NVIDIA is trying to limit the operation of unofficial modifications that enable multi-frame generation/MFG on RTX 40 graphics cards. Via the NVIDIA App, developers secretly distributed an OTA update to DLSS-G providers (branch 310.9) that changed the names of the temporal core and descriptor structures, temporarily disrupting mod injection. However, enthusiasts bypassed the new hurdle almost immediately.
When announcing the RTX 50 architecture, the Greens positioned multi-frame generation (creating multiple interpolated frames instead of one) as strictly hardware-specific. Some argue that the hardware optical flow accelerator on the RTX 40 chip is unable to produce multiple intermediate frames, and that for correct frame spacing, a dedicated frame queue hardware metering unit, which is only present in the RTX 50, is urgently needed. Modders were quick to emulate this technique from FSR frame generation, which showed that no hardware unit was needed to arrange the frames, and now they have bypassed the limitations of using DLSS MFG since it was originally purely software.
Update 310.9 disabled the modification for only a few hours. Overnight, the author released multiple hotfixes at once, restoring operation of DirectX 12 and Vulkan projects in 3x, 4x and higher modes, which have integrated the Streamline pipeline.
MFG’s current situation repeats the story surrounding DLSS 5 neural rendering. Initially, NVIDIA announced the technology as exclusive to the RTX 50 as well, but it was forced to officially confirm support for the Ada Lovelace generation after a series of fans ported it to all previous RTX and even AMD Radeon graphics cards.