Has the art team of *Hogwarts Legacy* taken sides too—Is DLSS 5 really just an AI filter?

Compared with the generative AI image creation that players are familiar with, the core of the controversy around DLSS 5 is actually more like a clash between “real-time rendering upgrades” and “conceptual misunderstanding.” If you compare it to the years of debate on the PC platform, from DLSS and FSR all the way to frame generation, this divide is nothing new—just more heated this time.

The issue began when Nvidia unveiled DLSS 5, triggering one of the fiercest technical controversies of the year. Technical media outlets and experts, including Digital Foundry, have generally given it very high praise. On the other side, anti-AI voices remain persistent, with some directly defining it as an “AI filter” and even arguing that it would damage the original artistic style.

But veteran game artist Georgian Avasilcutei, who has worked on projects such as Life is Strange, Dishonored 2, and Hogwarts Legacy, publicly came out in support of DLSS 5. His view is very clear: many critics do not truly understand what this technology is doing. In his words, DLSS 5 does not rely on prompts to “generate a new image,” nor does it arbitrarily alter character faces, models, or textures out of thin air.

Viewed from another angle, DLSS 5 is closer to a “low-cost restoration” of high-end rendering quality. The core logic presented in the original text is that it reconstructs and enhances lighting and shading based on data already present in the game itself. In other words, the models are unchanged, the textures are unchanged, and the lighting setup has not been overturned; what changes is the final on-screen quality of light and shadow, as well as the upper limit of how materials like skin and hair can look.

Avasilcutei also used comparison images to make the point: the same character model can look dramatically different under standard rasterized rendering with a basic shader versus being placed in a better ray tracing and shading setup; but that change does not mean it has become “a different character.” This is actually easy to understand—like the difference between a PS5 performance mode and a PC high ray tracing mode in the same game: the character is still the same character, yet the visual impression can feel almost like a generational leap.

From a development perspective, this statement is more valuable as a reference than simple corporate marketing. Art teams naturally want their work to be presented under the best possible lighting, but real-time rendering has long been constrained by performance limits. By contrast, if DLSS 5 can truly raise the quality of constrained lighting and shading performance without altering core assets, then it is more like “unlocking the upper limit” for artists rather than “redrawing” the art for them.

Of course, the controversy will not end just because one veteran artist spoke up. The most sensitive point in the player community right now is still the word “AI” itself. Avasilcutei even mocked some opponents by invoking the “Dunning-Kruger effect,” arguing that some of the criticism is more of a knee-jerk reaction than something grounded in technical understanding.

If we compare it with similar technical approaches, my judgment is this: what DLSS 5 really needs to answer going forward is not whether its name includes AI, but whether, once implemented in actual games, it is stable, clear, and closer than native solutions to the effect developers actually want. What PC players should focus on is never the conceptual label, but real gameplay comparisons. As long as it does not alter models, replace textures, or remake characters, then it is simply not in the same category as generative AI image creation.