Insomniac: 60FPS no moreNovember 7th, 2009Insomniac Games is "probably" going to turn its back on its long-standing commitment to 60FPS gameplay for its Ratchet & Clank franchise.
According to the developer, there is no marketing premium, or review score value, in supporting the highest-possible refresh rate common to all HDTVs. "We want to give you guys, our fans and players, the best looking games you can buy on a console," says engine director Mike Acton, and 60FPS actually gets in the way of that.
To back up his arguments, Acton has produced interesting data based on a large number of game reviews, which indicates that while there is a clear link between graphics and final score, there is little to no evidence that frame-rate has as much influence. He also polled readers of the Insomniac website, and found that while 16 per cent of respondents were firmly in favour of 60FPS, most are not, with the majority favouring a solid frame-rate that doesn't interfere with the gameplay.
Curiously, Acton seeks to play down the already small minority who don't agree with his overall findings by pointing towards general sales figures, saying that the 16 per cent figure can't possibly be true. Also interesting, especially in an argument couched in terms of sell-through figures, is the lack of focus on the mega-selling franchises that target 60FPS gameplay: the Gran Turismos, the Forzas, the FIFAs and, of course, the Call of Duties. Combined sales of all those probably account for a pretty significant chunk of the marketplace, and in the case of FIFA and Modern Warfare 2, a big slice of this year's Christmas sales.
In terms of Insomniac's decision to back away from 60FPS gameplay, it is perhaps not surprising when you consider overall trends. Back in the era where arcade gameplay and 2D sprite-based action was the norm, 60FPS was a given whether you were playing the latest coin-ops or powering up your latest (NTSC) SNES or Mega Drive game; it was the accepted standard. But in the tumultuous move to 3D gaming during the mid-nineties PlayStation era, things changed irrevocably. The enormous leaps in processing power required to create 3D imagery meant that 60FPS throughput from the console was only possible in the minority of cases: undemanding sports titles, the occasional shooter, and fighting games like Tekken and Virtua Fighter.
Two console generations later and 30 frames per second is the norm. Games are more slowly paced and definitely laggier than the 2D generation, but the majority of the audience has become conditioned to them, and, crucially, gameplay styles have shifted to sit more in sync with the lower frame-rates.
While Acton describes Insomniac's latest Ratchet episode, Crack in Time, as a 60FPS title, this is a somewhat best-case description of the overall flow of the game. Looking at the raw stats after Digital Foundry analysis, the performance doesn't quite reach the levels Insomniac aspired to, and - amazingly, bearing in mind the wonderful quality of the graphics - the game is actually sub-HD, albeit with the highest-possible image quality we've seen using its chosen upscaling technique.
Let's have a quick peek at some pixel analysis first, to confirm the findings. Edge analysis can sometimes seem inconclusive when looking at this game, but Ratchet & Clank seemingly works by merging the two buffers generated in the process of creating anti-aliasing. While most of the edge-smoothing effect is lost, Insomniac can get away with the generation of significantly lower framebuffer while making it look pretty damn close to native 720p. Lower resolution means that more frames can be rendered per second and this is undoubtedly a key aspect in A Crack in Time running as smoothly as it does.
It should not be understated how much of a technical achievement this AA buffer merge technique actually is in terms of Ratchet & Clank's implementation. Many games have attempted this technique (the PS3 versions of the WWE games, for example), but none has been convincing enough to fool the human eye into thinking that the game is anything other than sub-HD. To its immense credit, Insomniac appears to have made an impressive breakthrough here.
Similar to the last Ratchet game, we peg A Crack in Time at 960x704 in the final analysis, but the proprietary AA buffer merging technique does an astonishingly good job of creating the effect of native 720p, albeit with a slight blur. Moving down to a solid 30FPS would effectively double the amount of time Insomniac has with which to render a frame - more detail, more objects, more overdraw (it's this latter element that seems to cause the most frame-rate issues in this game). Although there are few complaints about the image quality, the developer could also shift to the native 720p resolution with 2x multi-sampling anti-aliasing - the standard set by many other first-party exclusives in the Sony stable (Uncharted 2, God of War III and MAG, just for starters).
