Not sure how I missed this, great breakdown. The Choosers are evaluated from Top-To-Bottom btw, so it will indeed choose the first Database that meets all the conditions.
I apologize if this is a dumb suggestion. But maybe there is an option to solve the problem with offst using the scheme they suggested? That is, do not try to insert the usual schemes with blending and ik controllers, and write an analog of movement component where instead of the trajectory of the character will be a vector where he aims. Make bases for different overlays and add as if a second layer of motion matching responsible only for layers with weapons (upper body / arms)
@@nick_lyrick Yeah you kind of lost me, not really sure what your trying to say but it sounds like you want to combine motion matching and layering together. That would increase the cost I would suspect by a moderate amount. As far as different Bases for different overlays it needs to be a Neutral Pose, if say you dynamically layer it using a pose meant to be for a state where he is aiming the weapon and you apply that over the weapon it will cancel it out so that he is in the lowerbody pose not the overlay pose. It would be like saying 2 - 1 = 1 1 + 1 = 2 Now your back at 2, Aka the Lowerbody pose.
@@unrealdevop Uh, excuse me. English is not my first language. I'm just looking at their implementation of aim offset for head rotation. There's a curve. I was thinking that it is possible to do the same thing for the arms only with a sharper curve It also seems that most of the animation selection is based on trajectory. It seems that if you look at the concept in general, animations with aiming need a reference point, the same as the trajectory for movement animations, etc. Some kind of arrow component. To which you can adjust the hands. The fact that one animation will cancel another in the example you described above is clear. But motion matching still looks like a black box. A lot of things are not clear how to integrate
@@nick_lyrick Motion Matching isn't going to work great with something like that. I imagine they would have added that as an example if they thought it was even remotely plausible.
Not sure how I missed this, great breakdown. The Choosers are evaluated from Top-To-Bottom btw, so it will indeed choose the first Database that meets all the conditions.
I apologize if this is a dumb suggestion.
But maybe there is an option to solve the problem with offst using the scheme they suggested?
That is, do not try to insert the usual schemes with blending and ik controllers, and write an analog of movement component where instead of the trajectory of the character will be a vector where he aims.
Make bases for different overlays and add as if a second layer of motion matching responsible only for layers with weapons (upper body / arms)
@@nick_lyrick Yeah you kind of lost me, not really sure what your trying to say but it sounds like you want to combine motion matching and layering together. That would increase the cost I would suspect by a moderate amount. As far as different Bases for different overlays it needs to be a Neutral Pose, if say you dynamically layer it using a pose meant to be for a state where he is aiming the weapon and you apply that over the weapon it will cancel it out so that he is in the lowerbody pose not the overlay pose. It would be like saying
2 - 1 = 1
1 + 1 = 2
Now your back at 2, Aka the Lowerbody pose.
@@unrealdevop Uh, excuse me. English is not my first language.
I'm just looking at their implementation of aim offset for head rotation. There's a curve. I was thinking that it is possible to do the same thing for the arms only with a sharper curve
It also seems that most of the animation selection is based on trajectory.
It seems that if you look at the concept in general, animations with aiming need a reference point, the same as the trajectory for movement animations, etc. Some kind of arrow component. To which you can adjust the hands.
The fact that one animation will cancel another in the example you described above is clear.
But motion matching still looks like a black box. A lot of things are not clear how to integrate
@@nick_lyrick Motion Matching isn't going to work great with something like that. I imagine they would have added that as an example if they thought it was even remotely plausible.