mirror/AssetStudio
1
0
Fork 0
mirror of https://github.com/aelurum/AssetStudio.git synced 2025-05-27 22:00:23 -04:00
Code Issues Packages Projects Releases Wiki Activity

more complete CompressedMesh (m_Skin, m_Normals, m_Tangents)

Browse Source
This commit is contained in:
Perfare 2018-09-08 23:14:18 +08:00
parent c9a98d7163
commit 14ad9431b9
2 changed files with 140 additions and 547 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
AssetStudio/Classes/AnimationClip.cs
View File

@ -103,7 +103,7 @@ namespace AssetStudio
if (numChunks == -1)
numChunks = (int)m_NumItems / itemCountInChunk;
var end = chunkStride * numChunks / 4;
var data = new float[end];
var data = new List<float>();
for (var index = 0; index != end; index += chunkStride / 4)
{
for (int i = 0; i < itemCountInChunk; ++i)
@ -124,11 +124,11 @@ namespace AssetStudio
}
}
x &= (uint)(1 << m_BitSize) - 1u;
data[index + i] = x / (scale * ((1 << m_BitSize) - 1)) + m_Start;
data.Add(x / (scale * ((1 << m_BitSize) - 1)) + m_Start);
}
}
return data;
return data.ToArray();
}
}

681
AssetStudio/Classes/Mesh.cs
View File

@ -1,104 +1,8 @@
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.IO;
using System.Collections;
using SharpDX;
/*Notes about handedness
Converting from left-handed to right-handed and vice versa requires either:
a) swapping any 2 axes
b) flipping any 1 axis
c) any odd combinations of a&b
An even number of combinations will result in the same handedness, but in a different perspective.
Also, rotating the axes is just that, the same handedness in a different system.
Y-up or Z-up are not requirements OR defining characteristics of a handedness, they are just common that way.
unt is left-handed with Y-up
Aircraft View
Y Top Y Up
| |
| Z Nose |
| / |
| / |
+--------X X--------+
Right to the \
left \
Z towards viewer
3ds Max is right-handed with Z up
Aircraft View
Z Top Z Up (Viewcube Top)
| |
| (-Y) Nose away Y |
| / from \ |
| / viewer \ |
+--------(-X) (VC Back) +--------X to the right
Right (Viewcube Right)
FBX and Maya are both right-handed but with Y up! (90 degree rotation on X = samme handedness)
Aircraft View
Y Top Y Up (Viewcube Top)
| |
| Z Nose |
| / |
| / |
+--------(-X) +--------X to the right
Right \ (Viewcube Right)
\
Z towards viewer
(Viewcube Front)
Exporting FBX from Max, vertex components are ALWAYS written as they are in max: X Y Z.
The Axis option only affects GlobalSettings and PreRotation in the FBX file.
"Z-up" option:
UpAxis=2,Sign=1 <=> ViewCube Up = FBX vertex[2] <=> Max Z = FBX Z <=> Maya Y = FBX Z
FrontAxis=1,Sign=-1 <=> ViewCube -Front = FBX v[1] <=> Max -(-Y) = FBX Y <=> Maya -Z = FBX Y
CoordAxis=0,Sign=1 <=> ViewCube Right = FBX v[0] <=> Max X = FBX X <=> Maya X = FBX X
no PreRotation
"Y-up" option:
UpAxis=1,Sign=1 <=> ViewCube Up = FBX vertex[1] <=> Max Z = FBX Y <=> Maya Y = FBX Y
FrontAxis=2,Sign=1 <=> ViewCube Front = FBX v[2] <=> Max -Y = FBX Z <=> Maya Z = FBX Z
CoordAxis=0,Sign=1 <=> ViewCube Right = FBX v[0] <=> Max X = FBX X <=> Maya X = FBX X
PreRotation -90,0,0 to bring Original Up (FBX Z) to ViewCube Up when importing
PreRotation means only the geometry is rotated locally around pivot before applying other modifiers. It is "invisible" to the user.
Importing FBX in Unt, Axis settings and PreRotations are ignored.
They probably ignore them because the order of vertex components is always the same in FBX, and Unt axes never change orientation (as opposed to Max-Maya).
Vertex components are loaded as follows:
Unt Up(Y) = FBX Y
Unt Front(Z) = FBX Z
Unt Left(X) = FBX -X
Technically, this is a correct handedness conversion, but to a different system, because the model is not properly oriented (plane nose is down).
