After 3 re-uploads, he finally has the balls to post the video to the public. Here's the pinned comment.
Everything surrounding this dude is just tiresome
Steve from Gamers Nexus downloaded the video. This is ILLEGAL under 17 U.S.C. § 1201 of the DMCA. "Fair use" is not a valid defense. The statute imposes strict liability for the act of circumvention itself.
Consider the points made in Louis Rossmann. "I'm ending @TheDMCALawyer's career: Randall Scott Newman disgraces the bar." YouTube, 26 Feb. 2026, https://youtu.be/xkH3n3ZPhwY
Yes, in that video Louis Rossmann grossly and deliberately violated the DMCA. What he is doing is illegal.
In other words,
The law is retarded and will continue to get even more retarded under these circumstances
The law is fine. Piracy should not be tolerated, and IP rights should be protected. Circumventing DRM is against the law, as it should be. You crying about it is inexcusable. The only people campaigning to change it are, I would presume, criminals. I remember Louis, for example, talking about Plex and torrenting movies. Not a good look, to say the least.
Oh dear, I falsely thought you were being facetious
Are you unaware that Akilah Hughes v Carl Benjamin happened 6 years ago and it was a total Sargon victory? And that's just one example. If Steve wanted to, he could rip out Kevin's head and shit down his neck, legally, and make him pay for the privilege, just with this one precedent.
Edit: it may amuse you that the above poster was so salty, he couldn't even reply in the thread, but instead went and called me an.. ekhm.. and I quote
Feel free to point and laugh at this RETARD FAGGOT
It's not the substance that is being disputed, it's the framing surrounding how other influencers cover the subject. Objectivity is what you'll find in things like Wikipedia articles (most of the time), not essays and exchanges of drama.
| Claim | Verdict | Notes |
| AMD/Intel cannot simply "clone" Nvidia's multi-year head start in upscaling/neural rendering | Accurate | Nvidia does have a long head start + dedicated Tensor hardware across RTX generations + a mature training/dev pipeline. Current materials still position DLSS as powered by Tensor cores and DLSS 4.5/5 as continuing, rather than replacing, that stack. AMD's current public positioning for FSR 4 is not "we matched Nvidia on all hardware", but that FSR 4 is built specifically for RDNA 4 / RX 9000-series-class hardware. Only Intel is a bit different with XeSS-SR having a more cross-vendor path, with fallback support via DP4a and cross-vendor shader-capable hardware. |
| Therefore AMD will never truly match Nvidia in quality | Overstated | "Hard to catch" != "can never match". It's entirely plausible that Nvidia keeps having the stronger overall stack while AMD/Intel close the gap in some aspects. |
| Hardware Unboxed's framing of "Is AMD about to catch up?" is inherently misleading because the race is structurally fake | Meh | Vendors are not running the same strategy, but it is perfectly coherent journalism to ask whether MD is "catching up" in output quality, feature quality, or practical user experience. Kevin's complaint isn't nonsense, but it does not invalidate the question or headline format by itself |
| Advocating broader support for RNDA 2/3 owners is the wrong consumer position, people should instead advocate a hardware-agnostic solution | Partly correct in principle, but too absolutist | Two separate questions in there. Is hardware-agnostic upscaling better for the industry? Yes. Solutions that can be deployed across ecosystems reduce fragmentation and vendor lock-in and help wider install bases. Does it mean it's wrong to ask AMD to extend support to older AMD cards? No. Still a legitimate consumer mask. If AMD can make an acceptable reduced-complexity model for older hardware, that's a real benefit to current users. To the extent "broad industry-optimal solution" and "only legitimate thing to advocate" are equivocated, false, otherwise true. |
| Anything that would work on RDNA 2 should work on any mainstream non-AMD GPU | Not established, bad case of overstating | At a very abstract level, if a technique can be implemented using broadly available shader instructions / low-precision integer paths, then cross-vendor support may be possible. Unfortunately, "possible in principle" != "should work equivalently in practice". There's real constraints. So the claim is too broad as stated. |
| No, it's not vendor-agnostic. Stop forgetting about consoles! | Mixed, but important valid point | If "vendor-agnostic" == "usable across the actual production ecosystem that shapes modern game development, including consoles" then it's a more practical standard than "can a PC GPU from another vendor run some version of the algorithm". He is also right about the broader production logic (console baselines do heavily influence engine budgets, content budgets, and optimization priorities -> these things then propagate to PC) and the industry dynamics. However, that doesn't make "runs on multiple PC vendors" a meaningless category. |
