Sixfortyfive
He who pursues two rabbits gets two rabbits.
I just bought my first Wii U Virtual Console game (Super Mario Bros: The Lost Levels). While trying it out on the Gamepad, I couldn't help but feel as if the game was less responsive than it should be. SMB2J is a game that I'm pretty familiar with and played a lot on the original Wii VC, and I currently have an original NES with SMB1 hooked up to a CRT TV, so it's easy for me to cross-check with original hardware. And after doing so, SMB2J definitely seems to be a bit off on the Gamepad.
While I consider input lag to be a pretty big deal, it usually isn't obvious to me and detrimental to my gameplay unless it exceeds 2 frames. And by all accounts, the Gamepad has been lauded for having extremely low latency compared to typical TVs; the only measurements I've seen in reviews peg it at about 1 frame. This kind of puzzled me, as it would imply that either the Wii U Virtual Console has pretty lousy and laggy emulation overall, or that the Gamepad performs worse with Virtual Console games than it does with native Wii U games. I tried to research the issue online but couldn't really find anything conclusive, so I decided to do some testing of my own.
I hooked the Wii U up to a CRT VGA monitor, using component cables, 480p resolution settings, and an XRGB-3 in B1 mode to bridge the gap from component to VGA. The XRGB-3 adds approximately 1ms-to-2ms of lag, which is a fraction of a single frame and pretty negligible in the grand scheme of things. These are the only means I have at my disposal to hook the system up to a CRT in progressive scan, as I have no CRTs with component inputs. Then I held up the Gamepad next to the monitor and recorded some test footage with both screens in the shot to see how much the Gamepad lags behind the CRT. Unfortunately, the only camera I have on hand maxes out at 30fps, which means that the footage is less precise than I'd like it to be on these particular 60fps games, but I think what I've captured is still useful.
First, I loaded Super Mario 3D World to check the performance on a retail Wii U game. Here's a short frame-by-frame capture (at 30fps) of the start of Mario's jump:
Frame 1: Mario's feet are planted firmly on the ground. The jump button is pressed.
Frame 2: On the CRT, Mario's feet have left the ground slightly and there's a slight puff of smoke beneath him to indicate movement. The Gamepad is lagging behind a little bit, but you can see that he's started to move and that the puff of smoke is partially visible. The last frame of his standing animation seems to be interpolated with the first frame of his jumping animation (again, a consequence of capturing a 60fps game with a 30fps camera).
Frame 3: Mario has completely left the ground on both screens. The Gamepad is a little behind, but not by much. In this snapshot, the Gamepad appears to be displaying the same frames of animation that the CRT is in the current and previous shots blurred together.
I recorded several jumps like this, and the results for all of them looked pretty consistent with this example. The Gamepad isn't that far behind the CRT: maybe 1, not more than 2 frames (at 60fps). Definitely seems to corroborate the Gamepad's reputation for low latency video streaming.
Then I tried the same thing with The Lost Levels:
Frame 1: Mario grounded. The jump button is pressed.
Frame 2: Mario has completely left the ground on the CRT. He's still planted firmly in place on the Gamepad. No motion blur in this shot.
Frame 3: Mario can still be seen on the ground in the Gamepad shot, 2 frames after the start of this example. In each jump I captured in 3DW, it never took more than 1 shot before some movement was seen on the Gamepad. This would seem to indicate that Lost Levels is actually suffering from more delay than 3D World on the Gamepad. I captured several jumps in this game as well, and again, this behavior was pretty consistent throughout all of them.
Frame 4: Mario finally clears the ground in both shots.
Then I decided to capture some shots of the timer in both games:
In SM3DW, every time the clock dropped by 1 second on the CRT, it also dropped by 1 second on the Gamepad...
In fact, out of the several dozen transitions that I captured, only in this one can you kind of see that the Gamepad has a slight after-image of the previous number in the middle shot.
Then I did the same in SMB2J, and every transition was like this in the several dozen that I captured, with the Gamepad always taking 1 extra snapshot before catching up.
