Gamasutra: The Art & Business of Making Gamesspacer
View All     RSS
October 23, 2014
arrowPress Releases
October 23, 2014
PR Newswire
View All





If you enjoy reading this site, you might also want to check out these UBM Tech sites:


id's John Carmack Chooses Framerate Over Graphical Fidelity
id's John Carmack Chooses Framerate Over Graphical Fidelity
August 22, 2011 | By Frank Cifaldi

August 22, 2011 | By Frank Cifaldi
Comments
    20 comments
More: Console/PC, Programming, Art



Despite a company history of leading the industry in graphic fidelity, id Software's upcoming Rage -- while still among the best-looking games out there -- intentionally prioritized framerate over graphics in development.

According to the company's legendary programmer John Carmack, that decision was a "hard-fought battle" internally.

"I made the conscious decision that the user is going to get more value out of running at a higher framerate than me making the pixels pretty," he told Gamasutra as part of a larger interview.

"I think I could have made the game look better at 30 hertz. We could have had some more design freedom," he admits, though he says the decision to focus on keeping the game at 60 frames per second will look better to just about everyone, including those who can't tell the difference between 60 and 30.

"Most people get at least a subliminal feel about it," he says. "It's more responsive. It's crisper. It's smoother."

But framerate perception has its limits. Carmack tells us he conducted a number of experiments to see if players responded to a doubling of framerate, having them play a demo in both 60FPS and 120FPS.

"Interestingly, almost nobody can tell the difference between 120 and 60," he says, adding that this "means that we've got this benefit curve on here, and 60 is kind of right at the knee."

This, in addition to technological advances like the iPhone 4's nearly print-level resolution, is "cool thing that in many ways, we are approaching sort of the biological limits of what betterness we can display," he says.

More insight from one of the most fascinating figures in video game development is available in our full interview.


Related Jobs

Avalanche Studios
Avalanche Studios — New York, New York, United States
[10.22.14]

UI Programmer
Avalanche Studios
Avalanche Studios — New York, New York, United States
[10.22.14]

UI Artist/Designer
Wargaming.net
Wargaming.net — Hunt Valley, Maryland, United States
[10.22.14]

Lead UI Engineer
Wargaming.net
Wargaming.net — Chicago, Illinois, United States
[10.22.14]

Lead UI Engineer










Comments


Clinton Keith
profile image
Agreed about 60Hz. We did an experiment with a game where we pulled in all the detail we could (draw distance, etc) just to see what the game felt like at 60Hz. It was amazing....it felt more crisp, responsive and smooth.



Unfortunately we did it so late in development that we couldn't ship it at 60Hz. The urge to keep adding "stuff" in the game and accept "mostly 30Hz" usually overpowers the will to maintain 60Hz. You need a guy like Carmack to put pressure on to make it happen ;)

Harry Fields
profile image
A consistent framerate is much more attractive than a sometimes 60, sometimes 40, sometimes 27 hz rate. If you can't keep it at a constant 60, go for a pretty constant 30.

Kevin Fishburne
profile image
That is a damn good point, which outside of my own head I've never heard spoken. A fixed frame rate is much more important than a high one. No one will notice it's "slow" until it slows down.

Mathieu MarquisBolduc
profile image
I dont know, going from 30 to 60 fps means surrendering HALF your graphical content. People tend to think of it as prettier pixels, but its mostly about graphical content, like having more npcs on screen for example. Everytime I see "crowds" of barely a dozen people in games I cringe. And Im kind tired of the trend of having a maximum of about 6 npcs on screen, especially in shooters.



Personally I'll take more & less static content over the subliminal feel of 60 fps.



Insomniac did some research too...



http://www.insomniacgames.com/how-much-does-framerate-matter/

Michael van Drempt
profile image
But I'd take that research as it was intended - informal and largely report-based (as opposed to solid metrics of some kind). Also, it was talking about console games, and when you're looking at a PC FPS (the genre, not the framerate), where you're whipping the mouse around and doing split-second turns, framerate can affect gameplay a lot more than in a game where the turn speed is already limited to accommodate the thumbstick controls.



Also, I'd point out that the findings were:



A higher framerate does not significantly affect sales of a game.

A higher framerate does not significantly affect the reviews of a game.



