With PlayStation Spectral Super Resolution (PSSR), Sony has joined the machine learning-based image reconstruction race started by Nvidia almost six years ago. At Digital Foundry, we're excited to see this type of technology come to the console space, and so as part of our recent exclusive PS5 Pro capture opportunity, we spent some time using the PS5 Pro and Ratchet and Clank: Rift Apart to get a series. of reconstruction-based shots that we know from previous experience really puts reconstruction technologies to the test. To what extent can PSSR surpass the FSR-level technology that consoles typically use? And how close is PSSR to the recognized state of the art: Nvidia DLSS?
First, let's discuss the caveats and limitations. We compared the PS5 Pro to the PC at rough quality settings, but it's impossible to get a complete match. Secondly, Rift Apart uses dynamic resolution scaling and the PC implementation is very, very different, so we count the pixels of each individual test shot on the Pro and then precisely match the resolution in DLSS using a tool called DLSSTweaks, while we were limited to FSR 3.1. Quality mode, which typically uses a slightly lower base resolution than the Pro. Beyond that, while we tried turning off motion blur on PS5 Pro for clarity, turning it off only seems to reduce the intensity of the effect, so which we had to match it in our PC captures.
Finally, and perhaps the biggest caveat of all, this is just one game and the quality of improved implementations varies from title to title. We haven't seen enough of PSSR to draw stronger conclusions like we can with FSR and DLSS, but there's a good chance that what we're seeing here with PSSR could be indicative. For now, though, it's mostly a face-to-face between Ratchet and Clank.
I highly recommend watching the video above as the very nature of the exercise means that video is a better means of communicating the differences and, as you'll see when you watch it, comparisons to FSR 3.1 were the first order of business. AMD's compute-based scaler is widely used in the console space, and in many scenarios this is the technology that PSSR seeks to supplant. I started by searching for 'disocclusion failure'. A third-person game like Rift Apart is great for testing this: it's essentially the quality of newly revealed details that have little to no “story” to take advantage of previous frames.
This is my main complaint with the quality of FSR 3.1: it fails to effectively smooth out newly revealed details, “failing” in motion. Fortunately, like DLSS and XeSS, PSSR does not suffer from this problem, and it is as obvious as day. Ratchet and Clank runs at relatively high resolutions on PS5 Pro, so this gap in quality may widen when it comes to much lower quality input pixel ranges, something we'll test in due course.
FSR also has problems with particles, whether it's the particles when Ratchet picks up his first weapon or the confetti that flutters in the background during the opening stage. The particles have a faint, ghostly appearance when processed with FSR, to the point where confetti can appear and disappear. PSSR does a comparatively superior job and actually appears to feature more confetti. That's not actually the case: it's just that the PSSR particles are more temporally consistent and remain on the screen in stark contrast to FSR 3.1. Hologram posters also look better with PSSR. FSR 3.1 lacks anti-aliasing coverage here, and while PSSR isn't perfect, the presentation is less jagged around the edges.
To view this content, please enable targeting cookies.
Finally, processing grass in the first stage is challenging for FSR, to the point where it can apparently present fluid artifacts that appear almost liquid. Machine learning-based scalers tend to discriminate better and PSSR attempts to fix that problem. It's still not perfect, but I'd say it manages to keep the grass from looking as weird as it does with FSR. Considering this is the first publicly seen version of PSSR, we can safely assume that it manages to do the job required of it: it offers a substantial improvement to improve quality, but keep in mind that we're still dealing with relatively high base resolutions. It's at 1080p, and even lower, where PSSR really needs to prove itself.
Comparisons to Nvidia DLSS are perhaps of less practical use. Unlike FSR 3.1, you will never see a Sony console using this improved technology. More comprehensively, however, DLSS holds the crown as the most successful machine learning-based scaler on the market, a position Nvidia has earned with almost six years of iteration, so we can get an idea of how close it has Sony state with its debut. iteration. In all of my test clips, I noticed a sort of sizzling sound in motion. I'm not exactly sure what it is, maybe some kind of meta-stability, but it's in all the footage of the game I've seen so far. To be fair, it's also not present in FSR 3.1, and I also found that in some areas, DLSS resolves moving geometry with less aliasing.
This is also confirmed when comparing a rapidly moving object. PSSR does a better job than FSR on the move, but against DLSS it doesn't smooth out aliasing either. Another thing you may notice about PSSR is how it produces a smoother image than DLSS. Nvidia's scaler no longer has an inherent post-process sharpener built into the model, instead games only add it if they want to. And even with post-process refinement, I think we can see how the resolution of the PSSR is noticeably smoother. This can be seen as a detriment, but I found it much preferable to cases where the Ratchet game's post-process sharpener is cranked up to maximum. Of course, the resolution of a reconstruction technique is essentially subjective and a matter of personal taste.
Less subjective is something curious I noticed: the ghost ray tracing on the PS5 Pro version of Ratchet and Clank, seen in its RT reflections. Combining ray tracing with image reconstruction is complex and the pattern noise introduced in the reconstruction can have a large impact on the output image. Insomniac says the sampling pattern is customized to best suit PSSR. Although reflections in movement have negative aspects, there can be greater stability in certain aspects. For example, at lower quality settings, checkerboard RT reflections clearly show checkerboard artifacts with DLSS. PSSR merges the checkerboard correctly, offering higher resolution output.
Beyond these differences, we'll need more time with PSSR on this and other titles to render a more definitive verdict. Based on current testing, there are areas where PSSR is far superior to FSR 3.1, the technology it is most likely to replace, and, on the face of it, is easily preferable. Assuming these differences translate to other games, PS5 Pro owners may notice significantly improved upscaling quality. The issues with stability are something to keep an eye on in the future, to see if they show up in other games, and if not, if they are improved in future iterations of PSSR.
But perhaps it's games like Alan Wake 2 that best prove their worth: an internal resolution of 864p in its 60fps performance mode, upscaled to full 4K is a daunting task for the technology, and we're looking forward to testing it, along with other titles. That may have taken FSR technology too far on the standard PS5 and Xbox consoles. The PS5 Pro release date of November 7 is approaching and we'll have more coverage as soon as we can.