What is frame generation? An honest look at DLSS 3 and FSR 3
Frame generation can nearly double the FPS number on your screen — our calculator shows an RTX 4070 jumping from 57 to 97 fps at 4K with it on. But here's the part most explainers skip: those extra frames don't make the game respond any faster. Frame gen is a smoothness tool, not a performance fix.
It's one of the most marketed PC features in years, and one of the most misunderstood. So let's be straight about what it does, what it doesn't, and when it's actually worth turning on.
What does frame generation actually do?
Your GPU renders two real frames, and frame generation creates a brand-new frame in between them by guessing what the in-between moment looked like — then displays it. So the screen shows more frames than the card actually rendered. Nvidia's DLSS 3 (RTX 40-series and newer), AMD's FSR 3, and Intel's XeSS version all do the same basic thing, and most can roughly double the displayed frame rate.
It's a close cousin of upscaling (DLSS/FSR Super Resolution), but not the same: upscaling renders fewer pixels and reconstructs detail to gain real performance, while frame generation inserts extra frames for visual smoothness. Many games let you run both at once.
How much FPS does it add?
Here's a Ryzen 5 7600 + RTX 4070 in Cyberpunk 2077 at 4K, from our FPS calculator, with upscaling and frame generation toggled on and off:
On paper it's transformative — 57 to 126 fps. But notice we labelled that column displayed FPS, not real performance. That distinction is the whole story.
The honest catch: it doesn't lower input lag
A generated frame is interpolated between two real ones, so it carries no new information about your input — your mouse and keyboard are still sampled at the base frame rate. That 97 fps reading looks smoother, but it feels about as responsive as the 57 fps underneath it. Frame generation actually adds a little latency on top, because the pipeline has to hold a frame back to interpolate against, which is why Nvidia makes its Reflex latency-reduction tech mandatory whenever DLSS frame generation is on (GamersNexus).
We show the boosted number in our calculator because that's what your FPS counter will read — but it's the kind of thing we'd rather you understand than be sold. Smoother image, same responsiveness.
Why you need a decent base frame rate
As Nvidia itself puts it, frame generation is for turning playable frame rates into high ones — not unplayable ones into playable. Aim for a base of at least 60 fps before switching it on. Below that, two problems compound: the latency floor is already high, so the unresponsive feel is obvious, and the gaps between real frames are larger, so the guessed frames produce more visible artifacts — smeary edges, warped UI, ghosting on fast motion. Using frame gen to drag 30 fps up to 60 is exactly the wrong job for it.
When should you turn it on?
- Turn it on for single-player games on a high-refresh monitor (120 Hz+), when you already have a 60+ fps base and want maximum smoothness. This is where it shines.
- Leave it off for competitive shooters — the input lag matters more than the smoother picture, and that's exactly where you feel it.
- Don't use it as a rescue for a build that can't hit 60 natively. Lower settings, drop resolution, or turn on plain upscaling first — see our fixing a bottleneck guide.
- No point on a 60 Hz monitor — there's no refresh headroom to display the extra frames.
DLSS 3 vs FSR 3 vs XeSS
The big practical difference is hardware support. Nvidia's DLSS frame generation is locked to RTX 40-series and newer and is generally the cleanest. AMD's FSR 3 frame generation is open and runs on a far wider range of cards — including older GPUs and even Nvidia ones — at some cost to image stability. Intel offers its own for Arc. Whichever you have, the rules above are the same: get a solid base frame rate first, then let frame gen carry it the rest of the way.
The smoothness you feel still comes down to your real frame rate and its 1% lows. Frame generation polishes a good experience; it can't manufacture one.
Frequently asked questions
Does frame generation increase input lag?
Slightly, yes. The generated frames are interpolated, so your inputs still register at roughly the base frame rate — and the technique adds a little latency on top, which Nvidia Reflex (required for DLSS frame generation) works to offset. The displayed FPS rises, but responsiveness tracks the real frame rate underneath.
Is frame generation worth it?
For single-player games on a high-refresh monitor, often yes — it makes 60 fps look like 100+ with little downside. For competitive shooters, usually no: the input lag matters more than the smoother image.
What base FPS do I need for frame generation?
Aim for at least a 60 fps base before turning it on. Nvidia's own guidance is that frame generation turns playable frame rates into high ones, not unplayable ones into playable — below about 60 it tends to feel sluggish and show more artifacts.
Does frame generation cause visual artifacts?
It can, mostly on fast-moving objects, UI elements, and screen edges, since the inserted frame is a guess between two real ones. The higher your base frame rate, the smaller the gaps it has to fill and the fewer artifacts you will notice.
Can any GPU use frame generation?
It depends on the version. Nvidia's DLSS frame generation needs an RTX 40-series or newer; AMD's FSR 3 frame generation works on a much wider range of GPUs, including older and non-AMD cards; Intel has its own for Arc. Check what your game and GPU support.
See your real base frame rate first — then decide if frame gen is worth it.