The good news is that based on our time with the console version of Crimson Desert, the PC build delivers exactly the kind of experience I was expecting. Yes, it can be demanding, but a mainstream Zen 2 processor like the Ryzen 5 3600 paired with a modest GPU like the Nvidia GeForce RTX 4060 can deliver reasonable results at 1440p using DLSS 4.0 balanced mode. The key to good performance ultimately comes down to just one setting: lighting quality. Keep away from max and cinematic settings, dial in ultra, and you'll see an extraordinary boost to frame-rate.
From there, it's all about the fine tuning, which is where we add the final flourishes via Digital Foundry optimised settings. Thanks here to contributor Rayan who put in the hours testing every single graphics setting in the game, understanding what they actually do and then figuring out which offers the best balance between fidelity and performance. And to offer an alternative viewpoint, we have Pearl Abyss' own settings selection. Well, for the purposes of this piece, we're using the PS5's balanced mode, but we've already published PC equivalent settings for all of the graphics options available on the base and enhanced Sony consoles.
Practically speaking, for a game using ray tracing across the board, the overall experience with optimised settings is fine, thanks to a shader compilation burn on boot that caches the required PSOs. Even on the mainstream Ryzen 5 3600 CPU from 2019, frame-times are relatively smooth, with only one noticeable stutter spotted in the first hour of play. This game is predominantly GPU limited, and overall performance on this kit hangs mostly in a 50-60fps window, with a couple of exceptions.

The scripted fight scenes at the beginning of the game can be heavy on the GPU, reaching a nadir in the 40s, but generally frame-rates are higher and genuine drops under 60fps are rare in the first hour. However, you definitely need a reasonable level of CPU power to maintain the 60fps threshold, especially in crowded scenarios like the pitched battle on Bug Hill. In these heavy areas, the 50-60fps window drops back to maybe the late 30s upwards, and enraging city guards in Demeniss also hits the CPU hard - perhaps more so than Bug Hill. Put simply, the more entities in active play, the more CPU power you need.
Interestingly, frame generation is an option, putting the RTX 4060 into a mid-70s to 90fps window, even on Bug Hill. Without frame-gen, PC latency metrics sit in the mid-40s, which is actually fine, but Frame Generation incurs around 15ms to 20ms of extra latency. Overall, I've certainly dealt with worse on this PC, suggesting very graceful scaling with a more modern CPU and something with a bit more power than the relatively modest RTX 4060.
Before diving into optimised settings, a quick word on the graphics menu: it's not bad, offering the right amount of configurable settings without needing a game restart. However, it frustratingly lacks on-screen indications of what the settings actually do, and the preview image offers no way to compare changes, which is a definite annoyance. VRAM limitations? Actually, Crimson Desert is fine on an 8GB GPU - just keep away from the cinematic texture quality option and limit yourself to high, which basically looks the same. If your GPU has more than 8GB, you're weapons-free to use the higher options - not that you'll notice much difference.
In terms of optimised settings, the video up top shows you what every graphics option actually does, but curiously, ray tracing on/off does very little, offering only a tiny 1-2fps boost on our RTX 4070 Ti test system. Crimson Desert extensively uses radiance caches, screen-space techniques, and SDF fallbacks, with RT adding extra detail using a small number of rays per pixel. Disabling RT doesn't change the core lighting, but shifts more of the workload onto the fallbacks, meaning RT is best left enabled as it improves light bounce, shadowing, and reflections more obviously.
In stark contrast to RT, the Lighting Quality setting is easily the biggest impact on GPU resources, especially at the Max setting. Max is required to use ML-based ray reconstruction and ray regeneration denoisers, which absolutely hammers performance, impacting it by up to 30 percent. By dropping down a quality preset from max and avoiding ray reconstruction, you effectively double performance, illustrating how stark the difference is. We chose ultra though, which gains eight percent of performance vs cinematic and looks much the same.
We also recommend dropping volumetric fog quality down from cinematic to the low preset, while model quality offers us a nice performance boost moving from cinematic down to ultra, making that the sweet spot. Beyond that, there are very few tangible performance improvements to be had - a few percentage points here and there, but sometimes you'll find that quality reductions don't actually seem to deliver that much more performance.
|
Digital Foundry |
PlayStation 5 |
|
|
Model Quality |
Ultra |
Medium |
|
Texture Quality |
High (8GB GPUs) Ultra/Cinematic (8GB+ GPUs) |
High |
|
Shadow Quality |
High |
High |
|
Ray Tracing |
On |
On |
|
Lighting Quality |
Ultra |
Medium |
|
Reflection Quality |
Ultra |
High |
|
Advanced Weather Effect |
Off |
Off |
|
Water Quality |
High |
Medium |
|
Foliage Density |
High |
Medium |
|
Volumetric Fog Quality |
Low |
Medium |
|
Effect Quality |
Cinematic |
High |
|
Simulation Quality |
Cinematic |
High |
|
Post-Processing Effect Quality |
High |
Medium |
With optimised settings complete, I tested out the Ryzen 5 3600/RTX 4060 PC at 1440p with DLSS 4.0 balanced mode across four different configurations: max (without ray reconstruction tech), cinematic, DF optimised and the equivalents to PS5 balanced mode.
In one scene, I gained a 39 percent performance increase just by using cinematic mode instead of max. DF optimised settings gave us a further 11 percent boost to frame-rate, while PS5 balanced mode equivalents were three percent faster than our own selection - but I'm happy with the choices made in bumping up so many settings to significantly higher presets in most cases, in exchange for a small hit to performance.
Meanwhile, in a more demanding area, there's a 50 percent increase in performance simply by dropping the lighting quality down from max to cinematic, while DF optimised settings deliver a useful 14 percent boost to frame-rates over cinematic. Curiously, in that scene, the equivalent PS5 balanced mode settings delivered no improvement to performance at all over DF optimised.
In a game like this, I'm sure users will be finding plenty of bugs, but my time with the game was trouble-free. It's a shame there's not more scalability in the options, but if the game produces pleasing enough results at 1440p resolution on an RTX 4060 paired with what is now an ancient CPU, I'd say it's in pretty good shape. Beyond this look at the PC release, we'll be reporting back soon with more on the console versions of the game - there's certainly plenty to get through!





Comments 8
For some reason having both ultra, cinematic, and max is weird to me. It's just a naming thing, but it's odd.
@MittenFacedLass yeah it's pretty lame. I guess the one messing things up is cinematic, especially as it is used in the industry for things that are not related to IQ (like a 30fps cinematic mode). Low Medium High Ultra Ultra+ Max would be a little clearer
I find the title of the article a little click baity, I prefer the more explicit titles that we usually get on the site.
I’ve put 20+ hours into the game, but most of that has been trying to dial in the graphics settings. 9800x3D+4090+120hzLG C4 42” @4k… this game looks awful. I’ve tried a lot of stuff, watched HUB, BenchmarKing, and now DFs optimised videos, and poured over reddit threads. I’ve tried capping fps in the Nvidia app, forcing sync on, using smooth motion instead of FG, low latency forced on, low latency ultra, changing the install location from one SSD to another… the game loads and holds super low polygon assets (like a house) until you’re only meters away and then loads the high poly version. Ignoring the RT reflection sparklies that reminds me of Hogwarts Legacy at launch, DLSS upscaling below quality looks awful without RR enabled, and appears to do something weird where it oversharpens textures but then applies blur on top, so the worst of all possible worlds. Reminds me of Monster Hunter Wilds. All this is still potentially workable if the graphics settings behaved themselves… but in my experience (and I hope I’m in the minority), the game appears to reset settings between sessions, unless you first select one of the presets at the top. Further, even if it does appear to retain the outward graphic setting selection, it doesn’t actually apply the setting to the game. This may be why for some graphics settings, people aren’t seeing any visual change, either in fidelity or fps. It makes ‘tweaking’ settings an impossible task, if you can’t trust the game to apply the settings. I noted in one of the patch notes the devs saying they had fixed exactly this behaviour… but I’m on the latest version in Steam and it is still happening. So as much as I want to play the game and enjoy/lament its weird combination of game mechanics (horse crouching - lament, horse drifting - enjoy), I’m leaving the game alone until the devs or modders get the game into at minimum, a configurable state. I really enjoyed Dragon’s Dogma 2 running on Performance quality thanks to the modders fixing the DLSS/RT blur… I’m hoping something can be done about CD also. To anyone not having these graphics issues, I congratulate you on your good fortune.
