RonsonPL wrote: ↑20 Aug 2026, 05:40
kyube wrote: ↑20 Aug 2026, 05:16
Have you considered pairing up those low frame rate games with frame generation?
Nvidia Smooth Motion, DLSS FG or LSFG (fixed 2x/3x) do wonders in unoptimized titles.
This would allow you to run ULMB2 @ 120Hz, which is noticably sharper than ULMB2 @ 60Hz.
I do hold some hope for FG, but it's still far from perfect and I'll need a Pulsar or OLED monitor to properly test it to form an opinion. I'm afraid I might decide it needs 120+fps real frames to be usable, but hopefully not.
My new PC is still in the boxes, some boxes are still in the stores (not purchased) so it may be a while before I form any judgement.
Thanks for the info though!
Pulsar doesn't really need framegen as badly as OLED does, as the organic purity of VRR strobing can work well enough if you keep your framerates in the triple digits. Pulsar does a fantastic job of motion blur reduction that you typically don't need framegen help most of the time.
Now framegen can help push 60fps content to 120fps+ needed for the VRR strobing of Pulsar, so that's where 2x framegen works well there.
Now if you use an OLED:
There's merits to using framegen as a blurbusting method, but I ensposue "GO BIG OR GO HOME".
2x framerate = 50% less motion blur on OLED, if you have Hz headroom
4x framerate = 75% less motion blur on OLED, if you have Hz headroom
10x framerate = 90% less motion blur on OLED, if you have Hz headroom
So from this:
2x framegen = not enough to be worth it as a blur busting tool instead of BFI for me
Old framegen = not high quality enough to be worth it (artifacty interpolation or AI slop supertanker)
So you pay lots of money for good 4x framegen (e.g. DLSS 4.5)
Framegen is a good new blurbusting option but it works better with modern games (e.g. CP2077) than retro games (e.g. SEGA). This is because the better framegen (e.g. DLSS 4.5) with the least "AI slop artifacts" requires direct access to the GPU's Z-buffer and game motion vector data.
I recently did a playthrough of CP2077 using DLSS 4.5 framegen (DLSS spatial upscale turned off + DLSS framegen only operation as a blur busting method). I was able to play RTX OFF graphics at over 300fps, at nearly CRT motion clarity, in Cyberpunk 2077.
OLEDs are the best way to use framegen-based method of blur busting, since DLSS 4.5 can unofficially do 6x framegen with RTX 5080 and 5090.
I tested:
1. DLSS 4.5 spatial - upscaling only (Transformer Model)
2. DLSS 4.5 temporal - framegen only (No spatial upscaling)
<-- My new CP2077 preference
3. Both
There are artifacts with all 3 approaches, but there's less artifacts with approach 1 or 2, than with approach 3. If you're using a low-pixel-density display such as a 45" 240Hz OLED ultrawide at 1440p, then item 2 often works better with OLED blurbusting. If you're using 360Hz or up, you'll be tempted to attempt 6:1 DLSS 4.5 framegen to see how it blurbusts a 360Hz+ OLED.
In CP2077 specifically, I often see artifacts with all approaches, but artifacts of approach 3 can sometimes be more distracting than approach 1 or approach 2 alone. Surprisingly, approach 1 and approach 2 was almost equal in "distracting-factor" when settings were carefully tuned. The distractions are mercifully sufficiently rare and non-annoying enough, that the blur busting factor upgrade exceeded BFI for me personally, at least for the specific game (Cyberpunk 2077 + Liberty DLC + new playthrough) since I kept the full color and brightness.
I did have to make do with a minor amount of motion blur -- but less motion blur than a Pioneer Kuro Plasma TV. Not bad for flickerless-based blur busting methods, and light years away from garbage yesteryear interpolation.
For 90% of content in Cyberpunk 2077, Item 2 looked almost as good as running framerates organically in CP2077, and I got the delicious CRT-quality colors with "better than plasma" motion clarity in CP2077 -- without needing the flicker of impulse-driving methods. You will occasionally notice motion artifacts of framegen, but instead of noticing it 90% of the time in ancient interpolation approaches, I only notice about 10% of the time or less (nonobjectionably)
Brute framerate-based motion blur reduction means 5x framerate = 80% less motion blur on OLED. Framegen isn't grandpa's artifacty laggy Sony Motionflow anymore, and doesn't look as bad as AI slop if you're using NVIDIA's latest ($$$$ argh argh) = DLSS 4.5. I don't like the artifacts of Lossless Scaling, but it's another brute framerate-based blurbusting option too.
Be noted latency is problematic with framegen if you can't keep base framerates above 60fps (e.g. 60->240 or 60->360). If you're stuck with a laptop-based GPU, you maybe tolerate as low as 40-45fps base framerate if you have to "make do" with a solo game like CP2077, but it's worthless for esports at this time until lagless approaches arrive.
Colors are delightfully saturated on OLED, but OLED have poor "built in hardware BFI", so brute framerate-based blur busting is the only way to make OLED outperform plasmas on some render-heavy games like CP2077 that rarely hits 200-500fps without the help of good framegen.
Whether framegen becomes weaksauce AI slop, or useless laghell, or beautiful blur busting:
- Depends on the game (e.g. CP2077 vs esports)
- Depends on display (low-Hz LCD versus high-Hz OLED). LCD GtG means blur busting upgrades to framerates is a bit weaker ROI.
- Framegen algorithm quality (rudimentary blackbox-in-middle vs highend gpu/driver native with direct access to zbuffer and motionvector data)
Now that said, playing CP2077 on the Pulsar is delightful but framerates fall low enough often that you really need to try to keep it in the triple digits to make Pulsar work well during CP2077.
Right blur busting tool for the right job!