In an effort to help PC VR developers bring their content to Quest 2, Meta has ported Showdown, an old UE4 VR graphics showcase, to the headset as a case study in optimization best practices.

Showdown is a UE4 PC VR demo originally made by Epic Games back in 2014 to show off high-fidelity VR graphics running at 90Hz on a GTX 980 GPU at a 1,080 × 1,200 (1.3MP) per-eye resolution.

Eight years later, you can now run Showdown on Quest 2 at 90Hz on the headset’s Snapdragon XR2 chip at 1,832 × 1,920 (3.5MP) per-eye resolution.

Meta ported the short demo as a case study in optimizing PC VR content to run on Quest 2.

And while the app has been heavily optimized and doesn’t look as good as its PC VR counterpart—decent anti-aliasing, lighting, and high-res textures are missing—it shows that developers don’t have to shy away from lots of objects, particles, and effects just because they’re targeting Quest 2.

The video above looks slightly worse than the experience in the headset itself due to a low-ish bitrate recording and the visibility of fixed foveated rendering (lower resolution in the corners of the image), which is significantly less visible in the headset itself due to blurring of the lens. Here’s Showdown running on PC if you’d like to see a comparison.

It’s not the best-looking thing we’ve seen on Quest 2, but it’s a good reminder that Quest 2’s low-power mobile chip can achieve something akin to PS2 graphics at 90Hz.

Meta’s Zac Drake published a two-part breakdown of the process of profiling the app’s performance with the company’s App Spacewarp tech, and the process of optimizing the app to run at 90Hz on Quest 2.

The GTX 980 GPU (which Showdown originally targeted on PC) is at least six times more powerful than the GPU in Quest 2… so there was a lot of work to do.

While the guide is specific to projects built with UE4, the overall process, as surmised by Drake, applies to optimizing any project to run on the headset:

  1. Get the project building and running on Quest
  2. Disable performance intensive features
  3. Measure baseline performance
  4. Optimize the stripped down project
  5. Optimize individual features as we re-enable them
  6. Re-enable feature
  7. Measure performance impact
  8. Optimize as needed

Although it’s plenty possible to get ambitious PC VR games running on Quest 2, building from the ground up with the headset in mind from the outset is sure to bring better results, as developer Vertical Robot is hoping to prove with its upcoming Red Matter 2.

This article may contain affiliate links. If you click an affiliate link and buy a product we may receive a small commission which helps support the publication. See here for more information.


  • Adrian Meredith

    That doesn’t look very good to my eyes based on that video. The story of poor quality you’d expect from a downgrade port. Red matter 2 will be way more impressive I bet.

    • It’s not imo. RE4 is a much better and far more powerful showcase of how good a game can look, sound and indeed play on Quest 2.

      • Pootasso pick

        i love re:4 vr. not the best showcase of graphics on quest 2 tho. it IS a 2004 game.

        • I haven’t seen a better looking and more polished fps-type game on Quest 2 personally. I know some games seem like they have more shininess and stuff, but I don’t think there’s a game I’ve played or seen that just looks better all round. And, for the most part, it’s mostly down to the quality of those 2004 textures above all else, which, in many ways, are far more suited to Quest 2’s limitations than almost every other modern VR game tries to do on the using using shaders and the like that it just doesn’t show off very well. Something like Vader Immortal for example, simply does not look better imo, unless people aren’t really paying attention to what they are actually seeing.

          • Christian Schildwaechter

            The Quest 2 may not be a computational powerhouse, but it can do better than 2000 PS2 or 2001 GameCube graphics. The RE4 port to Quest isn’t based on the original 2004 GameCube release, but on the 2011 HD remaster for Xbox 360 and PS3, which is why the textures are not just a pixilated mess in VR.

            Whether it is the best demonstration of Quest 2 graphics is up for debate. I probably would pick another example simple because RE4 pretty much started the “brown phase” of video games, where for several years into 7th gen (way too) many titles basically rendered everything through a mud filter to make it more real and gritty and distance themselves from “childish” cartoon graphics in popping colors.

