DLSS 5 Sneaks Into Games Before Release
An early, unofficial build of Nvidia’s DLSS 5 technology has leaked and is already being injected into a handful of games that aren’t yet optimized for it. The result is a mix of comedy, discomfort, and, for some, exactly what they wanted.
The image above — often labeled “Claude Strife” — shows a screenshot that appears to run DLSS 5 at extreme settings. Unlike the usual DLSS, which uses machine learning to upscale low‑resolution graphics while staying close to the original look, DLSS 5 tries to enhance the image with AI‑generated details that mimic a “photorealism” prompt. Whether the Final Fantasy 7 Rebirth screenshot is a genuine DLSS 5 example is unclear; attempts to trace its source led to a dead‑ended Reddit thread, but it resembles other experimental images.
Stylized characters are a poor fit for this approach. The AI‑driven upscaling can make faces look overly textured, exaggerating skin pores, cheek and jaw separation, and sharpening expressions that were originally neutral. The effect is a distinctive, artificial sheen that many compare to the “studio‑lighting” look of AI‑generated portraits, reminiscent of Grok images on X.
In the RenoDX Discord community, users share mixed reactions. One comment reads, “Bruh, my Hogwarts legacy student turned into a 35‑year‑old divorce lawyer,” highlighting how the technology can reshape character models in unexpected ways. Another post shows a before‑and‑after of a Starfield character: the original flat, textureless skin becomes noticeably more human after the tweak, and many find the improvement satisfactory.
Opinions are split. While some on the Discord channel celebrate the new visual fidelity, a ResetEra discussion expresses horror at the AI‑driven alterations. Given my own reservations about generative AI, I find myself in the “not in my house” camp — appreciating the volume and texture added to flat faces, yet disliking the unmistakable AI‑generated aesthetic that feels out of place.
How are you receiving this technology?