So Unt adds a -90 degree rotation, similar to the FBX PreRotation, to bring the nose to Front(Z).
Except it does it as a regular rotation, and combines it with any other rotations in the Transform asset.
Converting from Unt back to FBX, the same vertex conversion cannot be applied because we have to take into account the rotation.
Option 0: export vertices and transformations as -X,Y,Z and set FBX option to Y-up without PreRotation!
the result will be Max Z = FBX Y, Max -Y = FBX Z, Max X = FBX X => final order -X -Z Y
Option 1: export vertices and transformations as -X,-Z,Y and set FBX options as "Z-up".
The -90 rotation exported from Unt will bring the model in correct orientation.
Option 2: export vertices and transformations as -X,-Y,-Z, add -90 PreRotation to every Mesh Node and set FBX options as "Y-up".
The -90 rotation from Unt plus the -90 PreRotation will bring the model in correct orientation.
Remember though that the PreRotation is baked into the Geometry.
But since the -90 rotation from Unt is a regular type, it will show up in the modifier in both cases.
Also, re-importing this FBX in Unt will now produce a (-90)+(-90)=-180 rotation.
This is an unfortunate eyesore, but nothing more, the orientation will be fine.
In theory, one could add +90 degrees rotation to GameObjects that link to Mesh assets (but not other types) to cancel out the Unt rotation.
The problem is you can never know where the Unt mesh originated from. If it came from a left-handed format such as OBJ, there wouldn't have been any conversion and it wouldn't have that -90 degree adjustment.
So it would "fix" meshes that were originally sourced form FBX, but would still have the "extra" rotation in mehses sourced from left-handed formats.
*/
namespace AssetStudio
{
public class Mesh
@ -112,7 +16,6 @@ namespace AssetStudio
private ChannelInfo[] m_Channels;
private StreamInfo[] m_Streams;
public List<BoneInfluence>[] m_Skin;
//public Dictionary<int, float>[] m_Skin;
public float[][,] m_BindPose;
public int m_VertexCount;
public float[] m_Vertices;
@ -160,15 +63,6 @@ namespace AssetStudio
public ushort frequency;
}
public class PackedBitVector
{
public int m_NumItems;
public float m_Range = 1.0f;
public float m_Start;
public byte[] m_Data;
public byte m_BitSize;
}
public class BlendShapeData
{
public class BlendShapeVertex
@ -278,156 +172,19 @@ namespace AssetStudio
return result;
}
private static uint[] UnpackBitVector(PackedBitVector pakData)
{
uint[] unpackedVectors = new uint[pakData.m_NumItems];
//int bitmax = 0;//used to convert int value to float
//for (int b = 0; b < pakData.m_BitSize; b++) { bitmax |= (1 << b); }
//the lazy way
//split data into groups of "aligned" bytes i.e. 8 packed values per group
//I could calculate optimized group size based on BitSize, but this is the lazy way
if (pakData.m_BitSize == 0)
{
pakData.m_BitSize = (byte)((pakData.m_Data.Length * 8) / pakData.m_NumItems);
//don't know, don't care
}
int groupSize = pakData.m_BitSize; //bitSize * 8 values / 8 bits
byte[] group = new byte[groupSize];
int groupCount = pakData.m_NumItems / 8;
for (int g = 0; g < groupCount; g++)
{
Buffer.BlockCopy(pakData.m_Data, g * groupSize, group, 0, groupSize);
BitArray groupBits = new BitArray(group);
for (int v = 0; v < 8; v++)
{
BitArray valueBits = new BitArray(new Boolean[pakData.m_BitSize]);