            This doesn’t reflect on the quality of the game or the graphics, but it kind of screams “early 2000s”, which is not the first impression you want to generate when trying to demonstrate that the Quest 2 is capable of up-to-date graphics. RE4 was always a great game, and the Quest 2 port is quite an achievement, but I’m pretty sure it is the only Quest game where even the sky looks like mud.

          • Yeah, I agree with the whole “brown phase” of so many games from that era, and I hated that too. Although, strangely, when playing through RE4 VR, it never bothered me, even though it did with the original and a whole lot of games that fell foul of the same issue. It’s similar to the whole metallic/plastic look that sooo many games suffered following on from the original release of Gears of War.

          • Pootasso pick

            yeah.

  • RE4 VR already showed me that 90% of Quest 2 developers simply aren’t taking full advantage of the system. Rather, they’re porting PC VR games and just turning off all the shaders and stuff, which makes them look sh*t for the most part. When a developer actually develops properly with the Quest 2 in mind in the first place, and actually tries to create visuals that play fully it its particular strengths and weaknesses, it’s capable of some great things. I want more games at the all round quality of RE4 VR, and if this demo can give some more developers a kick up the ass, then I’m all for it. Make your textures and models and lighting and so on actually for Quest 2!

    • Jonathan Winters III

      Agreed. When porting, the first thing to go is usually real time lighting and shadows, which severely downgrades the visuals. Many graphically impressive Quest 2 games have shadows and real-time lighting (Battle Sister, Ven, etc), unfortunately many devs just don’t know how to optimize well for Quest 2 or they don’t have the budget to give time to optimize properly.

  • Roger Bentley

    looks like trash blade and sorcery and RE4 look way better

  • xyzs

    Damn, that should just be enough to teach the world and especially Meta that it was way too early to kill PC VR………..

    • Mr.Philgood

      Meta didn’t kill PC VR, the market did. Meta released the Rift S and Quest 1 at the same time at the same price point. Quest 1 was a lot more successful, that’s why they discontinued PC VR.

      • Andrew Jakobs

        To be honest, Rift S was an outsourced headet with minor input from Oculus, they were already putting all their own efforts into the Quest.

      • Scott Herdliska

        What’s funny is that the Rift-s is a better experience than the Quest 1.

      • The Rift S was way better than the Q1 for pcvr lol

    • Arek A

      Agree. just bought my rift back.

  • PsYcHo

    this suposed to look good?looks way worse than robo recall from 2016 and that was an actual game wtf the buildings look hideous,the textures are comparable to 2003 vice city and soo much aliasing,better stick to sharp but old games like re4 they are more pleasant on the eyes.

  • Thud

    I can’t believe that’s it’s been 8 years since this was released! holy s***.

  • Bob Jenkins

    it still looks awful lmao

  • Cless

    Good for a Quest 2 I guess…? It looks on par with 2016 PCVR games though.
    I really hope there are no more headsets build with the XR2, its ancient at this point and this video is clear proof of it.

    • namekuseijin

      it’s not like a mobile chip of today, 2 years after XR2, will magically handle Crysis in VR…

      there are new mobile chips every year, but their increments are small, 15% or so. you don’t go from PS2 to PS3 with that… and indeed it took consoles 5 years for that leap…

      • Cless

        Yeah, never said that. Just pointing out that if any headset comes out with the XR2, it would be laughable at this point. The problem is, the rumors are pointing at that currently…

        • namekuseijin

          “yeah, never said that, only implied…”

          • Cless

            Didn’t imply anything either, I meant exactly what I said on my second answer. Stop putting words in my mouth. Wouldn’t you be upset if you bought a new phone just to discover that it has an old CPU from years ago? I would, specially since this things aren’t that cheap.

