A game can look more realistic without becoming a game you would rather play.
That is the question behind our recent XOTIC PC video about NVIDIA DLSS 5. The technology is interesting, and some of the visual changes are striking. But when we look at those changes, we also want to know what happens to the choices that gave the game its identity.
Does the new lighting strengthen the atmosphere? Does a character still feel like the same person? And when a familiar world looks different, is that difference something we actually want?
There is room to admire the engineering and question the result.
What DLSS 5 is changing
NVIDIA describes DLSS 5's new feature as 3D-Guided Neural Rendering. It uses information from a game's rendered frames, including color and motion data, to generate more lifelike lighting and material detail. That makes this discussion different from simply asking whether an upscaler produces a sharper image. NVIDIA's DLSS 5 announcement
The technology has also moved beyond its initial reveal. NVIDIA's September 2026 release information identifies NBA 2K27 as its debut game, with support for GeForce RTX 50 Series hardware and GeForce NOW. That release should not be confused with every community experiment circulating under the DLSS 5 name. NVIDIA's September update
For this conversation, though, the most important question is visual: does the treatment fit the game?
Realism is a creative choice
A shadow can do more than suggest where the light is coming from. It can hide a threat, make a room feel unwelcoming, or keep your attention on a particular part of the scene.
The same is true of a face. Its proportions, surface detail, and lighting all contribute to how we read the character. A smoother or more polished appearance might look appealing in isolation while feeling less appropriate in the story.
That is the concern we raise in our video. Game developers work within technical limits, but they also make purposeful decisions. Not every exaggerated shape or unusual color is a problem waiting for more processing power.
Sometimes the less realistic version is the version with more personality.
That does not mean photorealism is the wrong goal. In a game built around a lifelike presentation, added detail may strengthen the experience. We simply would not treat realism as a universal measure of quality.
Developers have controls, and those controls matter
Calling DLSS 5 an "AI filter" captures a reaction to the appearance of some clips, but it does not fully describe how NVIDIA says the technology works.
NVIDIA documents developer controls for structure and tone intensity, along with masking tools that let teams adjust where enhancements apply. Its explanation also describes choosing between models and treating characters differently from environments. NVIDIA's developer-control overview
That is important context. The existence of those controls does not mean every viewer will like every result. It does mean the debate should account for how a studio chooses to use the technology.
Our preference is for that choice to remain deliberate. If a particular treatment makes a scene more convincing while preserving its mood, great. If it makes a distinctive world feel less like itself, we would rather keep the original presentation.
What the World of Warcraft clip shows
One example that caught our attention is Chen Yang's World of Warcraft + DLSS 5 (3 Passes) video, which supplied the game frame used in this article's hero image.
The creator describes the footage as a real-time recording using three neural-rendering passes through an add-on on an unofficial server. They report an RTX 5070 Ti and a 30 FPS lock. Those are the uploader's stated conditions, not XOTIC PC benchmark results or evidence of an official Blizzard integration. Watch the source video
The clip gives us something concrete to discuss: a familiar game world presented with a different surface appearance and atmosphere.
Some players may find that exciting. Others may prefer the way the original visual style communicates the world. Both reactions can be reasonable. An experimental configuration also should not stand in for every possible developer-tuned implementation.
How we would judge the result
A dramatic screenshot is a starting point. For a more useful comparison, we would look at the same scene in motion and ask:
- Does it preserve the mood? Brighter or more detailed is not automatically more appropriate.
- Do characters still feel consistent with their world? Faces and materials should support the game's visual language.
- Does it hold up while playing? A still image cannot tell us everything about motion or responsiveness.
- What are we actually looking at? An official implementation, a community mod, and an edited concept video need different context.
We would also want to try it ourselves before making performance claims. This article is our perspective on the visual debate, not a hands-on DLSS 5 review.
Our take
We want graphics technology to give artists more ways to create memorable worlds and players more ways to enjoy them.
We can appreciate what NVIDIA is building without deciding that every game should move toward the same kind of realism. The strongest result is one that feels right for the game, whether that means a more lifelike environment or preserving an unmistakably stylized character.
Where would you draw the line? Would you welcome more realistic environments while keeping character designs closer to their original appearance, or judge the whole experience game by game?
For more PC gaming perspectives and practical hardware advice, explore the XOTIC PC blog.