for (int b = 0; b < pakData.m_BitSize; b++)
{
valueBits.Set(b, groupBits.Get(b + v * pakData.m_BitSize));
}
var valueArr = new int[1];
valueBits.CopyTo(valueArr, 0);
//unpackedVectors[v + g * 8] = (float)(valueArr[0] / bitmax) * pakData.m_Range + pakData.m_Start;
//valueBits.CopyTo(unpackedVectors, v + g * 8);//doesn't work with uint[]
unpackedVectors[v + g * 8] = (uint)valueArr[0];
}
}
//m_NumItems is not necessarily a multiple of 8, so there can be one extra group with fewer values
int endBytes = pakData.m_Data.Length - groupCount * groupSize;
int endVal = pakData.m_NumItems - groupCount * 8;
if (endBytes > 0)
{
Buffer.BlockCopy(pakData.m_Data, groupCount * groupSize, group, 0, endBytes);
BitArray groupBits = new BitArray(group);
for (int v = 0; v < endVal; v++)
{
BitArray valueBits = new BitArray(new Boolean[pakData.m_BitSize]);
for (int b = 0; b < pakData.m_BitSize; b++)
{
valueBits.Set(b, groupBits.Get(b + v * pakData.m_BitSize));
}
var valueArr = new int[1];
valueBits.CopyTo(valueArr, 0);
//unpackedVectors[v + groupCount * 8] = (float)(valueArr[0] / bitmax) * pakData.m_Range + pakData.m_Start;
//valueBits.CopyTo(unpackedVectors, v + groupCount * 8);
unpackedVectors[v + groupCount * 8] = (uint)valueArr[0];
}
}
//the hard way
//compute bit position in m_Data for each value
/*byte[] value = new byte[4] { 0, 0, 0, 0 };
int byteCount = pakData.m_BitSize / 8;//bytes in single value
int bitCount = pakData.m_BitSize % 8;
for (int v = 0; v < pakData.m_NumItems; v++)
{
if ((bitCount * v) % 8 == 0) //bitstream is "aligned"
{//does this make sense if I'm gonna compute unaligned anywhay?
for (int b = 0; b < byteCount; b++)
{
value[b] = pakData.m_Data[b + v * (byteCount+1)];
}
if (byteCount < 4) //shouldn't it be always?
{
byte lastByte = pakData.m_Data[bitCount * v / 8];
for (int b = 0; b < bitCount; b++)//no
{
//set bit in val[byteCount+1]
}
}
}
else
{
//god knows
}
unpackedVectors[v] = BitConverter.ToSingle(value, 0);
}*/
//first I split the data into byte-aligned arrays
//too complicated to calculate group size each time
//then no point in dividing?
/*int groupSize = byteCount + (bitCount + 7)/8;
int groups = pakData.m_Data.Length / groupSize;
int valPerGr = (int)(pakData.m_NumItems / groups);
byte[] group = new byte[groupSize];
for (int g = 0; g < groups; g++)
{
Buffer.BlockCopy(pakData.m_Data, g * groupSize, group, 0, groupSize);
for (int v = 0; v < valPerGr; v++)
{
unpackedVectors[v + g * valPerGr] = BitConverter.ToSingle(value, 0);
}
}
//m_Data size is not necessarily a multiple of align, so there can be one extra group with fewer values
int lastBytes = pakData.m_Data.Length % groupSize;
int lastVal = (int)(pakData.m_NumItems - groups * valPerGr);
if (lastBytes > 0)
{
Buffer.BlockCopy(pakData.m_Data, groups * groupSize, group, 0, lastBytes);
for (int v = 0; v < lastVal; v++)
{
unpackedVectors[v + groups * valPerGr] = BitConverter.ToSingle(value, 0);
}
}*/
return unpackedVectors;
}
private static int GetChannelFormatLength(int format)
private static int GetChannelFormatSize(int format)
{
switch (format)
{
case 0: //float
case 0: //kChannelFormatFloat
return 4;
case 1: //half float
case 1: //kChannelFormatFloat16
return 2;
case 2: //byte float
case 2: //kChannelFormatColor
return 1;//TODO actual is 4
case 3: //kChannelFormatByte
return 1;
case 11: //int
case 11: //kChannelFormatInt
