The Effects of Frame Rate and Resolution on Users Playing First
Person Shooter Games
Mark Claypool
CS DepartmentWorcester Polytechnic
Institute
http://www.cs.wpi.edu/~claypool/papers/fr-rez/
Kajal ClaypoolFeissal Dama
CS DepartmentUniversity of Massachusetts,
Lowell
January 2006MMCN, San Jose, California 2
Computer Games and Performance
• Latest computer games push capabilities of hardware in “quest” for more detailed, realistic graphics
• Single game runs on varied hardware– PC : Old (600 MHz P3, 32 MB Video) or New (3 GHz P4, 256 MB Vid)
– Platform: PC, Console (i.e. Xbox), Hand-held (i.e. PSP)– Result: Uneven hardware capabilities, opportunities for
performance tuning
• Key factors for game performance are:– Frame Rate – higher frame look smoother, provide
more temporally precise feedback– Frame Resolution – higher resolutions look better,
provide more visually precise feedback
January 2006MMCN, San Jose, California 3
Motivation
• Unfortunately, often cannot have both high Frame Rate and high Frame Resolution– Ex: Hand-held devices have constrained resources
(small screens, limited power)– Ex: Older computers (often, only 1 year!) cannot run
latest games at maximum frame rate or resolution
• Tradeoff between Frame Rate and Frame Resolution– Higher resolutions mean lower frame rates and vice
versa
• How are frame rates and resolutions chosen?– Game console designers and hand-held designers
choose resolution for user•Frame rate may depend upon processing load
– PC gamers choose it by “feel”– Not guided by science
January 2006MMCN, San Jose, California 4
Related Studies
• Passive Users [1,2,3,4,5,6,7]– Users assess video with various frame rates and
resolutions– Generally, decrease resolution then decrease
quality, but decrease in frame rate less so • Active Users [8,9,10,11]
– Users perform tasks under various frame rates and frame resolutions
– Generally, extremely low frame rates impact performance, but frame rates of 4+ can be acceptable
• Overall - more passive than most games and tradeoffs not compared
• Our goal – Effects of Frame Rate and Frame Resolution on User Performance for Games
January 2006MMCN, San Jose, California 5
Outline
•Introduction (done)
•Methodology (next)
•Results
•What’s Going On?
•Conclusions
January 2006MMCN, San Jose, California 6
Methodology Outline
•Select game
•Build custom map
•Select parameters
•Build test harness
•Solicit users
•Analyze results
January 2006MMCN, San Jose, California 7
Methodology – Select Game
• First Person Shooter (FPS)– Popular genre, especially for online play– Requires intense player interaction with time-critical
decisions•Impairment to the display quality can cost virtual lives
• Quake III Arena– Still fairly popular (~700 active servers via GameSpy)– Representative of current FPS games in terms of
perspective, weapon choices and gameplay – Allows control of frame rate and resolution at startup
January 2006MMCN, San Jose, California 8
Methodology – Custom Map
= spawn point
Opponent Platform
(Chasm)
User Platform
• Minimize effects of other players– Use Bot (Xaero)
• Minimize movement– Chasm cannot be
jumped– Wall behind player so
doesn’t fall off accidentally
• Maximize aiming and shooting– No cover (save wall to
protect spawn point)– Use of railgun that
cannot be rapidly fired– One shot per kill
January 2006MMCN, San Jose, California 9
Frame rate: 30, Resolution: 640x480
Example
January 2006MMCN, San Jose, California 10
Methodology – Demographics
Demographics provided once, before maps started
January 2006MMCN, San Jose, California 11
Methodology – User Perception
User perception provided after each map played
January 2006MMCN, San Jose, California 12
Methodology – Parameters
•5 frame rates: 3, 7, 15, 30 and 60 fps – Ranges previously studied– May appear during normal game play
•3 frame resolutions: 320×240, 512×384, 640×480
– Hand-held devices to low-end PCs– Observed trend may interpolate to higher
resolutions
January 2006MMCN, San Jose, California 13
Methodology –Test Harness
For each player …
• Gather demographics
• Play Quake III– Very high frame rate (80 fps)– Very high frame resolution (1024×768 pixels)
• Shuffle (Frame Rate, Frame Resolution) combos
• For each combo …– Play Quake III with (Frame rate X, Resolution Y)– Gather user perception
January 2006MMCN, San Jose, California 14
Methodology – User Solicitation
•Two-week period
•Game played on isolated PC in lab
•Range of enticements:– Enter raffle for three $50 gift certificates– Extra credit for courses– Refreshments for participants
January 2006MMCN, San Jose, California 15
Outline
•Introduction (done)
•Methodology (done)
•Results (next)
•What’s Going On?
•Conclusions
January 2006MMCN, San Jose, California 16
Aggregate Statistics
• 60 users provided “clean data” – 64 participated, but 4 removed because ended
early• Age:
– Most (~75%) 16-25 years old (ugrad CS students)– Almost 25% over 25 years old (grad CS students)
• Gaming:– Over 65% played games over 1 hour per week– 25% played 6+ hours per week– 50% casual gamers, moderate at shooters
• Gender:– About 20% female
•Only one more than casual gamer, compared to about 65% for males
January 2006MMCN, San Jose, California 17
Performance and Frame Rate
(Resolution: 512x384 pixels)
January 2006MMCN, San Jose, California 18
Performance and Frame Resolution
(Frame Rate: 15 fps)
January 2006MMCN, San Jose, California 19
Perception and Frame Rate
(Resolution: 512x384 pixels)
January 2006MMCN, San Jose, California 20
Perception and Frame Resolution
(Frame Rate: 15 fps)
January 2006MMCN, San Jose, California 21
Outline
•Introduction (done)
•Methodology (next)
•Results (done)
•What’s Going On? (next)
•Conclusions
January 2006MMCN, San Jose, California 22
60 fps 15 fps
7 fps 3 fps
Effects of Frame Rate on User Performance
January 2006MMCN, San Jose, California 23
Example – 15 Frames per Second
January 2006MMCN, San Jose, California 24
Example – 7 Frames per Second
January 2006MMCN, San Jose, California 25
Example – 3 Frames per Second
January 2006MMCN, San Jose, California 26
Effects of Frame Resolution on User Performance
640 x 480 320 x 240
January 2006MMCN, San Jose, California 27
Example – Resolution 320x240
Frame rate: 30
January 2006MMCN, San Jose, California 28
Conclusions• Frame Rate larger impact on performance than Frame
Resolution– Frame Rate critical for adequate game performance
•Frame rates of 3 fps and 7 fps not playable•60 fps provides 7-fold increase over 3 fps
– Frame Resolution has little effect on user performance•Users as effective at 320×240 as at 640×480
• Frame Rate and Frame Resolution both important for user perception– Effect of frame resolution similar to effect of frame rate– Top frame rate tested (60 fps) shows limit– Top resolution tested (640x480) not at limit
•Perceived quality increases linearly with square pixels• Dramatically different previous research on video
– Showed converse, that Frame Resolution mattered more– Suggests challenges in designing devices for games and
video
January 2006MMCN, San Jose, California 29
Future Work
•Other aspects of First Person Shooters– Different map conditions– Movement
•Other display tradeoffs– “Quality” from graphics effects
•Anti-aliasing, realistic water/grass …
•Additional demographics studies– Gender, age, gaming experience …
•Other computer games– Real-Time Strategy, Sports …
The Effects of Frame Rate and Resolution on Users Playing First
Person Shooter Games
Mark Claypool
CS DepartmentWorcester Polytechnic
Institute
http://www.cs.wpi.edu/~claypool/papers/fr-rez/
Kajal ClaypoolFeissal Dama
CS DepartmentUniversity of Massachusetts,
Lowell