Playing the latest Ratchet & Clank, it's clear that while Uncharted developer Naughty Dog has scooped plenty of plaudits for its tech, Insomniac perhaps isn't getting the credit it deserves. The game is immensely detailed, throws about tons of those difficult alpha transparencies with abandon, renders absurd amounts of objects at almost any given point and works tirelessly in attempting to sustain 60FPS. The only problem is that the developer is so ambitious that it's just too much for the engine to cope with. To its credit though, Insomniac keeps the v-sync fully engaged at all time.
The first level - with Clank centre-stage - does a decent job of maintaining 60FPS with just a few exceptions. However, a couple of minutes into the video, the Ratchet stages show a much lower average frame-rate.
View this video in HD In Insomniac's own research, one of the conclusions reached was that a solid, sustained frame-rate was important: more important than 60FPS. "Frame-rate should be as consistent as possible and should never interfere with the game," Mike Acton says. "A solid frame-rate is still a sign of professional, well-made product. When there is a trade-off for frame-rate, it needs to be clearly worth it... it must introduce clear improvements on what the player sees, and never used as an excuse to not optimise the game or art."
It's an ethos a developer of Insomniac's status isn't likely to disappoint on should it decide to move to 30FPS. However, there are a couple of massive alarm bells being rung in Acton's post that I need to take issue with. First of all is the concept that "a drop in frame-rate is interestingly seen by some players as a reward for creating or forcing a complex setup in which a lot of things must happen on the screen at once. As in, 'Damn! Did you see that? That was crazy!'"
There is
some element of fun in stressing a game's engine to breaking point when gameplay isn't really an issue, so I can kind of see where Mike Acton is coming from, looking at it from a tech perspective, but let's hope that this "complex set-up" doesn't require much in the way of player response, because controller latency in these types of situation is definitely going to be compromised.
The second thing is the composition of Acton's very first sentence: that Insomniac is committed to "the best looking games you can buy on a console". What surprises me about the line is that graphics are simply part of an overall package, and gameplay should be king, and there are many situations in which 60FPS is crucial to the very core of the game.
While there are many, many strong arguments that the next Ratchet & Clank game would be better off targeting a solid 30FPS, there is also the sense that Insomniac's reasoning and approach to its research in Acton's blog post was a touch skewed. I'm curious if he considered whether the smoothness of the visuals is actually a core component of a game's visual appeal, as opposed to being some other kind of nebulous factor. All those graphics scores will presumably take into account the frame-rate, and marks will be lost if gameplay is slow, jerky or intrusive.
Acton also introduces some level of supposition to his argument based on his chats with gamers that backs up his thinking, but I suspect that actual focus testing would produce some interesting results. For example, Resident Evil 5 running on a good PC can look exactly the same as its console counterpart, but if you're running it at 60FPS, animations are smoother and more realistic, and gameplay is obviously more responsive. Put x amount of gamers in front of both versions, and I'm fairly sure that they would say that the PC version "looks better" and could well score higher. While it's true that many players may not be sensitive to frame-rate, it is also important to point out that gamers (and indeed many reviewers) don't have the vocabulary to articulate how important it is and how they relate to it.
However, the real elephant in the room is the concept of game marketing - the vast majority of which puts the emphasis firmly on image definition over the importance of a high frame-rate, and it's here that it is almost impossible to disagree with Insomniac's reasoning.
The first taste anyone has of any particular game comes from screenshots, and a screen generated in 33.33ms (i.e. a 30FPS game) can and usually will look better than one generated in 16.67ms (60FPS). And after the screenshots, the chances are that the marketing will move up to video - which the vast majority of the potential audience will see on the internet. Streaming assets online are invariably 30FPS, the same frame-rate that Insomniac will presumably be targeting with its next game.
