Feature: Animation changes #95
Conversation
Recent versions of three use an animation system that fits better with how WoW stores animation data. It's still helpful to split animation tracks apart, but it's no longer necessary to create keyframe objects for each individual keyframe.
Newer versions of three can work directly with arrays of values, including for complex types like vectors and quaternions. Since parsing dense content like animations into full-blown objects is rather overkill, we can now start to move away from that approach.
| @@ -0,0 +1,15 @@ | |||
| import r from 'restructure'; | |||
|
|
|||
| class Color16 { | |||
There was a problem hiding this comment.
It'd probably be okay to call this something like SignedFixed16 and read from the stream as an int16le rather than uint16le, but I find it weird to do that for values that can't be negative (like transparency and color values).
Hence... Color16 -- your one-stop shop for values between 0.0 and 1.0 stored as shorts with values between 0 and 32767.
|
This PR will break Wowser until wowserhq/wowser#145 is merged. |
| @@ -0,0 +1,3 @@ | |||
| import r from 'restructure'; | |||
|
|
|||
| export default new r.Array(r.floatle, 3); | |||
There was a problem hiding this comment.
Isn't this the same as float3? I'd be totally in favour of normalizing all of these to make things more clear 👍
There was a problem hiding this comment.
My thought was that we should have a consistent naming scheme, and float3 would conflict with things like float16 (for a single fixed-point 16-bit float).
Since float3 is a vector made up of 32-bit floats, I thought maybe we should move toward vec3float32 as a more consistent name.
If you're cool with that, I can search and replace all float3s to vec3float32s.
There was a problem hiding this comment.
How would we differentiate between an array of three float values – currently float3, vec3float32 – and an object with x, y, z properties – currently Vec3Float?
There was a problem hiding this comment.
Recap from Gitter: array types now use <type>array<len> as a naming convention, like so:
float32array3 -> [1, 2, 3]
This should resolve naming confusion with types that involve named properties.
* Removed all uses of float2 and float3 * Renamed array vector types to <type>array<len> convention
|
🎉 Awesome! |
three's new animation system makes it quite a bit easier to work with the data present in M2 animations. This PR brings Blizzardry more in line with what three expects, and produces somewhat leaner output-- particularly when serializing and deserializing.