What they were not addressing was whether a higher framerate actually makes a game better, which would be a much more case-by-case judgement. In most games, probably not, but Carmack already made the point that it makes a bigger difference in driving games than in other genres, so that would have informed the decision as well.

Glenn Sturgeon
profile image
But with insomiac things are a lot different, like they are stuck on console (moderate hardware) only developement, thier games don't stay at 60fps all the time and thier engines dont look nearly as good as id software's.

Rage on consoles running 60fps will look "at least on par" with todays best looking console titles and well beyond anything insomniac has done. I love insomniac but they are more focused on sales than top end proformance and base alot of ther 60fps theories on " review scores" that come from people that have little clue beyond trivia about games since they spent thier college time studying english, not games, design or programming.

Insomniac wishes thier games would sell as well as id's do.

Joe McGinn
profile image
Good call, not enough developers sticking to their guns on this.

David Holmin
profile image
I can only speak for myself, but 60Hz looks and feels much better. My experience with 30Hz is that quick movements and events with short duration feel really stuttery and jerky.

Curtis Turner - IceIYIaN
profile image
Besides being stuck in crappy 60 fps, because of old tv's/consoles, CarmacK has probably locked in a fps number because of the complaints from Quake 3. www.ESReality.com has talked about this a lot, I've only glanced over it. Mainly it has to do with the strafe jump, speed jumping. Players with a higher fps, I think 120/125, could do longer jumps.



Next time you run those crappy "tests", force your players in a 99++ fps on the game/monitor. After a long time, THEN switch the monitor to 60 hertz. I bet that'll strain their eyes and you'll hear the truth.

David Holmin
profile image
The game running at 60Hz probably won't stop you from setting your monitor to 120Hz - and voila - no strain on your eyes.

Arnaud Clermonté
profile image
Actually it's new (LCD) monitors that are stuck at 60.

I remember 10 years ago, most CRTs could go up to 90 Hz, and I was playing Quake3 v-synced at 90 fps.



David Holmin, no, you don't increase the game framerate by just changing the monitor's frequency...

Bob Johnson
profile image
60 fps is much better. Metroid Prime is a great example of the silky smooth nature of 60 fps.

Kevin Fishburne
profile image
While there's something to be said about the fluidity of a game at 60 FPS (assuming it can maintain it consistently), I have to disagree with the man here.



Saving Private Ryan runs at 24 FPS and looks way better than Rage. It also looks pretty good on an old TV with NTSC. Why does it still look so good despite the horrid FPS and resolution? Because the graphical content is photorealistic. When everything in your scene looks right, the frame rate and resolution become much less significant. Those two things should be the -last- thing a graphic designer considers and thought of as bonuses.



There's a lot that can be done at a lower frame rate and resolution to make a game look and feel great such as motion blur, FSAA, higher polygon counts, LOD, depth of field, etc. Employ all the usual tricks of fidelity, throw in a few more for good measure, and ramp down the FPS and resolution to make it possible. When a game looks like Saving Private Ryan we'll have finally arrived.

Mathieu MarquisBolduc
profile image
I hate to argue since I agree with you, but to be fair, I think only way to emulate the sampling function of a real-life movie camera (which is somewhat gaussian depending on the tech, versus the dirac function of a renderer) is to render at the highest possible fps then downsample to 24 using the aforementioned gaussian function. So if you want something that looks like Saving Private Ryan in motion, you'll still need a high fps, at least as a middle step :)



To me its really about putting more content: more npcs, dynamic objects, npcs that dont look all the same or all wear helmets and masks so you can see their face, people with long hair, etc. For most games, twice as much visual content is better than a slight improvement in fluidity.

John Mawhorter
profile image
Even 60 isn't good enough, but in this era at least it is the right commitment to make. I can tell the difference (and quality of my play) between 60, 80, and 100 in Counterstrike 1.6, and I think 100 should be the industry benchmark, but given the graphics sells titles argument I don't think a very popular opinion.



@Kevin Your example is rather poorly picked. In a movie you are watching the screen. In a game you are reacting as quickly as possible to events on screen. Sure photorealism might in some way help, but pure frame numbers clearly and obviously affect your reaction time, aim, skill, combos, etc. etc. etc.

