Last week, Ubisoft released its latest Star Wars open-world game, Star Wars: Outlaws. Star Wars: Outlaws is powered by the Snowdrop engine, so it’s time now to benchmark it and examine its performance on the PC.
For our benchmarks, we used an AMD Ryzen 9 7950X3D, 32GB of DDR5 at 6000Mhz, AMD’s Radeon RX580, RX Vega 64, RX 6900XT, RX 7900XTX, NVIDIA’s GTX980Ti, RTX 2080Ti, RTX 3080 and RTX 4090. We also used Windows 10 64-bit, the GeForce 560.94, and the Radeon Adrenalin Edition 24.8.1 drivers. Moreover, we’ve disabled the second CCD on our 7950X3D.
Ubisoft has included a lot of graphics settings to tweak. PC gamers can adjust the quality of Microdetail, Shadows, Particles, Terrain Tessellation and more. There are also options for Film Grain, DoF, Motion Blur and Chromatic Aberration.
Do note that the game has Ray Tracing, even on its lowest settings. In other words, you can’t completely disable RT. So, what are the RT effects that the game uses by default? RTGI, RT reflections, and RT shadows.
Not only that but the PC version comes with support for RTXDI. RTXDI is NOT RTGI. Instead, RTXDI replaces all of the game’s direct lighting. This will allow almost all light sources to cast shadows. Plus, the RTXDI shadows will look better and more realistic than the shadowmaps that the game used in its non-RT version. DF did a great job at explaining RTXDI, so make sure to watch their video.
The PC-only features do not stop here. Ubisoft has also added support for DLSS 3.5 Frame Generation and Ray Reconstruction. Oh, and there is also support for AMD FSR 3.0 FG, meaning everyone can use Frame Generation to hit higher framerates. And believe me. You’ll need it in Ultra settings.
Massive Entertainment has not included a built-in benchmark tool. This comes as a surprise as Avatar, which was powered by the same engine, did have one. Since there isn’t a built-in benchmark, for our tests, we used the first city you encounter in the game. This appears to be a demanding area.
From what we could see, Star Wars: Outlaws is mostly a GPU-bound title. Our NVIDIA RTX 4090 was used to its fullest, even at 1080p. So, there wasn’t really a point simulating different CPU configurations.
At Native 1080p/Ultra Settings/No RTXDI, our top four GPUs were able to push framerates higher than 60fps at all times. The RTX2080Ti was also able to provide a smooth gaming experience with a minimum of 58fps and an average of 64fps.
At Native 1440p/Ultra/No RTXDI, the only GPUs that were able to run Star Wars: Outlaws with more than 60fps were the AMD Radeon RX 7900XTX and the NVIDIA RTX 4090. The AMD Radeon RX 6900XT and NVIDIA RTX 3080 were also able to provide a playable experience, provided you use a FreeSync/GSync monitor.
Finally, at Native 4K/Ultra/No RTXDI, the only GPU that was able to provide a smooth gaming experience was the NVIDIA GeForce RTX 4090. Again, this is at Native 4K and the game has Ray Tracing effects on Ultra settings, even without using RTXDI.
And since we mentioned RTXDI, here are some comparison screenshots. The non-RTXDI screenshots are on the left, whereas the RTXDI screenshots are on the right. RTXDI can improve the game’s shadows. This can be very subtle at times. At other times, it can be very noticeable. In static screenshots, you may also think that the shadows in the last comparison are fine. In motion, though, the non-RTXDI version looks AWFUL as the grass shadows are full of aliasing.
The bad news here is that RTXDI comes with a HUGE performance hit in grassy areas. At 4K on Ultra settings with DLSS 3 Quality and Frame Generation, we witnessed drops to 50fps on the NVIDIA RTX 4090. That’s WITH Frame Generations. And, yes. That’s LOWER than Black Myth: Wukong with its Full Ray Tracing/Path Tracing effects. I don’t really know what’s going on here. However, you know there is something wrong when a non-path-traced game runs worse than a path-traced one.
DLSS 3.5 Ray Reconstruction also comes with a 10fps performance hit. This caught me off guard as I wasn’t expecting this big of a hit from it. There are visual benefits here, no doubt about that. But for a game that runs so poorly with RTXDI, it makes no sense to enable it even on the most powerful GPU. If, on the other hand, you don’t want to use RTXDI, you can enable DLSS 3.5 RR and enjoy its visual improvements.
It’s worth noting that Star Wars: Outlaws can scale down on older GPUs. By dropping to High, we were able to increase performance by 25% on the NVIDIA RTX4090 at Native 4K. Then, on Medium settings, we got an additional 12% performance boost.
Graphics-wise, Star Wars: Outlaws does not come close to what we’ve seen in the most recent Unreal Engine 5 games. The game also looks too soft right now, even at 4K. This appears to be related to the “Lens and Cinematic” setting. By setting this to Medium, the game will look sharper. However, it can cause additional aliasing. There are also numerous visual artifacts from the software-based RTGI. The animations of characters are also a bit rough, and nowhere close to what you’d expect from a triple-A title.
In all honesty, Black Myth: Wukong and Senua’s Saga: Hellblade 2 look like next-gen games compared to Star Wars: Outlaws. They also run better than Star Wars: Outlaws. So, this may give us an idea of how efficient UE5 can be when it comes to Ray Tracing. And although Horizon: Forbidden West does not have any RT effects, it can look better than the non-RTXDI version of Star Wars: Outlaws.
Now the good news here is that I did not experience any major stuttering issues. The game was also stable as I didn’t get any crashes. PC controls are also responsive. As such, the game plays great with a keyboard and a mouse.
All in all, Star Wars: Outlaws is a mixed bag. Performance is great without RTXDI, and the game can scale on older GPUs. However, there are various visual issues. For instance, shadows from the sun can look low-res and awful. RTXDI fixes this. The game also has blur, aliasing and ghosting issues, caused by the upscaling techniques, DLSS 3.5 Ray Reconstruction and the “Cinematic” setting. And when you disable them, you get other visual issues. For example, without DLSS 3.5 RR, you get A LOT of visual RT artifacts in shadowy places. And even with all its RT effects, Star Wars: Outlaws cannot match the visuals (and the performance) of Black Myth: Wukong or Hellblade 2. Or Alan Wake 2. So, let’s hope that the devs will bring some optimizations via some patches!
John is the founder and Editor in Chief at DSOGaming. He is a PC gaming fan and highly supports the modding and indie communities. Before creating DSOGaming, John worked on numerous gaming websites. While he is a die-hard PC gamer, his gaming roots can be found on consoles. John loved – and still does – the 16-bit consoles, and considers SNES to be one of the best consoles. Still, the PC platform won him over consoles. That was mainly due to 3DFX and its iconic dedicated 3D accelerator graphics card, Voodoo 2. John has also written a higher degree thesis on the “The Evolution of PC graphics cards.”
Contact: Email
The only way RTXDI made the game look more spectacular was in the first screenshot with the guy near the stairs. That Illumination was cool, every other screenshot is meh.
I think a lot of these effects are really about how we perceive them. Designers have spent years developing techniques to mimic the effects of real light so now that we have technologies that can just do direct lighting, a lot of the time it does feel like there is no change at the cost of a performance hit. We probably just don't see on the backend that some developer time/resources now get to be spent somewhere else instead of on creating a realistically lit scene. Now that the RT capable 3060 has become the most popular GPU on steam and the current gen of consoles are approaching their ends, we'll probably see way more RT implementations of a similar scale to this.
Should be called out for using the trash swgi amd friendly method rather than rtx gi which makes no sense whatsoever. They could still have that sw fallback.
Thanks for the work you did here John. Your results pretty much line up with what I got over my past week with the game. I played it on a 6700XT and just a few hours of the opening on a 3060 and i think my results line up with what your testing shows. I think the baked in RT shows that while it does incur a performance hit and doesnt look "that" much better than pre-baked lighting to some people, it does accomplish a lot and allows the devs to work on other aspects of their games as open worlds can have so many light sources. I can only imagine what the future holds for utilization of these RT suite.
Instead of anti-aliasing settings, they have "Upscaler" settings. I do not like the direction the game industry is going. This is a bad trend.
I've seen this alot nowadays. I was wondering how no one is pointing this out.
I keep commenting about it, but usually I get pounced on by NVIDIA/DLSS/TAA shills who want to bury dissenting opinions. As for media, something tells me NVIDIA is making sure none of them make negative comments about this trend.
True. Although DLSS itself is an anti-aliasing method, I have more fondness for DLAA.
I was under the impression that DLAA was the anti-aliasing from DLSS, just without the upscaling. Anyway, both are temporal, so I don't like them. I can't stand the ghosting and the excessive blurriness.
Granted I guess you could have been referring to Directionally Localized Anti-Aliasing instead of Deep Learning Anti-Aliasing (blasted NVIDIA).
It is, dlaa is the anti-aliasing method of dlss toolkit. Personally i like to employ another way – ie super-sampled (dsr dl 1.78 x) with data from dlss (quality).
Yeild's better results than dlaa in many titles and lightyears ahead of trash taa that's basically just a cheap smudge filter that washes away all fidelity.
It may be clearer, but you can get that with ReShade sharpening shaders at a lower performance cost (or even NVIDIA's sharpening in the NVIDIA Control Panel), and either way you still have the ghosting from the temporal frame blending that TAA and DLSS/DLAA do. I'd rather turn off all anti-aliasing, and use whatever ReShade anti-aliasing shader looks best in the game.
