Ce site est optimisé pour être consulté depuis un navigateur moderne dans lequel JavaScript est activé.

World rotation and scale values of attachment

jdtec86

I'm attempting some normal map lighting on my Spine character and I'm very close to having it working.

I have a couple of niggling issues which boil down to me not getting the correct 'final/world' rotation and or scale values for some attachments.

My code is as follows:

float
boneRotation = slot->bone->worldRotation,
attachmentRotation = attachment->rotation,
combinedRotation = boneRotation + attachmentRotation,
rotation = rad (combinedRotation);
                float
                    scaleX = slot->bone->worldScaleX * attachment->scaleX,
                    scaleY = slot->bone->worldScaleY * attachment->scaleY;
                
                for (int i = 0; i < Rect2D::NUMBER_OF_VERTICES; i ++)
                {
                    vertices [i].SetNormalMapValues (rotation, scaleX, scaleY);
                }[/quote]
This works most of the time but if I have a bone with scaleX -1.0, rotation 126. And a slot with rotation 126 my global rotation comes out to 252 - which is wrong - it should be zero as the scaleX -1.0 on the bone scaleX reverses the bone rotation which means you have BoneRot: -126 and AttachmentRot: 126 -> 0.0 (as appears in Spine and in spine-c runtime).

Thanks
Jon

BinaryCats

I think you might be running into a yIsDown problem on your bone I had an issue with it here: World Translation Of an animation

if you open up your json file and look at the bone in the setup pose skeleton you will see the bone rotation is either -180 or 180 (assuming the rotation of the bone in spine is 0). You need to take the bones initial rotation into account. Spine doesn't tell the full story, some of it data it shows you is different to the actual data.

edit

A Quick way to test this is to (in spine) select parent space, and select the bone in question if the green down arrow is pointing down, the bone is 'flipped' to start with...

you can not alter this, if you move the bone out the hierarchy the 'flip' will move onto the next child bone(s). You also cant change it in the json and import the data.

jdtec86

Thanks for the response, BinaryCats.

I didn't actually find a yIsDown value anywhere in the runtime. The json source and spine project seemed to be as I would expect. Are you using something other than the c-runtime? There is a flipY value but that's disabled for all my stuff.

I think I've finally got this working with a bit more tinkering so will post for future people or incase someone spots a mistake that isn't manifesting on my Spine project.

Basically I reverse the rotation of the attachment if the parent bones x scale is negative. I have working normal maps!

float flipRotation = (slot->bone->worldScaleX > 0.0f) ? 1.0f : -1.0f;

                float
                    boneRotation = slot->bone->worldRotation,
                    attachmentRotation = attachment->rotation * flipRotation,
                    combinedRotation = boneRotation + attachmentRotation,
                    rotation = rad (combinedRotation);
                
                float
                    scaleX = slot->bone->worldScaleX * attachment->scaleX,
                    scaleY = slot->bone->worldScaleY * attachment->scaleY;
                
                scaleX = (scaleX > 0.0f) ? 1.0f : -1.0f;
                scaleY = (scaleY > 0.0f) ? 1.0f : -1.0f;
                
                for (int i = 0; i < Rect2D::NUMBER_OF_VERTICES; i ++)
                {
                    vertices [i].SetNormalMapValues (rotation, scaleX, scaleY);
                }[/quote]

BinaryCats

Hello,

In Bone.cpp there is a function called spBone_isYDown it is used in spIkConstraint_apply1 (I haven't done much digging to find out how it works/how to use it, sorry)

Its good that you got your stuff working, however, bare in mind that with spine 3.0 flipx/y is being removed in favour for negative scale. what repercussions this will have on your project/code is unknown.

jdtec86

That's interesting. I'd only noticed the flipx/y recently and figured there was some reason to favour using it over simply flipping to a negative scale like I currently do for duplicate mirrored limbs. If it's being removed I won't have to worry about it after all.

I'll remember the spBone_isYDown function incase anything odd pops up in the future.

BinaryCats

Yeah, FlipX/y is a good feature! However, it does have one major draw back, the bones are flipped locally. meaning the angles in relative to the parent will change which as a result would break ik's.

http://esotericsoftware.com/spine-changelog spine 3.0 might be a while off yet 🙁 apparently