| Console baselines determine PC optimization to an important degree [presented in a flashback from a year ago] | Accurate | PC ports are often inherited from console-targeted memory budgets, CPU budgets, feature choices, temporal reconstruction assumptions, and content-production constraints. It is rare that PC ports are designed from a blank slate. |
| Bad visuals and poor optimization can be predicted from footage by noticing TAA abuse / reconstruction dependence | True, but risky if stated too strongly | One can absolutely sometimes predict trouble from footage. However, footage does not reveal CPU bottlenecks, the driver state, RT cost structure, shader complexity, frame pacing, scene-specific bottlenecks, to name a few things. So it's a useful heuristic, but not a decisive method. |
| "[I'm not going to address the side drama], moving on." | How convenient | Dude's spending a fair amount of time signaling that he will not dwell on the drama... while in fact dwelling on drama. Irrelevant for technical competence, but I'll point this out as a catch-all for everything unrelated to technical matters. |
| Influencers/news are the wind that controls market demand | Overstated, but substantially correct otherwise | Coverage influences demand and incentives, but it does not determine them unilaterally. However, review culture does help decide what graphics questions are in the foreground. |
| Vendor-agnostic image quality is pro-consumer, vendor-locked techniques are anti-consumer | Too binary as stated, but mostly correct | "the industry should not let proprietary reconstruction become the center of graphics progress while baseline, broadly available image quality stagnates or regresses" is a perfectly defensible consumer-side position. And obviously broadly deployable techniques are usually better for users because they reduce fragmentation and lock-in. |
| Karis's/Epic's philosophy is the weight holding down vendor-agnostic image quality | Overstated personalization, but substantially correct | Epic helped normalize a rendering philosophy that tolerates heavy temporal reconstruction and weaker native clarity in exchange for broader feature ambition. |
| FSR 1-3 were poor vendor-agnostic methods because they mirrored the Karis-gimped standard more than DLSS | A bit compressed, but correct | FSR 1 is a spatial upscaler, 2 and 3 are temporal approaches in the broader reconstruction family. However, all three largely operated within the same reconstruction-heavy paradigm that had already become dominant. |
| DLSS is essentially the most developed version of the Karis-gimped standard, plus ML hardware | Correct | DLSS is indeed the most mature, most heavily resourced, most hardware-accelerated member of technologies building on real-time rendering pipelines that lean heavily on temporal accumulation, reconstruction, and history-based stabilization. |
| AMD and Intel are "leeching" on Nvidia by creating DLSS-like tech and copying things | Bad rhetoric, but the direction is correct | Once Nvidia succeeded in defining the prestige frontier around ML-assisted reconstruction, rivals were indeed pushed into competing on that same terrain instead of redefining the game around cleaner vendor-agnostic rendering standards. Minus points for contemptuously calling perfectly ordinary competitive imitation "leeching" |
| If media says DLSS is better, Nvidia goes up; if media says native is better, Nvidia is incentivized to improve DLSS, therefore it's a lose-lose | Not sound, but the direction may be insightful in terms of strategy | Praising vendor-agnostic or native image quality does not only incentivize Nvidia because it can also pressure devs, engine vendors, AMD, Intel, and reviewers to pay more attention to baseline image quality than only feature race comparisons. However, "if discourse remains centered on Nvidia's feature leadership, even criticism can reinforce Nvidia's central position - the way out is to redirect attention to the quality and desirability of the baseline rendering standard itself" is indeed a good strategic observation. |
| The only thing that will lower Nvidia's market value is sustained focus on vendor-agnostic image quality | Unsupported as stated, but a weaker version is perfectly defensible | Obviously Nvidia's market position is shaped by a long list of factors beyond game reviewers. However, if Kevin's specific concern is the cultural and review discourse ecosystem that legitimizes reconstruction-heavy vendor-fragmented graphics progress, then the most direct corrective is indeed a sustained focus on vendor-agnostic image quality. |
| Being a marketing slave doesn't mean explicitly advertising DLSS, it means rewarding the culture that makes DLSS thrive | Plausible and probable | Under the structural definition of "marketing slave" as Kevin argues for (as opposed to "you literally praise Nvidia on command") his target is indeed media framing that normalizes and rewards a certain technological trajectory. That is, in that narrower sense, it is indeed a real possibility that a channel can oppose vendor misconduct in some contexts while still reinforcing a vendor-favorable technological narrative through the categories it chooses to emphasize. |
