DLSS 5 from NVIDIA: Why the “Beauty Filter” for Games Doesn’t Work

When NVIDIA introduced DLSS 5, the company presented the technology as the next big step in graphics: not just upscaling or frame generation, but a system capable of making the image more photorealistic in real time. In theory, DLSS 5 is supposed not just to improve sharpness, but to actually “enrich” the picture — lighting, materials, faces, and the overall cinematic quality of a scene.

But the reaction to the presentation showed that the industry has far more questions here than excitement. Because instead of “wow, this is the future,” a significant part of players and journalists saw something else in DLSS 5: an AI filter that makes games look “prettier” by template, but not always better.

The Reaction of Players and Journalists to the Presentation, The Instagram Beauty Filter Effect, The Uncanny Valley Effect, Is DLSS 5 Just a Regular AI Filter?, Is Targeted Use Possible?, What Happens Next?, Conclusion

The Reaction of Players and Journalists to the Presentation

The main conclusion from the first showcases of DLSS 5 is that the technology caused debate rather than admiration. And this debate is not only about image quality, but also about whether AI has any right at all to interfere so deeply with the artistic look of a game.

Many viewers noticed that after processing, character faces and scene materials looked more “glossy,” “smoothed out,” and as if slightly “retouched.” That is exactly why phrases such as AI slop, beauty filter, and Instagram-face for games almost immediately began to appear around DLSS 5.

At the same time, the problem is not that the technology looks “bad” in the literal sense. The problem is that it too often looks not like the work of artists, but like the work of an algorithm layered on top of the artists. And for game graphics, that is already a serious criticism.

The Instagram Beauty Filter Effect

The comparison with an Instagram beauty filter turned out to be the most accurate and the most painful for DLSS 5.

Why? Because the technology is indeed often perceived as a system that makes the image:

• slightly softer,

• slightly “cleaner,”

• slightly more cinematic,

• slightly more “perfect.”

In photography, that approach may work. But in games, it breaks one important thing: the authorial style.

Game graphics are not just a set of “dirty” pixels that need to be improved. Artists deliberately create:

• rough materials,

• harsh faces,

• hard shadows,

• gloomy or cold scenes,

• stylized rather than “perfect” textures.

When AI starts to automatically “beautify” all of that, it is not simply improving the image — it is standardizing it according to one neural-network standard of beauty.

And this is exactly where DLSS 5 currently looks especially controversial: it is trying to make the image not what it was intended to be, but what the model itself considers “more attractive.”

The Uncanny Valley Effect

The second major problem with DLSS 5 is the uncanny valley effect.

The technology strives for photorealism, but often ends up in a dangerous middle zone: the image no longer looks stylized, yet it still is not perceived as truly alive.

That is exactly why processed faces, lighting, and materials may cause not admiration, but mild discomfort. Everything seems “beautiful,” but something about that beauty feels unnatural.

This is especially noticeable in:

• skin,

• facial expressions,

• eyes,

• highlights on the face,

• the overall balance of light and shadow.

In other words, DLSS 5 often does not make a scene more convincing — it makes it too correct, and therefore less alive.

And that is the key problem: a game is not an advertising render and not an Instagram story. Sometimes unevenness, roughness, and imperfection work for atmosphere better than any “enhancer.”

Is DLSS 5 Just a Regular AI Filter?

Formally — no. And it is important to be precise here.

DLSS 5 is more complex than a regular filter because it:

• takes motion vectors into account,

• works not only on a single frame, but over time,

• tries to preserve image stability between frames.

So this is not a basic preset from a photo editor.

But here is the problem: judging by the way the technology affects the image, it really is perceived as a very smart AI filter layered on top of the final frame.

Based on the available explanations, DLSS 5 does not work as “a new way to render the world from scratch,” but rather as a system that takes an already formed image and improves it after the fact using an AI model.

That is exactly why many critics came to a conclusion that is unpleasant for NVIDIA:

DLSS 5 is not a new graphic reality, but a very advanced layer of post-processing. Technically, that may sound harsh. But from the point of view of perception, it is very close to the truth.

Is Targeted Use Possible?

And this is where DLSS 5 really does have a chance for the future.

Because the very idea of an AI tool for improving visuals is not bad. What is bad is that right now the technology is perceived as too global.

If such systems were used more selectively, they could be genuinely useful:

• for individual materials,

• for local lighting,

• for specific scenes,

• for narrow visual tasks.

In exactly that format, DLSS 5 could become not a “beauty filter for everything,” but an artist’s tool.

The problem is that right now the technology looks not like a finely tunable instrument, but like an automatic enhancement layer applied over the entire game. And that is what creates rejection.

In other words: DLSS 5 has potential, but only if NVIDIA learns to give developers real control, rather than just AI cosmetics by default.

What Happens Next?

With high probability, DLSS 5 is not going anywhere. On the contrary, the technology will almost certainly continue to be pushed heavily, because the promise of an “almost Hollywood-level” image is too tempting both for marketing and for major publishers.

But from here on, everything will come down to one thing: whether NVIDIA can turn DLSS 5 from an “automatic enhancer” into a controllable artistic tool. If not, the technology will get stuck with a very toxic image: not as a “new stage of graphics,” but as an expensive AI filter that kills a game’s style.

If yes, then attitudes toward DLSS 5 could still change dramatically. Because the very idea of using AI for careful visual enhancement is not absurd. It just looks too aggressive, too universal, and too self-confident in its current form.

Crimson Desert: A Beautiful Long-Running Project That Failed to Become a Revelation

Conclusion

The main problem with DLSS 5 is not that it “ruins graphics” or “doesn’t work.” It does work. Moreover, in some places it really does make the image look more impressive.

But more impressive does not mean better.

Today, DLSS 5 is perceived as a technology that too often tries to make a game prettier according to its own taste, rather than according to the logic of its artists and designers. And that is exactly why it currently looks not like a revolution, but like a very expensive and very controversial beauty filter for video games.

If NVIDIA learns to make this tool more precise, more local, and more subordinate to the artistic style of a specific game, then DLSS 5 still has a chance to become something truly important. But right now, the main conclusion after the presentation sounds very simple: game graphics cannot be improved the same way you improve selfies on Instagram.