The video discusses the recently leaked DLSS 5 technology from Nvidia, which was initially announced months ago but received overwhelmingly negative reactions. Critics felt DLSS 5 overly altered the artistic style of games, making characters look unnaturally “yassified,” akin to AI face filters, and worried it would encourage lazy development by replacing detailed facial rendering with upscaling shortcuts. Despite these concerns, industry figures like Todd Howard of Bethesda praised DLSS 5 as a major step forward in rendering for upcoming games like Starfield and Elder Scrolls 6. The presenter acknowledges the backlash and notes Nvidia has since introduced developer tools to allow selective application, such as masking faces to preserve artistic intent.
The video then shifts to a hands-on exploration of DLSS 5, demonstrating its performance and visual impact in Cyberpunk 2077. While DLSS 5 can improve the photorealism of NPC skin and environmental lighting, it comes at a significant cost, reducing frame rates by about 30%. The presenter notes that while some lighting and shadow adjustments look more natural, the performance hit is substantial, and the visual improvements may not justify the loss in smoothness for many players. The technology’s complexity is highlighted, including settings that allow developers to fine-tune its application and avoid affecting UI elements.
A major theme is how DLSS 5 struggles with games that have stylized or artistic visuals rather than photorealism. Because DLSS 5 is trained on real-world photographic data, it attempts to force stylized characters into a realistic mold, often resulting in uncanny, unsettling transformations. Examples include characters from The Last of Us and Fallout 4 looking distorted or aged unnaturally. The presenter argues that DLSS 5 works best in games aiming for true photorealism, such as Death Stranding 2 or Kingdom Come: Deliverance 2, where subtle lighting enhancements align with the game’s visual goals.
The video also raises concerns about the potential misuse of DLSS 5 by large studios and publishers. There is worry that companies might implement DLSS 5 primarily for marketing deals or performance shortcuts, bypassing artists’ creative input. Evidence is cited that some developers only learned about DLSS 5 when it was publicly announced, indicating a lack of consultation with art teams. This could lead to widespread adoption of DLSS 5 even in games where it undermines artistic style, driven by corporate interests rather than creative vision.
In conclusion, the presenter offers a nuanced view, acknowledging that DLSS 5 can enhance visuals in specific photorealistic contexts but is not universally beneficial and comes with heavy performance costs. They caution against its use in stylized games and emphasize the importance of preserving artistic intent. The technology is here to stay, with competitors like AMD developing similar neural rendering, so gamers and developers must be vigilant. Ultimately, the presenter personally opts not to use DLSS 5 due to the frame rate sacrifices and invites viewers to share their opinions on its future role in gaming.
