Sims 3:0x01D0E723

Vertex Format - VRTF
TypeID:	0x01D0E723
Game Version:	The Sims 3

Format

This is a Sims 3:RCOL chunk found embedded in MODL and MLOD resources.

DWORD Tag // 'VRTF'
DWORD Version // 00000002
DWORD Stride // Vertex data length (Bytes per vertex)
DWORD Count
DWORD IsExtendedFormat// boolean, always false but were it true the byte fields ahead would be DWORDs

--repetition Count packed D3DVERTEXELEMENT:
	BYTE Usage
	BYTE UsageIndex //some elements can appear more than once, each time the index increases
	BYTE Format
	BYTE Offset // Offset into the vertex declaration for the data

Element Usages

Usages, and the formats they are usually paired with.

Id	Type	Element Formats
0x00	Position	Short4, Float3
0x01	Normal	ColorUByte4
0x02	UV	Short2, Short4, Float2, Float4
0x03	BlendIndex	UByte4
0x04	BlendWeight	ColorUByte4
0x05	Tangent	ColorUByte4
0x06	Color	ColorUByte4

Element Formats

List of formats, and how to read them

Id	Type		SwizzleCommands
0x00	Float	1 float	1x(Swizzle32)
0x01	Float2	2 floats	2x(Swizzle32)
0x02	Float3	3 floats	3x(Swizzle32)
0x03	Float4	4 floats	4x(Swizzle32)
0x04	UByte4	4 bytes	1x(Swizzle32)
0x05	ColorUByte4	3 floats in 4 bytes. Order is little-endian, use the low order byte (4th) as an unsigned scaler value. For each remaining signed byte, if negative, add it, if positive, subtract it, then multiple the result by 1/scaler, or by 1/127 if scaler is 0 This entry needs explaining more clearly. In Little-endian, the low order byte is first, not fourth. The statements "add it" and "subtract it" do not explain what "it" is. Here and subsequently: "multiply by 1/xxx" - why not just "divide by xxx"? Pljones 23:07, 17 January 2011 (UTC) OK, here's my explanation (this generates values that match Wes Howe's tool fairly well): There are two methods, a general one and one for BlendWeight only, that are slightly different. The order of the three floats is the reverse order of the first three bytes, with the fourth byte being a scaling factor. Generate the scaling factor by subtracting the fourth byte from 255; if the result is zero, use 128 (or, for BlendWeight 256) instead. Each float is generated by taking each byte, subtracting 128 (except if BlendWeight) and then dividing by the scaling factor. Pljones 22:08, 18 January 2011 (UTC)	1x(Swizzle32)
0x06	Short2	2 floats in two words. Non-UV usage: Divide by 32767. UV usage: See below.	1x(Swizzle16x2)
0x07	Short4	3 floats in four words. Non-UV usage: The fourth word is an unsigned scalar (or use 32767 if zero). Divide by this scalar. UV usage: See below.	2x(Swizzle16x2)
0x08	UByte4N		1x(Swizzle32)
0x09	Short2N		1x(Swizzle16x2)
0x0A	Short4N		2x(Swizzle16x2)
0x0B	UShort2N		1x(Swizzle16x2)
0x0C	UShort4N	3 floats in four words. Non-UV usage: The fourth word is an unsigned scalar (or use 512 if zero). Divide by this scalar. UV usage: See below.	2x(Swizzle16x2)
0x0D	Dec3N		1x(Swizzle32)
0x0E	UDec3N		1x(Swizzle32)
0x0F	Float16_2		1x(Swizzle16x2)
0x10	Float16_4		2x(Swizzle16x2)

For UV usage, multiply by the MATD ShaderData entry for UVScales, element 0. If no such value exists, divide by 32767.

Based on contributions from delphy, karybdis and atavera