So, what gives? On one hand, I'm thankful that the Virtual Console itself doesn't seem to be extremely laggy (when played from the TV), but on the other, I'm still puzzled that the Gamepad seems to perform worse on the VC than for regular Wii U games. If anybody could link me some more thorough testing on this stuff or perhaps some precise 60fps or greater footage of their own, I'd very much appreciate it. I'd like to know exactly how big the difference is.
While I consider input lag to be a pretty big deal, it usually isn't obvious to me and detrimental to my gameplay unless it exceeds 2 frames. And by all accounts, the Gamepad has been lauded for having extremely low latency compared to typical TVs; the only measurements I've seen in reviews peg it at about 1 frame. This kind of puzzled me, as it would imply that either the Wii U Virtual Console has pretty lousy and laggy emulation overall, or that the Gamepad performs worse with Virtual Console games than it does with native Wii U games. I tried to research the issue online but couldn't really find anything conclusive, so I decided to do some testing of my own.
I hooked the Wii U up to a CRT VGA monitor, using component cables, 480p resolution settings, and an XRGB-3 in B1 mode to bridge the gap from component to VGA. The XRGB-3 adds approximately 1ms-to-2ms of lag, which is a fraction of a single frame and pretty negligible in the grand scheme of things. These are the only means I have at my disposal to hook the system up to a CRT in progressive scan, as I have no CRTs with component inputs. Then I held up the Gamepad next to the monitor and recorded some test footage with both screens in the shot to see how much the Gamepad lags behind the CRT. Unfortunately, the only camera I have on hand maxes out at 30fps, which means that the footage is less precise than I'd like it to be on these particular 60fps games, but I think what I've captured is still useful.
First, I loaded Super Mario 3D World to check the performance on a retail Wii U game. Here's a short frame-by-frame capture (at 30fps) of the start of Mario's jump:
Frame 1: Mario's feet are planted firmly on the ground. The jump button is pressed.
Frame 2: On the CRT, Mario's feet have left the ground slightly and there's a slight puff of smoke beneath him to indicate movement. The Gamepad is lagging behind a little bit, but you can see that he's started to move and that the puff of smoke is partially visible. The last frame of his standing animation seems to be interpolated with the first frame of his jumping animation (again, a consequence of capturing a 60fps game with a 30fps camera).
Frame 3: Mario has completely left the ground on both screens. The Gamepad is a little behind, but not by much. In this snapshot, the Gamepad appears to be displaying the same frames of animation that the CRT is in the current and previous shots blurred together.
I recorded several jumps like this, and the results for all of them looked pretty consistent with this example. The Gamepad isn't that far behind the CRT: maybe 1, not more than 2 frames (at 60fps). Definitely seems to corroborate the Gamepad's reputation for low latency video streaming.
Then I tried the same thing with The Lost Levels:
Frame 1: Mario grounded. The jump button is pressed.
Frame 2: Mario has completely left the ground on the CRT. He's still planted firmly in place on the Gamepad. No motion blur in this shot.
Frame 3: Mario can still be seen on the ground in the Gamepad shot, 2 frames after the start of this example. In each jump I captured in 3DW, it never took more than 1 shot before some movement was seen on the Gamepad. This would seem to indicate that Lost Levels is actually suffering from more delay than 3D World on the Gamepad. I captured several jumps in this game as well, and again, this behavior was pretty consistent throughout all of them.
Frame 4: Mario finally clears the ground in both shots.
Then I decided to capture some shots of the timer in both games:
In SM3DW, every time the clock dropped by 1 second on the CRT, it also dropped by 1 second on the Gamepad...
In fact, out of the several dozen transitions that I captured, only in this one can you kind of see that the Gamepad has a slight after-image of the previous number in the middle shot.
Then I did the same in SMB2J, and every transition was like this in the several dozen that I captured, with the Gamepad always taking 1 extra snapshot before catching up.
So, what gives? On one hand, I'm thankful that the Virtual Console itself doesn't seem to be extremely laggy (when played from the TV), but on the other, I'm still puzzled that the Gamepad seems to perform worse on the VC than for regular Wii U games. If anybody could link me some more thorough testing on this stuff or perhaps some precise 60fps or greater footage of their own, I'd very much appreciate it. I'd like to know exactly how big the difference is.