption of innocence is the basis of any competent legal system.

@Romans12 Pretty much my feelings. Playing through the remake now and it really is excellent. Spooky souls on the other hand was a disappointment. It should have never released with Silent Hill in the title, for its own sake.

Disappointing, but unsurprising. They should have kept the original team, and Chris Avellone.

@Oram77 Unless that monopoly, in a misguided attempt to see return on investment, tries to "westernize" anime and lands up butchering it much like Netflix.

If nothing else, monopolies represent a single point of failure.

It sunk a studio; I doubt there's any conspiracy. Part of the problem was marketing - I know people who refused to redeem the free copy offered by EGS because of how the marketing rubbed them the wrong way. A simpler explanation would be that people who enjoy that flavour of writing find themselves in the company of similar tastes. Just many amongst the existing player base have different tastes. Subjective.

With that said, I'll admit that around the 5 hour mark, the "saints" became less overtly aggravating and more just boring and forced. I'm not sure if the writing improved or I simply became desensitized to it. By that point though, the uninspired gameplay was already taking its toll. Not what I personally would call a good experience.