After receiving mixed feedback following its first reveal, NVIDIA has officially introduced the updated DLSS 5 at SIGGRAPH 2026. Rather than focusing solely on higher image quality, the company is now emphasizing flexibility, giving developers much more control over how AI enhances graphics in their games.
NVIDIA Responds to Community Feedback
When DLSS 5 was first demonstrated earlier this year, many players praised its impressive visual improvements but also expressed concerns. Some felt that the AI was making scenes look too different from the developers’ original artistic vision by altering lighting, textures, and materials more aggressively than expected.
NVIDIA has clearly listened to that feedback. The updated version is designed to give game studios the final say over how AI is used, ensuring that image enhancement complements the original artwork instead of replacing it.
More Control for Game Developers
The biggest addition in the updated DLSS 5 is a new set of developer controls. Studios can now fine-tune how strongly the AI processes different parts of the image through dedicated parameters that adjust image structure and tonal enhancement.
Developers can also decide which objects should receive AI processing. For example, they may choose to leave character faces untouched while applying enhancements only to environments, reflections, or lighting. This allows each game to retain its unique visual style while still benefiting from AI-powered rendering.
Three AI Processing Modes
Another major improvement is the introduction of three separate AI models, each optimized for different quality and performance targets.
Developers can switch between these models in real time depending on the scene, allowing games to balance visual fidelity and frame rate more effectively. According to NVIDIA, this flexibility makes it easier for studios to optimize games for different hardware configurations without sacrificing image quality.
Improved Performance
One of the biggest criticisms of the original DLSS 5 demonstration was that it relied on two GeForce RTX 5090 graphics cards. The updated version addresses that concern by introducing a more efficient neural network.
NVIDIA says the optimized model significantly reduces computational requirements, allowing the technology to run on a single modern RTX GPU. The company has also reduced rendering latency by processing frames more efficiently, making the technology better suited for fast-paced games.
Release Window
NVIDIA has not announced a final release date for DLSS 5, but development is continuing alongside several game studios. The company is expected to introduce additional developer tools before the technology becomes publicly available later this year.
While it remains to be seen whether the updated version will change public opinion, NVIDIA’s new approach represents an important shift. Instead of letting AI take over the creative process, the company is giving developers the tools to decide exactly how the technology should be used. That balance between automation and artistic control could become the biggest strength of the updated DLSS 5 when it officially launches.