return 4;
default:
return 0;
@ -829,7 +586,7 @@ namespace AssetStudio
if (m_Channel.dimension > 0)
{
chnMask |= 1 << chn;
stride += m_Channel.dimension * GetChannelFormatLength(m_Channel.format);
stride += m_Channel.dimension * GetChannelFormatSize(m_Channel.format);
}
}
}
@ -877,7 +634,7 @@ namespace AssetStudio
}
int[] componentsIntArray = null;
componentByteSize = GetChannelFormatLength(m_Channel.format);
componentByteSize = GetChannelFormatSize(m_Channel.format);
componentBytes = new byte[m_VertexCount * m_Channel.dimension * componentByteSize];
for (int v = 0; v < m_VertexCount; v++)
@ -1038,198 +795,120 @@ namespace AssetStudio
#region Compressed Mesh data for 2.6.0 and later - 160 bytes
if (version[0] >= 3 || (version[0] == 2 && version[1] >= 6))
{
//remember there can be combinations of packed and regular vertex properties
#region m_Vertices
PackedBitVector m_Vertices_Packed = new PackedBitVector();
m_Vertices_Packed.m_NumItems = reader.ReadInt32();
m_Vertices_Packed.m_Range = reader.ReadSingle();
m_Vertices_Packed.m_Start = reader.ReadSingle();
m_Vertices_Packed.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_Vertices_Packed.m_Data, 0, m_Vertices_Packed.m_Data.Length);
reader.AlignStream(4);
m_Vertices_Packed.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
#region m_Vertices;
var m_Vertices_Packed = new PackedFloatVector(reader);
if (m_Vertices_Packed.m_NumItems > 0)
{
m_VertexCount = m_Vertices_Packed.m_NumItems / 3;
uint[] m_Vertices_Unpacked = UnpackBitVector(m_Vertices_Packed);
int bitmax = 0;//used to convert int value to float
for (int b = 0; b < m_Vertices_Packed.m_BitSize; b++) { bitmax |= (1 << b); }
m_Vertices = new float[m_Vertices_Packed.m_NumItems];
for (int v = 0; v < m_Vertices_Packed.m_NumItems; v++)
{
m_Vertices[v] = (float)((double)m_Vertices_Unpacked[v] / bitmax) * m_Vertices_Packed.m_Range + m_Vertices_Packed.m_Start;
}
m_VertexCount = (int)m_Vertices_Packed.m_NumItems / 3;
m_Vertices = m_Vertices_Packed.UnpackFloats(3, 4);
}
#endregion
#region m_UV
PackedBitVector m_UV_Packed = new PackedBitVector(); //contains all channels
m_UV_Packed.m_NumItems = reader.ReadInt32();
m_UV_Packed.m_Range = reader.ReadSingle();
m_UV_Packed.m_Start = reader.ReadSingle();
m_UV_Packed.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_UV_Packed.m_Data, 0, m_UV_Packed.m_Data.Length);
reader.AlignStream(4);
m_UV_Packed.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
if (m_UV_Packed.m_NumItems > 0 && (bool)Properties.Settings.Default["exportUVs"])
var m_UV_Packed = new PackedFloatVector(reader);
if (m_UV_Packed.m_NumItems > 0)
{
uint[] m_UV_Unpacked = UnpackBitVector(m_UV_Packed);
int bitmax = 0;
for (int b = 0; b < m_UV_Packed.m_BitSize; b++) { bitmax |= (1 << b); }
m_UV1 = new float[m_VertexCount * 2];
for (int v = 0; v < m_VertexCount * 2; v++)
{
m_UV1[v] = (float)((double)m_UV_Unpacked[v] / bitmax) * m_UV_Packed.m_Range + m_UV_Packed.m_Start;
}
m_UV1 = m_UV_Packed.UnpackFloats(2, 4, 0, m_VertexCount);
if (m_UV_Packed.m_NumItems >= m_VertexCount * 4)
{
m_UV2 = new float[m_VertexCount * 2];
for (uint v = 0; v < m_VertexCount * 2; v++)
{
m_UV2[v] = (float)((double)m_UV_Unpacked[v + m_VertexCount * 2] / bitmax) * m_UV_Packed.m_Range + m_UV_Packed.m_Start;
}
if (m_UV_Packed.m_NumItems >= m_VertexCount * 6)
{
m_UV3 = new float[m_VertexCount * 2];