        • MeowMix

          The problem is, the rumors are pointing at that currently…

          The problem is, if you’re referring to Project Cambria, well gaming isn’t the focus. It’s said it’ll play Quest games, but the focus is productivity. Evren if Cambria had a newer SoC, the games won’t target the specs of Cambria, and will continue to target the Quest2 specs.

          • Cless

            I see! Then that is good. Hopefully we don’t see any gaming headset with it!

        • Andrew Jakobs

          Except if they were using faster chips, the headsets would also be much more expensive. Also at this time, there aren’t many mobile chips that have much better GPU’s (and being able to sustain that for a couple of hours on a battery).

          • Cless

            I mean… that is not how any of this works, is it? When a new chip comes out years later, its not more expensive, its usually the same price as the one its replacing, maybe slightly higher to account for inflation. They also tend to be more powerful and more efficient battery wise.

            If we start seeing $200 XR2 headsets, I won’t complain about those, I will complain about $500-1000 headsets having it though, no matter how “overclocked” it is.

            Sad thing is I just checked when the XR3 is suspected to come out… and apparently rumors say that is late 2023, that is… quite late. By then we will be having Nvidia 4000 series refresh about to come out on PCs…

          • Andrew Jakobs

            But there aren’t any new mobile chips at the moment. And you’re wrong about pricing, especially when bought in bulk for production, the older the chip gets, the cheaper it becomes, especially when a new release on the horizon. Also considering it takes a while to design a new headset around a new chip. Personally I still think the Quest is just for most of the design purely based on the reference model by Qualcomm, that also explains why the Pico Neo3 is so close in specs to the Quest.
            And ofcourse PC GPU’s will always be much better, but at this time the GPU’s are just way WAAAAAAY to expensive. I’m still even waiting for a simple RTX3060 to come down to MSRP prices as the hell that I’m gonna pay so much money for not even a midrange card. I thought back in november 2019 the 360 euro’s I paid for the RTX2060Super (which was really the cheapest available back then) was too expensive, as in the past the xx60 range was more like 200-260 euro range. Around me I see people alsow waiting on prices to drop. Maybe the cheapest solution is just to buy a complete new system as they seem to be cheaper as a new card alone (especially if your CPU is almost 10 years old, even though if it was back then the top of the line and still would be able to beat most of the newer semi-midrange CPU’s)..

    • Andrew Jakobs

      The Pico Neo3 is running the XR2 at a little higher speed.

      • Cless

        Damn, good enough I guess, but its not like there are any other alternatives either so…

  • Corellianrogue

    “but it’s a good reminder that Quest 2’s low-power mobile chip can achieve something akin to PS2 graphics at 90Hz.” It’s FAR beyond PS2. Especially when you consider it’s not just 90fps but also almost 2K per eye. (Edit: It’s actually over full 2K per eye in the total amount of pixels.) Even the PS3 could barely manage 1080p in a handful of games. I can’t remember if there are any full 1080p (as in 1920×1090) 60fps games on the PS3. All I can remember is Virtua Tennis 3 on the Xbox 360 being full 1080p 60fps. (I can’t remember whether Virtua Tennis 4 was as well.) Remember, the GameCube is more powerful than the PS2 and it could only run Resident Evil 4 at 30fps in 480p. Compare that to Resident Evil 4 VR.

    • namekuseijin

      yes, resolution and framerate (and 3D evidently) are way better than even PS3 could provide. But overall the graphics (overall polygonal detail and number of detail) is still more circa PS2 than PS3 – I’d say close objects like guns are very detailed like on PS3, but scenery is more like PS2…

      it’s undeniably a throwback, much like the Wii was. But the good news is that mobile chips evolve too. PS3 graphics were the point where diminishing returns started kicking in, when you can’t quite make out huge improvements instantly – that’s the era of Crysis, Uncharted, Gears of War, Assassin’s Creed. I bet Quest 3 will make most happy enough not to care about the distinction anymore. desktop chips seem to be economically on the way out…

      • Corellianrogue

        Many games look closer to PS2 just because they’re made by small indie developers, but the more graphically impressive games look a lot better than PS2 games. A good example is actually just a modded game, Doom 3 on the Quest 2. (Or even Quest 1.) The PS2 simply couldn’t do that even in 480p at 30fps. The original Xbox only just managed it and that was 2 or 3 (or maybe more) times more powerful than a PS2. The best graphical quality on Quest 2 games will probably be between an Xbox and Xbox 360 quality.

