35 replies to this topic

[CygnusX7]

There is a big difference from when I started playing and then after this change was made.
I often wish for MWO to be the way it was and the way a mech walks (feels) was definitely it.

[cSand]

Mystere, on 02 December 2015 - 01:08 PM, said:

Sometimes, a picture is definitely not worth a thousand words.

And besides, I think you missed my point.

Just yankin your chain a bit

[DaZur]

cSand, on 02 December 2015 - 01:17 PM, said:

Just yankin your chain a bit

cSand... The chain yanker. Like a Mentos but not quite as fresh....

[cSand]

DaZur, on 02 December 2015 - 01:24 PM, said:

cSand... The chain yanker. Like a Mentos but not quite as fresh....

generally leaves you desiring a mento after the fact


:|

[orcrist86]

All mechs are rigged for ik. Russ said so in a previous town hall. Just like standing animation is included for each mech. The problem with ik was 2 things. It ate player fps and messed up hsr and hit detection. Russ has also said that are looking to get it back in but need to see how it impacts things.

[Buster Machine 0]

Not to necro, but I never got into this game at launch due to the lack of Inverse Kinematics. It ruined the feeling of a mechwarrior game for me.

Even though I'm into the game now 3 years late, I'm still saddened by the immersion breaking leg movements.

[Ex Atlas Overlord]

Envisage, on 01 December 2015 - 11:53 PM, said:

Hey there, I don't know the term "Inverse Kinematics". Can you elaborate, and how does it change immersion.

Well...."Kinematics is the branch of classical mechanics which describes the motion of points (alternatively "particles"), bodies (objects), and systems of bodies without consideration of the masses of those objects nor the forces that may have caused the motion"

and...." an inverse function is a function that "reverses" another function. That is, if is a function mapping x to y, then the inverse function of maps y back to x."

So whatever functions they were using said "If the mech is standing like this, then the ground under it must look like this"....and the inverse of those functions would say "If the ground looks like this, then the mech will stand like this".

Edit: Y=a^x and Log(Y)=x are inverse functions of each other.

Edit 2: God damnit, did I fall for the necro?

Edited by The Atlas Overlord, 19 February 2016 - 11:39 PM.

[Juodas Varnas]

The Atlas Overlord, on 19 February 2016 - 11:11 PM, said:

Well...."Kinematics is the branch of classical mechanics which describes the motion of points (alternatively "particles"), bodies (objects), and systems of bodies without consideration of the masses of those objects nor the forces that may have caused the motion"

and...." an inverse function is a function that "reverses" another function. That is, if is a function mapping x to y, then the inverse function of maps y back to x."

So whatever functions they were using said "If the mech is standing like this, then the ground under it must look like this"....and the inverse of those functions would say "If the ground looks like this, then the mech will stand like this".

Edit: Y=a^x and Log(Y)=x are inverse functions of each other.

Edit 2: God damnit, did I fall for the necro?

Well, it's a good thread to necro.

Since Inverse Kinematics still hasn't been re-introduced into the game.
Which they really should.

I mean, seriously, mechwarrior games 20 years ago had proper IK.

[Cyborne Elemental]

I don't understand why they can't make it client side only.

When it comes to legs and hitboxes, this game is so sloppy because of low tickrate where I don't think it would really make much of an impact on gameplay if client side vs server side legs were slightly out of position.

As long as they're synced up in rotation positions, its not a huge deal IMO.

Don't know how the code operates in that regard, don't really care, but there has to be an easy way to make it not look like complete *** without hogging up server resources.

[MoonUnitBeta]

Kilo 40, on 01 December 2015 - 11:37 PM, said:

I'm pretty sure they fairly recently acknowledged the the absence of inverse kinematics, and gave a reason for why it's not in the game at the moment, but I'll be damned if I can remember where the post was, or why they said it's not in the game.

but yeah, I'd like to see it too.

The only one I remember where they mention that all the mechs are outfitted with everything is here:
where they have all the bells and whistles for when they "switch it on"

http://mwomercs.com/...12#entry4145712

feel free to put this link in your post, since it's on the front page, more people will see it.

Edited by MoonUnitBeta, 20 February 2016 - 02:40 AM.

[Buster Machine 0]

How much more latency would it take, it'll just take 3 or 2 points of data per leg. The X Y and Z axis for the knee, foot, and toe or just the foot and toe

[Aresye]

I still consider myself a noob with 3D work, but just from my small experience with rigging, IK is much more labor intensive, prone to joint-flipping with the slightest provocation, and very easy to completely break a rig. FK and IK are also not very friendly with each other. It takes a lot of custom scripting to get an effective IK/FK switch, which is even more prone to errors than a normal IK system.

MWO's servers would still have to keep track of each mech's root translation, so if the IK solve is server side, that's 3 separate points for each mech the server has to account for, compared to the 1 now. If the IK solve is client side, then ping and HSR issues come into play with discrepancies between the client and server, except instead of the mech's position being corrected for the client, you'd get a whole bunch of joint-flipping, which if severe enough, can actually fully break the IK solve and/or cause gimbal lock.

If you're confused by what I'm talking about, then download blender, find an IK rig, and see how everything works. You say things like, "Well if it's client side then it can easily be done," but you truly can't comprehend the complexities of IK systems until you've actually used and/or setup one yourself. There's a reason even with today's powerful game engines, that most game characters are built on FK rigs.

IK is great for animation, but it's just not quite there yet for fast-paced PvP gaming.

Dr Genocide, on 20 February 2016 - 03:01 AM, said:

How much more latency would it take, it'll just take 3 or 2 points of data per leg. The X Y and Z axis for the knee, foot, and toe or just the foot and toe

Pole Vectors and End Effectors, so at least 2 extra points of data per leg.

[Appogee]

Moonlight Grimoire, on 02 December 2015 - 12:21 AM, said:

I believe they did say it was something they wanted to go back to and put back into the game,

That is just the standard answer to any feature request in a Town Hall.

One Day™

Most of our current generation PCs would have the processing power necessary to perform the correct positioning of feet relative to terrain. I agree that the absence of this makes MWO sometimes feel unnecessary outdated. I'd like to see it reintroduced.

Dr Genocide, on 20 February 2016 - 03:01 AM, said:

How much more latency would it take, it'll just take 3 or 2 points of data per leg. The X Y and Z axis for the knee, foot, and toe or just the foot and toe

I would think it could be calculated client side. The client already knows the precise position of the terrain under every Mech foot, and could calculate the angles from the map data already in memory.

[SirSlaughter]

You simple don't get it

Inverse cinematics is lostech

[TexAce]

Here is an idea:

Since all mechs are being built with IK in mind and its just a switch on the server, why not turn it on for 1vs1, 2vs2, 4vs4 or 8vs8 matches? Server load is 33% less in 8vs8 in comparison to 12vs12 so the headroom would be sufficent to use IK.

[Nightmare1]

I wouldn't mind seeing this return.

Of course, I'd also like to see reciprocating Hill Climb (i.e. - downhill = faster travel) as well as my tearfully missed and long-awaited collisions and knock-downs.