| Gamers Nexus saying RTX Remix is cool proves the criticism | Too strong as stated, but partly fair otherwise | "praising RTX Remix without foregrounding its ecosystem-shaping downsides can reinforce Nvidia's preferred narrative" would indeed be legitimate criticism. But the literal accusation was overstated. |
| RTX Remix is a big ploy to turn modders into Nvidia minions | Unsupported as stated, but there is a serious structural concern | The structural concern of a Nvidia-led modding platform that encourages mod output to be built around Nvidia's rendering stack and feature priorities being capable of shifting hobbyist effort and modding prestige toward a Nvidia-centered ecosystem is real. But the "minions" framing is just bad packaging |
| Nvidia is turning graphics modding into a vendor-locked activity | "cannot run elsewhere at all" is too absolute, but a weaker formulation is substantially credible | Nvidia's own docs and public material describe RTX Remix as using a runtime for classic DX8/9 games and path-traced remastering, the toolkit/runtime are open source, and DXR-capable hardware can in principle run Remix-based content. However, even when something is not exclusive in the sense of a hard lockout, it can still be ecosystem-biased, as tools are designed by Nvidia, showcase materials foreground RTX features, performance/quality expectations are centered on Nvidia strengths, and the cultural prestige of the graphics modding scene can shift towards "RTX remaster"-like work |
| The mods can be played on any hardware with DXR, but the graphics technologies provided by RTX Remix are designed to perform poorly on other vendors and even on older Nvidia hardware | Partly wrong, partly overstated | Remix relies on DXR-class capabilities rather than an absolute hard lock to GeForce-only execution. As in, the official materials frame it around RTX remix runtime/toolkit and DXR/path-traced remastering, not around an outright "GeForce only" blockade. However, the high cost of path tracing, heavy RT workloads, and use of Nvidia's preferred stack do not by themselves prove intentional sabotage of AMD/Intel performance. As long as there are benign explanations like "Nvidia optimizes primarily for its own architecture" or "path tracing is expensive in general", Kevin would need more than "it runs worse elsewhere" to prove intentional anti-competitive design in the strong sense. It is certainly plausible that RTX Remix encourages rendering choices and content pipelines that tend to favor Nvidia's strengths and can have poor practical outcomes on weaker or non-Nvidia hardware, but "designed to perform poorly" is a step beyond what's shown in the video. |
| RTX Remix creates a need for DLSS performance gains | True, but limited | It is a true structural observation that RTX Remix-style rendering shifts the practical center of gravity towards the performance-recovery technologies that Nvidia leads in. The stronger intentional claim "therefore it was built in order to create that dependence" is plausible, but unproven. |
| RTX Remix doesn't provide perspective shadow maps, competent BRDFs, etc. | Handled too broadly, can be coherent if read charitably | Sloppy wording. No specification of what "competent" means, and "it doesn't provide perspective shadow maps" only matters if he's arguing the platform structurally steers modders away from certain more appropriate vendor-agnostic techniques. Charitably, RTX Remix does not give equal emphasis to a wider menu of more broadly deployable, artistically controlled, vendor-agnostic graphics improvements. |
| Gamers Nexus spent time on DLSS/RTX coverage instead of the underlying regression issue | True if the evidence he shows accurately represents what actually happened | The fact that reviewers often analyze the compensatory technology more than the underlying rendering regression said technology compensates for is a real pattern across tech and media coverage in general, not just with GN. However, it's not a knockout, because it only means that coverage is incomplete relative to Threat Interactive's framework. |
| TXAA vs SMAA | True | He is pointing out a real underlying historical contrast, in that there was a vendor-defined temporal AA path on one side and a more generally deployable AA approach on the other |
| Why did it take 13 years for a mainstream voice to show this? The answer is basically obvious corruption or cool-factor chasing | Overstated, but defensible | Mainstream coverage often focuses on the shiny compensatory technology than the older baseline alternatives that technology replaced, and that framing can distort what audiences think the real issue is |
| Mock "vendor-locked method worse than vendor-agnostic method, not cool enough; AI rebrand of resolution reduction, greenlight that investigation" sequence | Coherent | As Nvidia explicitly markets DLSS as neural rendering and AI-driven image-quality/performance improvement, it is perfectly plausible that "AI gaming" became a more clickbaity and culturally central story than "baseline AA/image-quality regression". |