for (uint v = 0; v < m_VertexCount * 2; v++)
{
m_UV3[v] = (float)((double)m_UV_Unpacked[v + m_VertexCount * 4] / bitmax) * m_UV_Packed.m_Range + m_UV_Packed.m_Start;
}
if (m_UV_Packed.m_NumItems == m_VertexCount * 8)
{
m_UV4 = new float[m_VertexCount * 2];
for (uint v = 0; v < m_VertexCount * 2; v++)
{
m_UV4[v] = (float)((double)m_UV_Unpacked[v + m_VertexCount * 6] / bitmax) * m_UV_Packed.m_Range + m_UV_Packed.m_Start;
}
}
}
m_UV2 = m_UV_Packed.UnpackFloats(2, 4, m_VertexCount * 2, m_VertexCount);
}
if (m_UV_Packed.m_NumItems >= m_VertexCount * 6)
{
m_UV3 = m_UV_Packed.UnpackFloats(2, 4, m_VertexCount * 4, m_VertexCount);
}
if (m_UV_Packed.m_NumItems >= m_VertexCount * 8)
{
m_UV4 = m_UV_Packed.UnpackFloats(2, 4, m_VertexCount * 6, m_VertexCount);
}
}
#endregion
#region m_BindPose
#region m_BindPoses
if (version[0] < 5)
{
PackedBitVector m_BindPoses_Packed = new PackedBitVector();
m_BindPoses_Packed.m_NumItems = reader.ReadInt32();
m_BindPoses_Packed.m_Range = reader.ReadSingle();
m_BindPoses_Packed.m_Start = reader.ReadSingle();
m_BindPoses_Packed.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_BindPoses_Packed.m_Data, 0, m_BindPoses_Packed.m_Data.Length);
reader.AlignStream(4);
m_BindPoses_Packed.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
if (m_BindPoses_Packed.m_NumItems > 0 && (bool)Properties.Settings.Default["exportDeformers"])
var m_BindPoses_Packed = new PackedFloatVector(reader);
if (m_BindPoses_Packed.m_NumItems > 0)
{
uint[] m_BindPoses_Unpacked = UnpackBitVector(m_BindPoses_Packed);
int bitmax = 0;//used to convert int value to float
for (int b = 0; b < m_BindPoses_Packed.m_BitSize; b++) { bitmax |= (1 << b); }
m_BindPose = new float[m_BindPoses_Packed.m_NumItems / 16][,];
for (int i = 0; i < m_BindPose.Length; i++)
{
m_BindPose[i] = new float[4, 4];
for (int j = 0; j < 4; j++)
{
for (int k = 0; k < 4; k++)
{
m_BindPose[i][j, k] = (float)((double)m_BindPoses_Unpacked[i * 16 + j * 4 + k] / bitmax) * m_BindPoses_Packed.m_Range + m_BindPoses_Packed.m_Start;
}
}
}
var m_BindPoses_Unpacked = m_BindPoses_Packed.UnpackFloats(16, 4 * 16);
throw new NotImplementedException();
}
}
#endregion
PackedBitVector m_Normals_Packed = new PackedBitVector();
m_Normals_Packed.m_NumItems = reader.ReadInt32();
m_Normals_Packed.m_Range = reader.ReadSingle();
m_Normals_Packed.m_Start = reader.ReadSingle();
m_Normals_Packed.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_Normals_Packed.m_Data, 0, m_Normals_Packed.m_Data.Length);
reader.AlignStream(4);
m_Normals_Packed.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
var m_Normals_Packed = new PackedFloatVector(reader);
PackedBitVector m_Tangents_Packed = new PackedBitVector();
m_Tangents_Packed.m_NumItems = reader.ReadInt32();
m_Tangents_Packed.m_Range = reader.ReadSingle();
m_Tangents_Packed.m_Start = reader.ReadSingle();
m_Tangents_Packed.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_Tangents_Packed.m_Data, 0, m_Tangents_Packed.m_Data.Length);
reader.AlignStream(4);
m_Tangents_Packed.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
var m_Tangents_Packed = new PackedFloatVector(reader);
PackedBitVector m_Weights = new PackedBitVector();
m_Weights.m_NumItems = reader.ReadInt32();
m_Weights.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_Weights.m_Data, 0, m_Weights.m_Data.Length);
reader.AlignStream(4);
