This is one gamer's view, not a lab test. The piece has no measured before-and-after frame rates, no resolution-specific numbers and no controlled image comparisons. Treat it as a starting checklist for your own testing, not a set of guaranteed gains.
Why a flagship GPU still needs settings discipline
NVIDIA launched the RTX 4090 in 2022 as a 4K gaming GPU. Its launch announcement said the card "consistently delivers over 100 frames per second at 4K-resolution gaming", though that claim was about the games of that time. It started at $1,599 and went on sale October 12, 2022.
Newer AAA games are heavier. Rozario can still max out most of them, but he doesn't want to sit near 60 FPS. That's a personal choice, and it's a sensible way to frame the exercise: pick an FPS target first, then spend your GPU budget only on effects you can actually see.
1. Ray tracing and path tracing
Rozario turns on path tracing in any game that supports it, but only for about 10 minutes to see what it looks like. Then he switches it off. His examples are the RT Overdrive mode in Cyberpunk 2077 and the Full Ray Tracing setting in Black Myth: Wukong. He says path tracing without DLSS upscaling and frame generation is "an absolute no-go" on his card.
Conventional ray-traced reflections and shadows get a more nuanced treatment:
- Judge them game by game, because the visual payoff varies with how each title implements them.
- If the difference is clearly visible, try Medium.
- Use DLSS upscaling to win back the lost frames.
He also avoids frame generation unless the game already runs at about 60 FPS without it. His reasoning is that it smooths motion but doesn't make the game feel more responsive.
NVIDIA's documentation supports the underlying distinction. It describes DLSS Super Resolution as reconstructing higher-resolution frames from a lower-resolution render. It describes Frame Generation as producing additional frames. Reflex is a separate technology that reduces system latency by synchronizing CPU and GPU work. The 60 FPS cutoff is Rozario's own rule of thumb, not an NVIDIA threshold.
2. Shadow quality: High instead of Ultra
Rozario says the jump from Low to High is obvious, and he wouldn't trade it away. Between High and Ultra, he says he can barely tell the difference unless he compares them side by side. He also finds that Ultra sometimes makes shadows look too sharp. Depending on the game, the top setting might raise shadow resolution, extend the distance at which detailed shadows appear, or soften edges.
Shadows are among the costlier settings, so this is a cheap place to save frames. The caveat is that every engine labels and splits shadow options differently. Battlefield 6, for example, separates Local Light & Shadow Quality from Sun Shadow Quality.
3. Volumetric lighting and fog: High or Medium
Volumetrics cover light shafts through trees, dust in a room, fog and clouds. Rozario wouldn't disable them or drop them to Low, because they carry so much atmosphere. He says dropping from Ultra to High, or to Medium in demanding games, gives a decent FPS boost without a noticeable loss. His test is whether god rays still look like god rays. He accepts slightly lower rendering resolution or a bit more noise at the edges.
For Battlefield 6 specifically, other guides lean much more aggressive than Rozario. A recent FPS-focused guide from fpstweak recommends Low and claims several percent more FPS in heavy fights. That is a different goal, because competitive players also want thinner smoke for visibility. It's a good reminder that the "right" setting depends on whether you value atmosphere or clear sightlines.
4. Grass and foliage density: High, not Ultra and not Low
Rozario likes dense vegetation in open-world games such as Red Dead Redemption 2 and Assassin's Creed Shadows, but he's happy to lose some of it for frames. He draws the line at High, since Low can make landscapes look barren. The control goes by different names depending on the game:
| Game | Setting name |
|---|---|
| Assassin's Creed Shadows | Scatter Density |
| Battlefield 6 | Undergrowth Quality |
Shacknews' Battlefield 6 settings explainer describes Undergrowth Quality as the setting for the quality and complexity of grass and other plant life on the map. Several community and media guides recommend Low for competitive play. Rozario's High is a deliberate step up from that, aimed at single-player-style immersion. Even in Battlefield, one guide (Karan Benchmarks) lists Undergrowth at High in its own optimized setup, so the answer varies with taste.
A practical way to apply this
- Set your FPS target and resolution first, for example about 100 FPS at 4K for single-player games.
- Start from a high preset, not from zero.
- Change one setting at a time and look at the image during real play, not only in a still comparison.
- Use a performance overlay to see whether you're limited by the GPU or the CPU. fpstweak notes that some options, such as mesh quality, lean more on the processor, so lowering GPU-heavy settings won't help in that case.
- Keep upscaling in mind as a separate lever from the four settings above.
- Only turn on frame generation if the base frame rate is already healthy.
The upgrade question
Rozario also argues that the 4090 is aging well. He says the RTX 5090 was "barely 30% faster at 4K" for about a 25% higher MSRP, so he skipped it. NVIDIA's own "up to 2x" claim for the 5090 relied on Blackwell plus DLSS 4, so it isn't a pure native-rendering comparison. Independent launch-era reviews reported smaller gains in ordinary 4K gaming, with TechSpot at roughly 30% and Tom's Hardware at 25% in its rasterization suite. Whether that justifies skipping an upgrade depends on your budget and display, and he says he'd consider dropping to 1440p before paying $2,000 or more for another card.
Bottom line
The advice amounts to this: spend your GPU budget where your eyes can see it. Shadows at High, volumetrics at High or Medium, and foliage at High are cheap compromises. Path tracing is better treated as a screenshot mode on a 4090 unless you're happy with upscaling and lower frame rates. These are one enthusiast's preferences, not benchmarks, so test them against your own games and your own tolerance for frame-rate dips.