NVIDIA wanted me to upgrade my RTX 4070 Ti for Multi Frame Generation, but I found another way

Gaming has been Samarveer’s greatest passion, and the Literature graduate in him takes immense joy in dissecting games for their themes, messages, and impact. Samarveer holds a deep appreciation of gaming, and considers the platform to be the most immersive and impactful across all media. He can be found engaging with gaming communities online, always ready to debate the finer points of ray tracing or itching to write an 8-page collegiate thesis on any game that impacts him emotionally.

When I bought my RTX 4070 Ti back in 2023, I didn't know that I was going to be upgrading my GPU multiple times over the coming years, without ever buying a new piece of silicon. Between newer DLSS models, Nvidia's Smooth Motion, and the endless tinkering I've done with DLSS 5, my GPU has been surprising me this far into its life.

I've spent the better part of this year discovering just how much more performance and visual fidelity I can squeeze out of hardware that Nvidia would much rather have me replace. This GPU's biggest selling point was 2x frame generation, and I was particularly proud of it, too. That changed after the RTX 50 series came out with Multi-Frame Generation (MFG) going up to 6x frame generation.

This was the only feature left for me to use on my GPU, and instead of waiting for Nvidia to finally add support for MFG to older RTX 40 series GPUs, I just went ahead and got it.

My RTX 4070 Ti has only gotten better with age

Nvidia's updates have given my last-gen GPU a second wind

Three years after purchase, my GPU has still been getting better with upgrades. Nvidia's Smooth Motion may have arrived exclusively for the RTX 50 series back in January 2025, but it was only a matter of months before my RTX 40 series GPU received the feature, too. I even used Smooth Motion extensively to realize how effectively it helped my frame-locked games.

DLSS 4.5 followed in 2026, and it brought newer upscaling models to my GPU. Thanks to improvements in image quality and temporal stability, DLSS 4.5 genuinely turned my three-year-old, 1440p-only card into a 4K-capable GPU overnight. The GPU itself isn't getting any younger, but its software is getting so much more interesting.

Then came DLSS 5, and naturally, it didn't take long for the open-source community to find a way to run Nvidia's neural rendering technology on older GPUs as well, including RTX 40 series cards and even some AMD graphics cards. The point of the matter here is that between community tools, unofficial implementations, and Nvidia's own increasingly sophisticated rendering techniques, my aging GPU has hardly felt obsolete at all.

There was, however, one conspicuous omission. Multi Frame Generation remained officially exclusive to RTX 50-series cards. Furthermore, unlike DLSS 5, Nvidia has made no announcement about any plans to bring MFG to older GPUs. As such, I decided to see if I could do something about that myself.

I finally brought Multi Frame Generation to my RTX 4070 Ti

NVIDIA drew the line at RTX 50, but modders had other plans

With the RTX 50 series GPUs, Nvidia introduced Multi Frame Generation, with up to 6x frame generation through AI-generated frames. This feature requires games to have native support, and truth be told, it's a rather convenient reason to spend money on newer hardware. Unfortunately, on paper, my RTX 4070 Ti simply wasn't invited to the party, regardless of how much GPU performance I had left to spare, or how many AI Tensor cores it had on the silicon.

Enter OptiScaler, the community tool that lets users experiment with upscaling and frame-generation implementations across any game that supports DLSS or FSR upscaling technology. Upgrading my GPU in this climate was simply out of the question, but thankfully, OptiScaler provided a way to work around the software restrictions separating my RTX 40 series card from the MFG feature set.

Getting everything working required some tinkering, but the payoff was immediately apparent in my testing. With MFG enabled through OptiScaler, my displayed frame rates climbed to three-digit numbers in Control Resonant and The Blood of Dawnwalker. It was impossible not to be excited about my own GPU all over again now that it was generating 4x frames instead of just 2x.

Here's how OptiScaler works around Nvidia's software lock

A little software trickery goes a long way

It's important to understand what OptiScaler is doing in order to unlock Multi Frame Generation on a GPU that doesn't officially support it.

Inside Nvidia's frame generation binaries are architecture checks which compare your PC's GPU architecture ID against the Blackwell-architecture ID to decide if MFG is supported.

A bit of a history lesson is required here — when Nvidia moved from DLSS 3's 2x frame generation to DLSS 4's MFG, it replaced Ada Lovelace's reliance on hardware Optical Flow Accelerator with an efficient software-based AI model. This new model ran entirely on AI Tensor cores, which are available on the RTX 40 series architecture as well, hence making the entire thing work. This model was faster, and used less VRAM, but most importantly, was backwards compatible with existing DLSS integrations.

OptiScaler effectively reaches into the frame generation module as it loads, and then rewrites the specific comparison instructions that check your GPU against Blackwell architecture IDs. It quite literally patches the individual checks inside the DLL file so that the frame-gen component behaves as if it were running on Blackwell architecture. In a nutshell, it makes your 40-series GPU report a 50-series ID to the game.

As a result, the game's own DLSS Frame Generation can advertise your GPU as MFG-compatible, unlocking 3x/4x/6x frame generation even on RTX 40 series cards. With the fake Blackwell architecture ID reported to the game, and since frame generation post DLSS 4 has been dependent on AI Tensor cores (which RTX 40 series cards have), even the last-gen card easily runs multi-frame generation.

There are drawbacks to using MFG on 40 series hardware, though

No free lunches, after all

Getting Multi Frame Generation running on my RTX 4070 Ti was exciting, but the workaround isn't without compromises. Ada lacks Blackwell's dedicated hardware Flip Metering, which helps coordinate frame delivery. Consequently, generated frames can arrive unevenly, producing judder even when the FPS counter looks spectacular.

Experimental OptiScaler forks attempt to address this through software-based pacing fixes, including options such as temporal fixes and legacy pacing. However, these aren't perfect substitutes for dedicated hardware, although that refinement could be on the horizon.

VRAM was my other real headache while using MFG on my RTX 4070 Ti. Running demanding games with ray tracing and even 2x generation pushed memory consumption higher than I'd have liked. So, even while OptiScaler was clearly letting me use 4x and 6x frame generation, it was at 3x frame generation with path tracing in Control Resonant that I ran headfirst into my 12GB VRAM ceiling, leading to crashes past that point.

GPU generations are rapidly becoming just software unlocks

Graphics cards used to be separated primarily by what their silicon could do. A "better" and "newer" GPU used to mean that it came with more shaders, faster memory, newer architectures, and dedicated hardware. These gave GPU manufacturers genuine reasons to push us all towards the next generation.

Increasingly, however, software compatibility is becoming the dividing line, with features restricted to newer products even when older hardware might manage them perfectly well under the right conditions.

I don't expect NVIDIA or AMD to abandon generational exclusivity when there's money to be made selling new GPUs. But projects like OptiScaler make me question how much of that obsolescence is actually inevitable.