    • Octogod

      Quest performance is sub PS2 when it comes to triangles and transparency. But yes, looking at FPS and resolution, it’s much higher.

      • Corellianrogue

        How does the PC version of Doom 3 run on even the Oculus Quest 1 with just a mod then? That game just isn’t possible on the PS2. If it were then it probably would have been released on it. What PS2 games look as good as Doom 3?

  • PcgamerModsAreSnowflakes

    Meta VR proves that you just can’t polish a turd.

    • XRC

      Unfortunately mythbusters proved, you can in fact polish a turd…

  • Sofian

    What makes you think it’s rendered at native resolution?

  • Of course Meta ported it, not Epic. Going to die without ever playing Robo Recall 2 at this point…

  • Tommy

    Yeah, still not impressed.

  • James Cobalt

    The demo doesn’t look as good as Vader Immortal or Red Matter, both of which also run on UE4. If all you wanted to show was the power of highly optimized UE4 games on Quest hardware, you can’t provide a better example than these full length experiences. But if you want to demonstrate (with source code, so you can follow along) how to actually go about converting an existing project, this is pretty great.

  • wowgivemeabreak

    Q2 has better graphics than PS2. I think the PS2 was so long ago that people have forgot what games on it look like. A game looking like this on PS2 would have blown people away at the time.

    As for this demo, the lack of better lighting really hurts it compared to the original. It looks decent but too dark. Perhaps it isn’t as dark in the headset as the video shows.

    • benz145

      I’m comparing it to the best looking PS2 games (ie: best graphics at the end of its life). I think it’s a pretty fair comparison if you’re looking at things like polygon counts, texture resolution, and lighting. Resolution and framerate are a different story surely.

  • d0x360

    It looks ok but they hide allot of issues with chromatic aberration. It’s also clearly low resolution made obvious by the jaggies.

    It’s also easy to render something pretty when it’s just a tiny slice. It’s no different than the robot street scene they made for the Rift. I STILL haven’t seen a PC game that looks as good as that… Well this but at lower res with some effects and physics removed.

    So don’t let this trick you into thinking there is some amazing next level that we will get on current hardware. Not without something like DLSS which isn’t ideal for VR due to latency caused by needing multiple frames of data for the temporal upscaling.

    • benz145

      AppSW is definitely a big boost to what can be done on Quest 2 (along the lines of the gains DLSS offers), and that’s partly what’s being demonstrated in this case study.

    • Christian Schildwaechter

      The collection of previous frames itself isn’t a problem for VR, due to the camera always being controlled by the user it is actually less of a problem. In pancake games any type of temporal anti-aliasing/upscaling will not work for a few frames whenever the camera perspective is switched, e.g. after a cut-scene, causing a short flickering until enough past frames/pixels have been collected. This rarely happens in VR, as taking camera control from the users is very disorientating, so you can be pretty sure to already have access to previous frames from almost the same perspective at all time.

      There is some latency added for reconstructing the image like with any type of post-processing, and other problems inherent to TAA, but these are unrelated to collecting multiple frames. It just means that temporal upscaling doesn’t work for everything, you have to balance the benefits of a shorter rendering time against the added time for upscaling and the loss in visual quality. Digital Foundry just compared DLSS 2.3 against FSR 2 (which in theory could work on the Quest 2), and got mostly around 1-2ms added frame time for resolutions similar to that of the Quest 2, but of course using PC grade GPUs. As long as you can fit this into your 11ms@90Hz, or 22ms with AppSW, the user will not notice any extra latency, so temporal upscaling could be very useful for VR.