All of a sudden all I see is AI this and upscale that in the menus. I don’t generally use anti-aliasing but I’m like, “where did it go”. We have lost the fight to DRM and now we are about to lose the fight to Fake Frames. I’m 41 and my hobby is basically dead. This is the only vice I have to close the world out for a couple of hours. Man this is sad. We are currently at a stage where it’s better to rent the games via subscription service, rather than buy them. Basically spend 15 bucks for the month, play what you want and comeback another month when there is something to play. Because buying games right now is pure rubbish. All of a sudden the subscription model doesn’t sound so bad because none of these games have a soul anymore, no sense in investing. They did this on purpose.
Sadly I have to agree. I find myself mostly playing older games these days, and only being interested in new ones where I can disable the anti-aliasing (even if it's just via config file).
I just played KOTOR and KOTOR2 for the first time a few months ago. As an FPS player I always thought the combat system looked stupid, but they were 1000x better than the new trash that keeps getting released by AAA studios.
And most of these games perform terribly at native resolutions. So you are basically forced to play the game with upscaling enabled. I don't mind this, if the upscaling is done right and doesn't introduce visual artifacts.
Ghosting is an artifact caused by Temporal Anti-Aliasing and temporal upscalers. It's unavoidable with the current implementations of the technology.
There is a TAA at native resolution option. It looks awful for some reason, but it's there.
I hate TAA. Is there an alternative to the temporal BS? Or an "Off" option?
Edit: The game has mandatory ray tracing, so of course they wouldn't allow an alternative to temporal anti-aliasing/upscaling. Without the temporal blending the lighting would look really weird, because of the spotty way that ray traced lighting is rendered.
I thought you were just talking about upscaling. TAA has been the norm for like a decade now, when’s the last AAA game that had MSAA? Like early Xbox 360 era.
I don't measure time in Xbox versions.
Red Dead Redemption 2 has MSAA. Not sure when that came out, or when the Xbox 360 was a thing.
Counter Strike 2 is forward rendered with MSAA (or CMAA2). Forward rendering is better for competitive shooters because of the lower latency.
Half Life Alyx is forward rendered with MSAA. Most VR games are forward rendered with MSAA since deferred rendering isn't as good for VR due to the higher latency, and VR needs the lower latency of forward rendering, plus TAA isn't good for VR for the same reason as motion blur (any visual artifacts can cause dizziness, headaches, and nausea).
Bohemia Interactive isn't a AAA studio, but from what I understand they still make forward rendered games with MSAA, they just call it "Hardware AA" for some reason.
Both Unreal Engine 4 and 5 have a forward rendering mode (they call it "forward shading") with optional MSAA that game devs can use. It's for VR and mobile games, but you can use it for regular flatscreen games as well.
I think Godot (which Unity devs switched to in droves when Unity introduced their install fees) only supports forward rendering with MSAA, so any games made with it will probably have MSAA.
Dude I’ve been a PC only gamer for 30 years and the idea of not knowing when the 360 was around is patently ridiculous.
Anyway, that’s a short list of games. Also Red Dead looks awful with MSAA, the tree foliage doesn’t even appear right. The game is designed around TAA and needs it. For better or worse that’s what games were designed around for 10 years, and now they’re designed around DLSS/FSR. I hate it too, but it is what it is.
I think you're taking about DitherTemporalAA (dithered transparencies) that a lot of games use in hair and sometimes foliage these days, which requires TAA to blend so it looks like normal transparencies. Technically MSAA can blend the edges as well, and MSAA looks better in RDR2 than TAA does if you use a high enough setting (I have a 1440p monitor though, so that may effect how it looks).
Note that DitherTemporalAA is just a cheap trick to try to gain some extra performance, and isn't even remotely necessary. This is one of the many bad decisions that game devs make these days that make TAA appear necessary, when in reality it isn't even remotely necessary.
A game that came out 2 years ago called Metal: Hellsinger uses SMAA, and it looks perfectly fine.
For more on why TAA is bad:
https://url.gt500.org/TAA
Was hoping you'd tryout the hidden "Outlaw" Graphics setting as I'm curious what fps a 4090 could get and if it even looks any better.
how can other games compete with ue5 games anymore.nanite means unlimited poly count while other games still have serious poly problems lod problems and other stupid problems .wtf .ue5 is a generation ahead of any engine out there
It's quite nice they did a gayme were you play as non-human.
Shame it looks like a turd.
Is it $20 yet for the top edition or another 6 months. 😉
Thanks for providing at least some results on high. Setting Ultra settings on everything is for dummies as the minor image quality improvement is rarely worth the performance cost. Personally, I would be more interested in benchmarks on high across the board, with perhaps one token low, med, high, ultra comparison for the top end GPU to see how things scale.
Problem with high in this game is it sets RT to medium, including reflections which don't even look good on high. Makes for a very messy image on shiny floors and whatnot.