NVIDIA officially releases DLSS 3.5: Using supercomputing AI models, the quality of light-tracing images is doubled

NVIDIA’s DLSS can be said to be one of the most successful and widely used AI training models currently. It is precisely because of the use of DLSS that the game frame rate of the graphics card has been greatly improved, thus facing the full light chasing special effects. Don’t worry at all, there will be no serious lagging. In particular, the launch of DLSS 3 has made it possible to achieve hundreds of frames in a full light-tracing environment. Now, in addition to improving the frame rate of the game, Nvidia also wants to improve the image quality of the game through AI. Recently, NVIDIA launched DLSS 3.5, which mainly uses AI models to greatly improve the image quality of light tracing, thereby making the game image quality more realistic.

At the Cologne Games Show, NVIDIA held a briefing and officially released the AI ​​neural network technology for light-tracing games, also known as DLSS 3.5, thereby improving the image quality of light-tracing games. As a reflection of the current technological progress in the game industry, light tracing has obviously improved the quality of game images. Of course, it is very difficult for graphics cards to simulate light, and it also puts forward higher requirements for graphics card performance, and graphics card manufacturers have also Using various technologies to optimize the generated light, such as noise reducers, can effectively reduce the noise of the light, thereby making the light more natural. However, in the past, this noise reduction method was independently designed by NVIDIA, and was unsatisfactory in terms of training volume and processing capacity. Now, with the significant increase in AI computing power, NVIDIA has also introduced AI and launched an AI network trained by NVIDIA supercomputers to allow light to have higher pixels, thus greatly improving the image quality of light.

NVIDIA stated that simulating light is a very complex process and requires far more calculations than other game technologies. When using supercomputer training, the training volume of DLSS 3.5 is 5 times that of DLSS 3, which is fully demonstrated The advantage of the rolled version is that it is trained by AI. Therefore, this training model has a certain degree of universality and can be applied to different games. It can also identify the performance of light in the game and perform specific optimizations, which is quite effective. intelligent. Judging from the game frame rate given by NVIDIA, it seems that DLSS 3.5 does not improve the frame rate much. It mainly improves the picture performance of light chasing games.

Similar to the previous DLSS 3, if you want DLSS 3.5 to be successfully applied, the support of game manufacturers is indispensable. Starting from the fall, there will be games that support DLSS 3.5, including "Alan Killer 2" and "Cyberpunk 2077" " and "Portal". In addition, unlike DLSS 3, the technology provided by DLSS 3.5 can be applied in RTX graphics cards.

It can be said that in recent years, NVIDIA has been booming in the use of AI, especially the emergence of DLSS, which has given NVIDIA the right time and place. In comparison, AMD's new technologies here have been delayed, and the flagship graphics card is also quite Powerless, if this continues, I am afraid that the market share of AMD graphics cards will become lower and lower.

The above is the detailed content of NVIDIA officially releases DLSS 3.5: Using supercomputing AI models, the quality of light-tracing images is doubled. For more information, please follow other related articles on the PHP Chinese website!