In other words, the entire awareness of the game in the minds of the audience will be built up in an environment where no more is required than a target frame-rate of 30FPS. Indeed, all the advantages of rendering to the higher frame-rate will be completely worthless right up until the point where the playable demo hits, and gamers will be able to feel the difference.
"Basically nobody sees 60FPS until they start playing the game. And at that point the game has already been sold. Sad but true." That's our old friend Sebastian Aaltonen, of Trials HD fame, posting on the
Beyond3D Forum. "A good example: We tried to get 60FPS videos of Trials HD around the net. But the support for 60FPS video is almost non-existent in all the popular video downloading sites. And even our Xbox dashboard video trailer was automatically compressed down to 30FPS to reduce its bandwidth."
However, Aaltonen sounds an optimistic note about the impact of 60FPS gameplay done right. "We have incredibly high demo-to-full-version conversion ratio. When people see the game at its full 60FPS glory, they will very likely buy the full version. Half a million unit sales for an Xbox Live Arcade game is pretty good sign of support towards 60FPS gaming."
More than that, a total of 52 million sales for the Gran Turismo franchise, solid in its support for 60FPS since it migrated to PS2, also suggests that - as a core part of the gameplay package - 60FPS can be of crucial importance. Not only that, but Infinity Ward's Call of Duty 4: Modern Warfare is also the highest-selling game of this generation, and it's hugely unlikely that the developer of either of these gaming behemoths will ever waver from their commitment to the highest possible frame-rates and the lowest
controller latencies.
So, what's the solution here? Is there any kind of middle ground? Looking at the performance profile of Ratchet & Clank, couldn't Insomniac lock the game at 45FPS to maintain the solid frame-rate, while still offering a superior level of feedback to the average 30FPS title? The short answer is "no", mostly because of the refresh rate of our HDTVs being, for the most part, 60Hz.
Locking FPS at anything other than 30 or 60 introduces noticeable and off-putting judder. Some frames are on-screen for longer than others, upsetting visual coherency. It's the same reason why watching an NTSC DVD movie shows jerky panning shots - it's trying to squeeze 24 frames into a 60Hz refresh. Compare it with a PAL DVD (sped-up to run at 25FPS, blending nicely with the 50Hz update of the screen) and the judder magically disappears. The effect in a video game, featuring far more frequent and dynamic motion, is exponentially amplified and simply doesn't look that good.
So, why will Polyphony Digital and Infinity Ward doggedly stick to 60FPS, while Insomniac is ready to drop down to 30? The bottom line is that the mechanics of some games are so important that they override the marketing imperative, or else creative ways are found to accommodate the lust for ever more impressive screenshots. GT wouldn't be GT without its physics, and they rely on 60Hz feedback (and the same goes for Forza 3 and Trials HD). Call of Duty, like Burnout Paradise, is a game based on ultra-low latency response and split-second decisions, and again, without 60Hz, the core experience would be compromised.
In terms of the marketing angle covered earlier, Polyphony generally doesn't release actual screenshots of its games - certainly not of GT5 (see Digital Foundry's piece on the
bullshot phenomenon) and in the case of Infinity Ward, its screenshot endeavours can be covered off very nicely by using the PC version running at 1080p. Forza 3 is a bullshot-free zone by reputation, but it's worth pointing out that the game's replay mode - running at 30FPS - effectively provides all the screenshot and video quality required for marketing purposes, without the pain of compromising the core appeal of actually playing the game.
With Insomniac's Ratchet & Clank, the platform-style gameplay simply doesn't require the level of response you get with a 60FPS frame-rate. And with all the respect in the world, as the performance graph shows, often you're not getting it anyway. 60FPS offers very little to the game's overall proposition to gamers, and Mike Acton is right: in terms of marketing value and potential sales, shifting the focus from frames to an overall improved image quality will probably yield dividends. It's why it's the right move for Insomniac and Ratchet & Clank, even if it's not difficult to disagree with many of the fundamental aspects of how the company seemingly came to its decision.