Putting together the Digital Foundry platform comparisons aren't easy. Pristine, lossless captures are acquired quickly filling multi-terabyte RAID arrays, clips are meticulously lined up, shots are extracted, magnified and compared. And then of course, performance is tested. But what if we could carry out most of the Face-Off process in one pass? What if we could play two versions of the same game simultaneously while grabbing all the capture data we need in the background? Well, that's what we're doing here in this video presentation, where John Linneman and I play Knack on PS4 and PS4 Pro using a single Dual Shock 4 controller, with two screens side-by-side.
Dual system control (we tend to call it 'dual-wield') was a hardware experiment I put together back in September 2014, in an attempt to streamline workflow and to make our comparison videos - then dominated by cut-scenes and very limited gameplay - a lot more dynamic. In fact, doing a quick search though our YouTube history, I found this never-published Battlefield 4 comparison which showed highly promising results during our first tests. But the truth is that while dual-wield produces some cool comparison clips, the combination of varying game logic and shifting analogue stick data means that de-sync happens very quickly, limiting its application.
But Knack? Well, that's different. Camera movement is carried out by the game, not the player, removing our most troublesome de-sync variable - the drift caused by the processing, averaging and smoothing of controller input. Xbox One and PS4 actually have different controller polling intervals too (4ms and 5ms respectively), and this also contributes to de-sync - but this is not a problem with Knack as we're comparing two PlayStations with the same sampling. Game logic still causes problems, but we can quickly re-sync the action by moving Knack to a corner of a room. In almost every way, we have a best case scenario here. A typical dual-wield session on most titles lasts a few minutes before we need to abort (some see immediate drift) but in the case of Knack, we could probably play through most of the game like this.
But could we actually do a live Face-Off? The answer is yes - and no. The video here has only one edit - where we swapped out a PS4 Pro for standard hardware - and we did manage to pick out the primary difference between Pro's 4K and 1080p modes. It seems that dynamic shadows are limited to just one light source in the 4K output mode (the sun) whereas the high performance option - which attempts to run Knack at 1080p60 - retains the same visual feature set as the base PlayStation 4 game.
But getting to the nuts and bolts of Knack's PS4 Pro features required additional work after filming was complete. During play we speculated that the 4K output mode wasn't native, and our best guess at the time was 1800p. Running in the captures and carrying out the pixel count, we settled on a native resolution of 3072x1728 (80 per cent of 4K on both axes). A strange shimmering we picked up on viewing the game at close-range on a 58-inch Panasonic 4K screen was confirmed looking at the captures, while other scenes exhibited a kind of vertical interlace artefact, suggesting a checkerboard upscale of some description.
Performance analysis remains very much an offline exercise of course. Here we could confirm that Knack's high performance mode mostly runs between 50-60fps, while the 4K output mode - by and large - matches the frame-rates posted by the game running on standard PlayStation 4 hardware. With the data in the bag, final resolution metrics and performance analysis was added to the 'live' video during the editing phase. And it's here that some other anomalies were noted - such as deactivated lighting effects in the 4K mode's background features (highlighted up-close fleetingly during an engine-driven cut-scene).
John and I had plenty of fun filming this, as it realised - fleetingly at least - our initial hopes a couple of years ago for producing better, faster comparison work. It's evolved into a nice tool for the DF toolbox though, and we've used it recently for great shots in Uncharted 4, Rise of the Tomb Raider, Hitman and to spectacular effect in FIFA 17. We even used it to play FIFA 17 and PES 17 simultaneously, which worked rather more successfully than we thought it would.
But as things stand, we'll have to continue producing our work in the established manner - until Knack 2 comes out, at least.