m_Weights.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
var m_Weights = new PackedIntVector(reader);
#region m_Normals
PackedBitVector m_NormalSigns_packed = new PackedBitVector();
m_NormalSigns_packed.m_NumItems = reader.ReadInt32();
m_NormalSigns_packed.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_NormalSigns_packed.m_Data, 0, m_NormalSigns_packed.m_Data.Length);
reader.AlignStream(4);
m_NormalSigns_packed.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
if (m_Normals_Packed.m_NumItems > 0 && (bool)Properties.Settings.Default["exportNormals"])
var m_NormalSigns = new PackedIntVector(reader);
if (m_Normals_Packed.m_NumItems > 0)
{
uint[] m_Normals_Unpacked = UnpackBitVector(m_Normals_Packed);
uint[] m_NormalSigns = UnpackBitVector(m_NormalSigns_packed);
int bitmax = 0;
for (int b = 0; b < m_Normals_Packed.m_BitSize; b++) { bitmax |= (1 << b); }
var normalData = m_Normals_Packed.UnpackFloats(2, 4 * 2);
var signs = m_NormalSigns.UnpackInts();
m_Normals = new float[m_Normals_Packed.m_NumItems / 2 * 3];
for (int v = 0; v < m_Normals_Packed.m_NumItems / 2; v++)
for (int i = 0; i < m_Normals_Packed.m_NumItems / 2; ++i)
{
m_Normals[v * 3] = (float)((double)m_Normals_Unpacked[v * 2] / bitmax) * m_Normals_Packed.m_Range + m_Normals_Packed.m_Start;
m_Normals[v * 3 + 1] = (float)((double)m_Normals_Unpacked[v * 2 + 1] / bitmax) * m_Normals_Packed.m_Range + m_Normals_Packed.m_Start;
m_Normals[v * 3 + 2] = (float)Math.Sqrt(1 - m_Normals[v * 3] * m_Normals[v * 3] - m_Normals[v * 3 + 1] * m_Normals[v * 3 + 1]);
if (m_NormalSigns[v] == 0) { m_Normals[v * 3 + 2] *= -1; }
var x = normalData[i * 2 + 0];
var y = normalData[i * 2 + 1];
m_Normals[i * 3] = x;
m_Normals[i * 3 + 1] = y;
var zsqr = 1 - x * x - y * y;
float z;
if (zsqr >= 0f)
z = (float)Math.Sqrt(zsqr);
else
{
z = 0;
var normal = new Vector3(x, y, z);
normal.Normalize();
x = normal.X;
y = normal.Y;
z = normal.Z;
}
if (signs[i] == 0)
z = -z;
m_Normals[i * 3] = x;
m_Normals[i * 3 + 1] = y;
m_Normals[i * 3 + 2] = z;
}
}
#endregion
#region m_Tangents
PackedBitVector m_TangentSigns_packed = new PackedBitVector();
m_TangentSigns_packed.m_NumItems = reader.ReadInt32();
m_TangentSigns_packed.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_TangentSigns_packed.m_Data, 0, m_TangentSigns_packed.m_Data.Length);
reader.AlignStream(4);
m_TangentSigns_packed.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
if (m_Tangents_Packed.m_NumItems > 0 && (bool)Properties.Settings.Default["exportTangents"])
var m_TangentSigns = new PackedIntVector(reader);
if (m_Tangents_Packed.m_NumItems > 0)
{
uint[] m_Tangents_Unpacked = UnpackBitVector(m_Tangents_Packed);
uint[] m_TangentSigns = UnpackBitVector(m_TangentSigns_packed);
int bitmax = 0;
for (int b = 0; b < m_Tangents_Packed.m_BitSize; b++) { bitmax |= (1 << b); }
m_Tangents = new float[m_Tangents_Packed.m_NumItems / 2 * 3];
for (int v = 0; v < m_Tangents_Packed.m_NumItems / 2; v++)
var tangentData = m_Tangents_Packed.UnpackFloats(2, 4 * 2);
var signs = m_TangentSigns.UnpackInts();
m_Tangents = new float[m_Tangents_Packed.m_NumItems / 2 * 4];
for (int i = 0; i < m_Tangents_Packed.m_NumItems / 2; ++i)
{
m_Tangents[v * 3] = (float)((double)m_Tangents_Unpacked[v * 2] / bitmax) * m_Tangents_Packed.m_Range + m_Tangents_Packed.m_Start;