@NandoCalrissian How so, if it is factually correct? My definition of clickbait is an obviously enticing headline that does not deliver.
The first image shows that permutations of lighting quality can double performance which is full delivery on the headline? And beyond that, it's mostly just scraps. The second bank of screenshots shows a best case scenario of just 15 percent of extra performance beyond that one settings change.
The usual PC settings normalization for consoles. Water doesn't change anything; even 1.5% is a gain, so we choose to remove it to normalize as usual.
DLSS 4.5 weighs more than 4, which remains comfortably better than PSS2, whose flaws are invisible, just as the first's were barely noticeable before. The usual improvements are made after the fact.
Scenarios for textures and lighting are carefully chosen to minimize the impact on the flagship consoles, and the circle can be closed, for now...
@Rich_Leadbetter Great question, I don't pretend to be an authority in this space, so what follows is entirely subjective. For me, clickbait is broadly about titles that give the impression of deliberately obscuring information I would need to decide whether the content is worth my time, regardless of whether the article itself delivers.
Allow me to share an example: "This common food can help you drop belly fat by 30% in 30 days" feels like clickbait, while "Asparagus can drop your belly fat in 30 days" does not. With the second title I have enough information to make my own call: either I already know about asparagus and its fat blasting properties, or I hate it so much I won't entertain reading an article about its benefits. Either way, that's my decision to make upfront.
Reading the internet in (the year of our lord) 2026 too often means clicking a link, quickly scanning through the article to find the information that was hinted at and obfuscated by the title, only to then decide whether I actually wanted to read it in the first place. I did exactly that with this DF article, which is what prompted me to comment.
DF content typically sets an exceptionally high bar for quality, honesty and clarity, which is precisely why a title that felt a little clickbaity stood out enough to warrant a comment.
Applying that logic here, a subjectively less clickbaity title might have been: "A single lighting quality setting can significantly boost Crimson Desert performance." I'm not particularly interested in the game, but that title would have given me enough to decide whether the technical breakdown of the lighting was worth my time.
Update after 30+ hours:
4090+9800X3D... I don't know if it was the game, or patience, or I got lucky in the order I set things...but:
DLSS 4.5 (L) set to Quality, FG on, RR off, Reflex on, vSync left to its own devices, all settings pushed to Cinematic/Max except vegetation because otherwise I can't find anything on the ground in a forest: ~100fps@4k, no weird shadows, still little normal map fizzle at times, lovely soft lighting, excellent clarity in motion. As good as DLAA if not better, as the base fps is up to around 57 and much steadier.
Low poly objects are still a thing, but I don't think I care anymore. With model quality set to Cinematic, enough of them are far away that I can tune them out.
Using Max Lighting plus DLSS 4.5 (L) set to Quality was the game changer for me (I think?). It's noticably 'better' than 4.0 and 4.5 with objects in motion, which the game has a lot of. I still have the blur at 100%, and depth of field enabled in the Accessibility section... I think that offsets the 4.5/L (over)sharpening at Quality, while retaining the clarity in motion.
If anyone else is struggling on PC, I suggest a restart whenever changing anything significant.
This game just became a bit of a stunner (for me).
Show Comments
Leave A Comment
Hold on there, you need to login to post a comment...