Kevin Fishburne
profile image
That is a good point, however as the subsequent poster suggested someone really needs to do some studies on this. If, for example, 90% of players say the graphics in the 30 FPS version are significantly better and the 60 FPS players only show a 10% performance gain, I'd say the FPS loss is negligible. Basically all these factors need to be studied.



One of the biggest factors I think is how far a target moves on-screen when you change orientation (aim/turn). If the mouse is sensitive and the target is distant, a lower FPS will make it more difficult to quickly fine tune your shot. A higher FPS for everything else is pretty much wasted other than the psychological effect of it simply feeling smooth.

John Mawhorter
profile image
Also I read a detailed page analyzing FPS from a games point of view that said nobody really knows the upper end of FPS that "matters" to the human eye. Obviously there is a point at which gains in accuracy/reaction speed aren't that high vs. the cost of getting there, but it could be 150, 300, 500. Carmack's user test was probably not very strictly controlled/conducted. While the difference is difficult to notice consciously there is a different feeling that over time can be detected. Differences of 20 or more FPS seem to be noticeable to me consciously after a while of playing. And the unconscious effect on reaction time/skill could really easily be tested for scientifically (why has noone done this?)

John Mawhorter
profile image
Also depending on the game engine (as pointed out above with Quake 3's longer jumps for higher FPS rates) your frame rate can be tied to updating game objects/events or not and therefore have even more of an effect on game feel. Since game feel is very important it certainly impacts player and reviewer opinions. Smoothness of gameplay is present in almost all good/great games and at least some of that depends on refresh rate, imo.

Amir Sharar
profile image
If I remember correctly, the United States Air Force did some research on this topic (which shouldn't be a surprise to any of us, makes sense) and they saw that 120 fps was the limit for most people. I'm sure there are exceptions to this rule, as the research showed the some people couldn't see past 60. Keep in mind that this is dealing with 2D displays, other research in regards to the human eye demonstrate that it operates at levels of 1/100 of a second and the brain takes that data and interprets it.



There is a reality that must be kept in mind with any discussion in regards to framerate. When we see things, we are able to see motion via blur. This is also captured when we record video (if the shutter speed allows for it). When we watch movies at 24fps, we may consider them acceptable but remember that blur is also captured to allow us to easier follow movement. The brain interprets blur to give us a sense of motion. If you looked out a window and a ball was thrown from right to left, you'd know it from the blur. As the ball travels in various directions your brain interprets all the light that entered your eye and gives you that image.



When we see 30fps in most games, we do not see this blur, making the game appear choppy. 60fps adds more to visuals allowing our brain to interpret blur with the information the eyes receive, just like in real life.



There are 30fps games like Project Gotham 3 and 4 that employ motion blur so well that the end effect is that it is a very smooth looking game, but still not to the same effect as a 60fps title.



So let's rid ourselves of these terms such as "psychological effect of being smooth" or bunk like that. The reality is that we brains interpret what our eyes see, and see things like motion blur to give us a sense of direction for moving objects. The end effect is realism, the brain considers the motion to be more like real life.



Ultimately, if one is striving for realism in their game, 60fps may be the way to go unless you are willing to put per object motion blur in your game (which could be very expensive). Also, per object motion blur doesn't tell us what the user is looking at, making it still artificial looking. So 60fps is the best bet as the user can choose to track whatever visual they want.



http://en.wikipedia.org/wiki/Saccadic_masking



Ever ride the subway and focus on a window frame, to note that the background has blurred? Then have you attempted to track the movement of the background with your eyes (thus eliminating the blur)?



I do think that ultimately that tests need to be done in regards to user perception (particularly among gamers) on the topic of graphical detail vs. framerate (I would also throw in anti-aliasing as well, as I think it could be the biggest factor in making things look more "realistic" apart from a believable lighting model), and in fact I was going to write a Gamasutra blog entry on this topic but my general laziness has prevented me from doing so, I don't think so now as it seems that others agree that some research needs to be done. :)

Kevin Fishburne
profile image
60 or even 120 fps isn't going to give you decent natural motion blur unless the screen's blurry. Poor rendering can be done faster intentionally. At 30+ FPS it feels slicker and smoother, but it's irrelevant as far as collision detection, network synchronization and user input are concerned. Those elements are generally paced at a simpler and/or less-frequent resolutions to balance the load.


none
 
Comment: