NVIDIA’s work on classics such as Quake II and Unreal demonstrates how ray tracing, path tracing and Physically Based Rendering can transform the appearance of 1990s games. It also raises a broader question: when is it legitimate to modernize only some aspects of a game while leaving others untouched?
A partial modernization is not inherently a problem. The decisive factor is whether the contrast between old and new elements is supported by a clear artistic choice.
When modern lighting exposes old geometry
1990s games were designed as coherent visual systems in which geometry, textures and lighting reflected the technological limitations of their time.
RTX Remix can introduce path tracing, indirect lighting, reflections and refractions, dynamic shadows and volumetric effects while replacing original textures with 4K PBR materials using albedo, normal maps, roughness, metallic and emissive properties.
The underlying mesh, however, may remain almost unchanged. A normal map can simulate surface detail but cannot increase polygon density or alter a mesh’s silhouette. Consequently, the more photorealistic the lighting becomes, the more visible low-poly geometry can become.
The artistic choice
This is not necessarily a flaw in RTX Remix itself. It is primarily a rendering framework, not an automatic asset reconstruction system.
Retaining original geometry can be a legitimate artistic decision. It may preserve a game’s visual identity, deliberately emphasize its retro character or create a meaningful contrast between old and new.
The problem arises when there is no such intention. If the objective is photorealism, modernizing lighting and textures while leaving extremely primitive geometry untouched can produce an inconsistent visual language.
The same principle applies to any remaster: elements should remain technologically old because they serve the artistic vision, not simply because they were not modernized.
RTX projects therefore demonstrate that visual fidelity has multiple dimensions. Rendering fidelity and geometric fidelity are distinct but interconnected, and improving one can expose limitations in the other.
The most convincing modernization is consequently not the one that updates the most components, but the one that makes a coherent decision about what to preserve and what to rebuild - from meshes and silhouettes to characters, animations and effects.
Partial modernization is not inherently a compromise. It becomes a compromise when the parts left untouched have no artistic reason to remain untouched.