m_Tangents[v * 3 + 1] = (float)((double)m_Tangents_Unpacked[v * 2 + 1] / bitmax) * m_Tangents_Packed.m_Range + m_Tangents_Packed.m_Start;
m_Tangents[v * 3 + 2] = (float)Math.Sqrt(1 - m_Tangents[v * 3] * m_Tangents[v * 3] - m_Tangents[v * 3 + 1] * m_Tangents[v * 3 + 1]);
if (m_TangentSigns[v] == 0) { m_Tangents[v * 3 + 2] *= -1; }
var x = tangentData[i * 2 + 0];
var y = tangentData[i * 2 + 1];
var zsqr = 1 - x * x - y * y;
float z;
if (zsqr >= 0f)
z = (float)Math.Sqrt(zsqr);
else
{
z = 0;
var vector3f = new Vector3(x, y, z);
vector3f.Normalize();
x = vector3f.X;
y = vector3f.Y;
z = vector3f.Z;
}
if (signs[i * 2 + 0] == 0)
z = -z;
var w = signs[i * 2 + 1] > 0 ? 1.0f : -1.0f;
m_Tangents[i * 4] = x;
m_Tangents[i * 4 + 1] = y;
m_Tangents[i * 4 + 2] = z;
m_Tangents[i * 4 + 3] = w;
}
}
#endregion
@ -1237,134 +916,71 @@ namespace AssetStudio
#region m_FloatColors
if (version[0] >= 5)
{
PackedBitVector m_FloatColors = new PackedBitVector();
m_FloatColors.m_NumItems = reader.ReadInt32();
m_FloatColors.m_Range = reader.ReadSingle();
m_FloatColors.m_Start = reader.ReadSingle();
m_FloatColors.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_FloatColors.m_Data, 0, m_FloatColors.m_Data.Length);
reader.AlignStream(4);
m_FloatColors.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
if (m_FloatColors.m_NumItems > 0 && (bool)Properties.Settings.Default["exportColors"])
var m_FloatColors = new PackedFloatVector(reader);
if (m_FloatColors.m_NumItems > 0)
{
uint[] m_FloatColors_Unpacked = UnpackBitVector(m_FloatColors);
int bitmax = 0;
for (int b = 0; b < m_FloatColors.m_BitSize; b++) { bitmax |= (1 << b); }
m_Colors = new float[m_FloatColors.m_NumItems];
for (int v = 0; v < m_FloatColors.m_NumItems; v++)
{
m_Colors[v] = (float)m_FloatColors_Unpacked[v] / bitmax * m_FloatColors.m_Range + m_FloatColors.m_Start;
}
throw new NotImplementedException();
}
}
#endregion
#region m_Skin
PackedBitVector m_BoneIndices = new PackedBitVector();
m_BoneIndices.m_NumItems = reader.ReadInt32();
m_BoneIndices.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_BoneIndices.m_Data, 0, m_BoneIndices.m_Data.Length);
reader.AlignStream(4);
m_BoneIndices.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
if (m_BoneIndices.m_NumItems > 0 && (bool)Properties.Settings.Default["exportDeformers"])
var m_BoneIndices = new PackedIntVector(reader);
if (m_Weights.m_NumItems > 0)
{
uint[] m_Weights_Unpacked = UnpackBitVector(m_Weights);
int bitmax = 0;
for (int b = 0; b < m_Weights.m_BitSize; b++) { bitmax |= (1 << b); }
var weights = m_Weights.UnpackInts();
var boneIndices = m_BoneIndices.UnpackInts();
uint[] m_BoneIndices_Unpacked = UnpackBitVector(m_BoneIndices);
InitMSkin();
m_Skin = new List<BoneInfluence>[m_VertexCount];
for (int s = 0; s < m_Skin.Length; s++)
int bonePos = 0;
int boneIndexPos = 0;
int j = 0;
int sum = 0;
for (int i = 0; i < m_Weights.m_NumItems; i++)
{
m_Skin[s] = new List<BoneInfluence>(4);
}
//read bone index and weight.
m_Skin[bonePos][j].weight = weights[i] / 31.0f;
m_Skin[bonePos][j].boneIndex = boneIndices[boneIndexPos++];
j++;
sum += weights[i];
int inflCount = m_Weights.m_NumItems;
int vertIndex = 0;
int weightIndex = 0;
int bonesIndex = 0;
for (weightIndex = 0; weightIndex < inflCount; vertIndex++)
{
int inflSum = 0;
int j;
for (j = 0; j < 4; j++)
//the weights add up to one. fill the rest for this vertex with zero, and continue with next one.