  • Octogod

    This is probably the worst demo they’ve ever released. It does the opposite of what Meta intended, showing that even engineers who made the hardware can’t make this look better than 2010 iPhone games.

  • pablito

    Here is the Cambria power formula:

    Understand, the XR2 chip has been running underclocked in the Quest 2 all ths time. There’s a 18% performance gain as a result of Cambria’s better headset design.

    Now then…

    Cambria max performance=(XR2 clockspeed + Applicatione Space Warp70%boost)*Dynamic Foviated Rendering with eye tracking at a 350% boost…

    Alien Isolation will run natvely with all effects on the Cambria. A few RAM issues may persist, but yeah… we,re there.

    • Andrew Jakobs

      There is no native Alien Isolation… sadly.. Shame they never finished the VR part. Still think it’s only a few days or a couple of weeks work to finetune the VR part, and then release it as a separate paid DLC..

    • Cless

      So… where are you claiming that 350% boost with Foveated Rendering? Because I’m calling bullshit to that IMMEDIATELY. No matter how early in the rendering pipeline you implement it, you are not getting 350% boost without losing serious quality (unless your lenses are absolute trash and hide it)

      • pablito

        Actually Meta’s own internal memo’s released/leaked: Sadley it’s Bradley was one of the first ones to comment on it. Upload VR had a couple articles stating a 350% boost. The 350% value is considered to be on the conservative side. This issue with performance boost has been out there for over two months now. Various You Tube channels are covering it in depth. Trust me, it’s very real.

        • Christian Schildwaechter

          Pretty sure you are mixing things up here. We recently heard about significant performance improvement from dynamic foveated rendering with eye tracking, but it wasn’t from internal Meta memos, it was from a closed door presentation by Unity showing eye tracking on the PSVR2. Using a very demanding scene, they got 360% the frame time performance with dynamic foveated rendering compared to baseline.

          First, this means a 260% boost over the 100% baseline, not a 360% boost. Second, they also talked about fixed foveated rendering, which gave them 250% of the baseline, or 150% boost. So in reality the actual performance increase by eye tracking compared to just statically rendering the outer areas at lower resolution, just like on Quest 2, was 360/250 = 1.44, or a 44% boost. Your assumed 350% boost factor from eye tracking is about eight times of what they actually got. That 360% number looked impressive, which is why everybody jumped on it, but it includes the 100% baseline and is compared to a non optimized scenario, which wouldn’t be realistic in performance bound game.

          There is also no way to know if this would translate to the Quest. The test scene used in the presentation was VR Alchemy Lab with lots of dynamic lights, making it very compute intensive. With increasing the FoV, the rendering cost grows exponentially, as you have to render a lot of geometry that covers only a very small part of the image. That is why they were able to get 250% baseline performance out of fixed foveated rendering, which removes a lot of less visible details, while the same on Quest 2 only gives you about 125% baseline performance (according to Meta). Due to the much weaker GPU, some of the compute intensive options aren’t ever used in Quest games, meaning that there is actually less room for optimization.

          Other performance tests by Tobii (used in PSVR 2) have shown much more moderate performance gains from eye tracking plus dynamic foveated rendering, around 40%, which isn’t a gigantic improvement over 25% from fixed foveated rendering. Application Space Warp provides “up to 70% better” frame rates, but this is for a best case scenario. And only the CPU part of the XR2 is underclocked on Quest 2, so that the GPU can be given more power and thus run at full speed. Cambria could slightly overclock the XR2 with better cooling, but it also has to drive 30% more pixels than the Quest 2. And with the extra compute requirement from eye and face tracking, we are most likely looking at a device that is able to run Quest 2 games with 30% more pixels at about the same performance as the Quest 2 (which would benefit from AppSW in the same way), but not significantly faster. So you should adjust your estimate way down.

  • pablito
  • Scott Herdliska

    Is this the “win” they think it is?

  • I’ve read their posts and while nice, they are not teaching anything special for optimization