| Influencers have been contributing to this imbalance for over a decade | Overstated, but defensible | If influential channels consistently push the vendor feature race more than they push baseline image stability and cross-vendor rendering quality, then they can help normalize the very trajectory TI is attacking |
| Supporting us instead of them makes that suppression end | Just rhetoric | At best, changing audience attention can change discourse margins, but it is by itself insufficient to guarantee a reversal of industry rendering trends. |
| DLSS 5 looks uncanny, but even if it looked great, it doesn't matter because it's not vendor-agnostic | Overstated, but defensible | A vendor-specific technique can still matter technically, aesthetically, and competitively, even if one is opposed to its effect on the ecosystem. That is, vendor lock-in is a valid criticism, but it does not remove technical significance. However, it is true that the underlying consumer problem remains unsolved if the improvement is delivered through a proprietary Nvidia-only path instead of through a broadly deployable rendering standard. |
| Baseline vendor-agnostic image quality regressed while "fake optimization" and reconstruction took over | True | "Modern" graphics indeed compensate for heavier rendering choices with reconstruction-heavy techniques, and that can mask or normalize a decline in baseline image clarity/stability relative to older, cleaner vendor-agnostic output. |
| NVidia marketing material / Assassins Creed Shadows comparison being "objective improvements" versus "AI art direction changes" | True | Once a neural model is changing the apparent lighting, it's a change of look that goes beyond a purely neutral restoration. Nvidia's own press wording ("Infuses Pixels With Photoreal Lighting and Materials") shows that it's already well beyond plain upscaling language. Given that this is the case, reviewers should not discuss DLSS 5 as if it were only a neutral image-clarification tool. |
| Material lighting response and AO are being implemented incompetently in "modern" games | Reasonable | This is substantiated by Nvidia's own examples, because Nvidia's marketing is highlighting lighting/material/AO-like qualities rather than only raw resolution. That is, the things Nvidia advertises as improved-by-DLSS-5 are areas where many contemporary games already have unsatisfying or fragile baseline results, ergo the AI layer works as a corrective on top of underlying rendering weaknesses. |
| Praising DLSS 5 is praising a band-aid instead of looking what the industry has already achieved | Polemical, but true | DLSS 5 is explicitly pitched as adding more photoreal lighting/material response at the pixel stage, not merely scaling a cleaner base image. The implication is that reviewers should distinguish between foundational rendering quality and a proprietary neural correction layer, and should not treat successes in the latter as equivalent to success in the former. |
the enemy of my enemy is a lolcow
This is untrue. They showed it over top pathtraced (ground truth) resident evil, and DLSS 5 ignored actual scene lighting, dulling highlights and creating light from nowhere. Their Zora demo is also pathtraced and exhibits the same nonsensical lighting with DLSS 5 enabled. Trace the shadows with DLSS 5 on vs off and with it on things are shaded with no regard for actual light, especially when with it off it is actual ground truth rendering because of path tracing. Kevin's entire diatribe over 'incorrect' or 'outdated' lighting models is the same as his previous diatribe obsessions over overdraw, he learns about something and becomes hyper fixated on that one thing believing he knows more than anyone about it. He hasn't actually worked on anything (and refuses to learn), so he doesn't understand limitations, time, or budget constraints.
I was referring to that specific Asscreed Shadows comparison image Kevin showed in the video, I only assessed the claim based on what he showed
No, this contradicts the available evidence. His pattern justifies the inference that he has real architecture-level competence but drags himself down with weak calibration, high ego, and chronic overreach. That is bad enough as it is, no need to invoke mount stupid, it didn't do anything to deserve this
Even if there were a vendor-neutral integer path, it's still the wrong approach according to Threat. There shouldn't be a reliance on any post-process temporal accumulation to begin with.
Absolutely
Even if such a neutral system existed, it would not answer the deeper objection TI makes. His actual position is that the industry relies too heavily on temporal/reconstruction-based methods in the first place (aka "Karis-gimped"). That is "we should decouple the interface from vendor branding" does not by itself imply "that makes the underlying temporal-based method acceptable"
No offense but did you use an LLM to write this?
No offense taken, I did rely a bit on LLM assistance in fact-checking because of how utterly dogshit search engines are in currentyear.
I don't think anyone outside of misguided influencers or Nvidia marketers claimed such a thing, no? Neither did TI criticize DLSS 5 for an inability to somehow magically "create accurate lighting"? It'd be like criticizing a horse for lacking diving skills