if (sum >= 31)
{
int curWeight = 0;
if (j == 3)
for (; j < 4; j++)
{
curWeight = 31 - inflSum;
}
else
{
curWeight = (int)m_Weights_Unpacked[weightIndex];
weightIndex++;
inflSum += curWeight;
}
double curWeightDouble = (double)curWeight;
float realCurWeight = (float)(curWeightDouble / bitmax);
int boneIndex = (int)m_BoneIndices_Unpacked[bonesIndex];
bonesIndex++;
if (boneIndex < 0)
{
throw new Exception($"Invalid bone index {boneIndex}");
}
BoneInfluence boneInfl = new BoneInfluence()
{
weight = realCurWeight,
boneIndex = boneIndex,
};
m_Skin[vertIndex].Add(boneInfl);
if (inflSum == 31)
{
break;
}
if (inflSum > 31)
{
throw new Exception("Influence sum " + inflSum + " greater than 31");
m_Skin[bonePos][j].weight = 0;
m_Skin[bonePos][j].boneIndex = 0;
}
bonePos++;
j = 0;
sum = 0;
}
for (; j < 4; j++)
//we read three weights, but they don't add up to one. calculate the fourth one, and read
//missing bone index. continue with next vertex.
else if (j == 3)
{
BoneInfluence boneInfl = new BoneInfluence()
{
weight = 0.0f,
boneIndex = 0,
};
m_Skin[vertIndex].Add(boneInfl);
m_Skin[bonePos][j].weight = (31 - sum) / 31.0f;
m_Skin[bonePos][j].boneIndex = boneIndices[boneIndexPos++];
bonePos++;
j = 0;
sum = 0;
}
}
bool isFine = vertIndex == m_VertexCount;
if (!isFine)
{
throw new Exception("Vertecies aint equals");
}
}
#endregion
PackedBitVector m_Triangles = new PackedBitVector();
m_Triangles.m_NumItems = reader.ReadInt32();
m_Triangles.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_Triangles.m_Data, 0, m_Triangles.m_Data.Length);
reader.AlignStream(4);
m_Triangles.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
if (m_Triangles.m_NumItems > 0) { m_IndexBuffer = UnpackBitVector(m_Triangles); }
#region m_IndexBuffer
var m_Triangles = new PackedIntVector(reader);
m_IndexBuffer = Array.ConvertAll(m_Triangles.UnpackInts(), x => (uint)x);
#endregion
}
#endregion
#region Colors & Collision triangles for 3.4.2 and earlier
if (version[0] <= 2 || (version[0] == 3 && version[1] <= 4)) //
if (version[0] <= 2 || (version[0] == 3 && version[1] <= 4))
{
reader.Position += 24; //Axis-Aligned Bounding Box
int m_Colors_size = reader.ReadInt32();
@ -1376,46 +992,23 @@ namespace AssetStudio
int m_CollisionVertexCount = reader.ReadInt32();
}
#endregion
#region Compressed colors & Local AABB for 3.5.0 to 4.x.x
else //vertex colors are either in streams or packed bits
#region Compressed colors
else
{
if (version[0] < 5)
{
PackedBitVector m_Colors_Packed = new PackedBitVector();
m_Colors_Packed.m_NumItems = reader.ReadInt32();
m_Colors_Packed.m_Data = new byte[reader.ReadInt32()];
reader.Read(m_Colors_Packed.m_Data, 0, m_Colors_Packed.m_Data.Length);
reader.AlignStream(4);
m_Colors_Packed.m_BitSize = reader.ReadByte();
reader.Position += 3; //4 byte alignment
var m_Colors_Packed = new PackedIntVector(reader);
if (m_Colors_Packed.m_NumItems > 0)
{
if (m_Colors_Packed.m_BitSize == 32)
{
//4 x 8bit color channels
m_Colors = new float[m_Colors_Packed.m_Data.Length];
for (int v = 0; v < m_Colors_Packed.m_Data.Length; v++)
{
m_Colors[v] = (float)m_Colors_Packed.m_Data[v] / 0xFF;
}
}
else //not tested
{
uint[] m_Colors_Unpacked = UnpackBitVector(m_Colors_Packed);
int bitmax = 0;//used to convert int value to float
for (int b = 0; b < m_Colors_Packed.m_BitSize; b++) { bitmax |= (1 << b); }
m_Colors = new float[m_Colors_Packed.m_NumItems];
for (int v = 0; v < m_Colors_Packed.m_NumItems; v++)
{
m_Colors[v] = (float)m_Colors_Unpacked[v] / bitmax;
}
}
throw new NotImplementedException();
}
}
else { uint m_UVInfo = reader.ReadUInt32(); }
else
{
var m_UVInfo = reader.ReadUInt32();
}
reader.Position += 24; //Axis-Aligned Bounding Box
reader.Position += 24; //AABB m_LocalAABB
}
#endregion