Kyu Puff
Smash Champion
The other day I was playing with PDub when we noticed something strange. He was playing Marth, and I crouched behind him as he up-tilted. I crouch canceled the up-tilt, and neither Ice Climber received any knockback. This might seem normal at first--you're not supposed to go anywhere when you crouch cancel--but the thing is, Nana can't actually crouch cancel. If the up-tilt hit both ICs, she would fly away.
We replicated the situation. Every time he up-tilted, Popo would take damage, but Nana would stay crouching in place, receiving no damage or knockback as if the move had completely whiffed. Our first thought was that maybe they had slightly different X-coordinates, causing the hitbox of up-tilt to connect with Popo but just barely miss Nana. This turned out not to be the case, as it worked with multiple spacings, and Popo seemed to be overlapping Nana completely.
Then PDub told me that something similar is known to happen with Yoshi's dash grab. Yoshi has a lot of trouble dash grabbing Sheik, because he tilts his head to the right, and his tongue "passes behind the bend in Sheik's knees." This didn't make any sense to me--every hitbox/hurtbox I'd ever seen was depicted in 2-Dimensions. But, apparently it is a thing; hitboxes have X- and Y-, as well as Z-, coordinates, and need to overlap with a hurtbox in all three dimensions in order to register a hit.
When Nana and Popo stand in place (or shield, or crouch), their hurtboxes are "staggered" along the Z-axis, even though they are aligned horizontally. Nana is further away from the camera, and Popo is closer. The result is that moves with a large enough Z-offset might only hit Popo or only hit Nana. We used the camera to examine Marth's u-tilt, and found that both the very beginning and very end of its animation "veer off" to the side, wihch explains why it could hit one Ice Climber but completely miss the other.
Yoshi's dash grab similarly has a large Z-offset. He tilts his head to his right, which is towards the camera if he's facing right, and away from the camera if he's facing left. The result is that if you're standing on the left side of Yoshi, he'll almost always grab Nana (which makes dash grab a useless move), but if you're on his right, he'll almost always grab Popo (which actually makes it fairly useful). Thus, if everything else is held constant, it's strictly better to fight Yoshi from the left side.
Most moves are relatively centered, so this mechanic rarely ends up making a difference. However, there are some moves which seem to hit one Ice Climber over the other a disproportionate amount of the time. Another move we tested is Samus's standing grab, which works the same way (except in close quarters, where it seems to be random).
One question I think would be interesting to investigate is this: How does the Z-position of a move's hitbox affect the way it interacts with both Ice Climbers? In particular, how do moves with disappearing hitboxes (i.e. grabs and projectiles) "decide" which Climber to hit? If a move isn't perfectly centered, does it have a higher probability of hitting the IC whose hurtbox overlaps it more? Or is it always 50-50, except in cases where the hitbox has zero overlap with one IC's hurtbox?
Here is a tool you can use to directly measure the Z-coordinates of different moves: http://smashboards.com/threads/physics-input-etc-display-in-develop-mode.391956/ I personally can't get it to run on my laptop, but hopefully someone with a faster computer will be able to flesh out this research.
We replicated the situation. Every time he up-tilted, Popo would take damage, but Nana would stay crouching in place, receiving no damage or knockback as if the move had completely whiffed. Our first thought was that maybe they had slightly different X-coordinates, causing the hitbox of up-tilt to connect with Popo but just barely miss Nana. This turned out not to be the case, as it worked with multiple spacings, and Popo seemed to be overlapping Nana completely.
Then PDub told me that something similar is known to happen with Yoshi's dash grab. Yoshi has a lot of trouble dash grabbing Sheik, because he tilts his head to the right, and his tongue "passes behind the bend in Sheik's knees." This didn't make any sense to me--every hitbox/hurtbox I'd ever seen was depicted in 2-Dimensions. But, apparently it is a thing; hitboxes have X- and Y-, as well as Z-, coordinates, and need to overlap with a hurtbox in all three dimensions in order to register a hit.
When Nana and Popo stand in place (or shield, or crouch), their hurtboxes are "staggered" along the Z-axis, even though they are aligned horizontally. Nana is further away from the camera, and Popo is closer. The result is that moves with a large enough Z-offset might only hit Popo or only hit Nana. We used the camera to examine Marth's u-tilt, and found that both the very beginning and very end of its animation "veer off" to the side, wihch explains why it could hit one Ice Climber but completely miss the other.
Yoshi's dash grab similarly has a large Z-offset. He tilts his head to his right, which is towards the camera if he's facing right, and away from the camera if he's facing left. The result is that if you're standing on the left side of Yoshi, he'll almost always grab Nana (which makes dash grab a useless move), but if you're on his right, he'll almost always grab Popo (which actually makes it fairly useful). Thus, if everything else is held constant, it's strictly better to fight Yoshi from the left side.
Most moves are relatively centered, so this mechanic rarely ends up making a difference. However, there are some moves which seem to hit one Ice Climber over the other a disproportionate amount of the time. Another move we tested is Samus's standing grab, which works the same way (except in close quarters, where it seems to be random).
One question I think would be interesting to investigate is this: How does the Z-position of a move's hitbox affect the way it interacts with both Ice Climbers? In particular, how do moves with disappearing hitboxes (i.e. grabs and projectiles) "decide" which Climber to hit? If a move isn't perfectly centered, does it have a higher probability of hitting the IC whose hurtbox overlaps it more? Or is it always 50-50, except in cases where the hitbox has zero overlap with one IC's hurtbox?
Here is a tool you can use to directly measure the Z-coordinates of different moves: http://smashboards.com/threads/physics-input-etc-display-in-develop-mode.391956/ I personally can't get it to run on my laptop, but hopefully someone with a faster computer will be able to flesh out this research.