C++ API Reference
ddsFloatReader.cpp
//-
// ==========================================================================
// Copyright 1995,2006,2008 Autodesk, Inc. All rights reserved.
//
// Use of this software is subject to the terms of the Autodesk
// license agreement provided at the time of installation or download,
// or which otherwise accompanies this software in either electronic
// or hard copy form.
// ==========================================================================
//+
// DESCRIPTION:
//
// Demonstrates how to write a .dds float file reader using the API.
// In Maya's image reading menu dialogs, you can select *.* to see all images, and
// then retrieve a dds extension file item to load the .dds file into Maya.
//
#include <maya/MPxImageFile.h>
#include <maya/MImageFileInfo.h>
#include <maya/MImage.h>
#include <maya/MFnPlugin.h>
#include <maya/MStringArray.h>
#include <maya/MIOStream.h>
#if _WIN32
#pragma warning( disable : 4290 ) // Disable STL warnings.
#endif
#include "ddsFloatReader.h"
#include <math.h>
using namespace dds_Float_Reader;
MString kImageFormatName( "DDS Float");
class ddsFloatReader : public MPxImageFile
{
public:
ddsFloatReader();
~ddsFloatReader() override;
static void* creator();
MStatus open( MString pathname, MImageFileInfo* info) override;
MStatus load( MImage& image, unsigned int idx) override;
MStatus close() override;
protected:
// Data members
unsigned int fWidth;
unsigned int fHeight;
unsigned int fNumChannels;
unsigned int fBytesPerPixel;
// File and header description
FILE *fInputFile;
DDS_HEADER fHeader;
};
//
// DESCRIPTION:
ddsFloatReader::ddsFloatReader()
: fWidth(0),
fHeight(0),
fNumChannels(0),
fBytesPerPixel(0),
fInputFile(NULL)
{
}
//
// DESCRIPTION:
ddsFloatReader::~ddsFloatReader()
{
close();
}
//
// DESCRIPTION:
/* virtual */
MStatus ddsFloatReader::close()
{
fWidth = 0;
fHeight = 0;
fNumChannels = 0;
fBytesPerPixel = 0;
// Close our file
if (fInputFile != NULL)
fclose(fInputFile);
fInputFile = NULL;
return MS::kSuccess;
}
//
// DESCRIPTION:
void * ddsFloatReader::creator()
{
return new ddsFloatReader();
}
// Need to swap byte order on Mac
//
inline void swap_endian(void *val)
{
#if defined(OSMac_)
unsigned int *ival = (unsigned int *)val;
*ival = ((*ival >> 24) & 0x000000ff) |
((*ival >> 8) & 0x0000ff00) |
((*ival << 8) & 0x00ff0000) |
((*ival << 24) & 0xff000000);
#endif
}
inline void swap_endian_half(void *val)
{
#if defined(OSMac_)
unsigned short *ival = (unsigned short *)val;
*ival = ((*ival >> 8) & 255) |
((*ival << 8) & 65280);
#endif
}
const double two_pow_neg14 = pow(2.0, -14.0);
//
// Half to float conversion
//
inline float halfToFloat(unsigned short val)
{
float outValue = 0.0f;
#if defined(OSMac_)
// Need to swap bytes on Power PC (Mac)
swap_endian_half( &val );
#endif
// Convert 16-value into a float...
double h_mantissa = (float) (val & 1023); // Mantissa = low order 10 bits
double h_exponent = (float) ((val >> 10) & 31); // Exponent = next 5 bits
unsigned int i_sign = (val >> 15) & 1; ; // Sign is the highest bit
double h_sign = (i_sign == 0) ? 1.0 : -1.0;
if (h_exponent != 30.0)
{
outValue = (float) (h_sign * pow(2.0, h_exponent-15.0) * ( 1.0 + ( h_mantissa / 1024.0 )));
}
else
{
outValue = (float) ( h_sign * pow(2.0, two_pow_neg14) * ( h_mantissa / 1024.0 ) );
}
//printf("[%d][%d] = inValue=%d, isign =%d, sign=%g, exp=%g, mant=%g, value=%g\n", y,x, *inPtr, i_sign, h_sign, h_exponent, h_mantissa, outValue);
return outValue;
}
//
// DESCRIPTION:
MStatus ddsFloatReader::open( MString filename, MImageFileInfo* info)
{
#if _WIN32
if ( ( fInputFile = fopen( filename.asChar(), "rb" ) ) == NULL )
#else
if ( ( fInputFile = fopen( filename.asChar(), "r" ) ) == NULL )
#endif
{
// Unable to open the file
return MS::kFailure;
}
// Read the DDS header in
//
if ( fread( &fHeader, 1, sizeof(DDS_HEADER), fInputFile) == sizeof(DDS_HEADER) )
{
swap_endian(&fHeader.fCapabilities.dwCaps2);
// Cube maps and volume textures are not supported currently.
//
if (( (fHeader.fCapabilities.dwCaps2) & DDSCAPS2_CUBEMAP_FLAG ) ||
( (fHeader.fCapabilities.dwCaps2) & DDSCAPS2_VOLUME_FLAG ))
{
close();
return MS::kFailure;
}
// Get dimensions of image
//
swap_endian(&fHeader.fWidth);
swap_endian(&fHeader.fHeight);
fWidth = fHeader.fWidth;
fHeight = fHeader.fHeight;
fNumChannels = 0;
if (fWidth==0 || fHeight==0)
{
close();
return MS::kFailure;
}
// Check for float formats
//
swap_endian(& fHeader.fFormat.fFlags );
if (fHeader.fFormat.fFlags & DDS_FOURCC_FLAG)
{
swap_endian(&fHeader.fFormat.fPixelFormat);
bool supportedFormat = true;
// Half float formats
switch (fHeader.fFormat.fPixelFormat)
{
// Half float formats
//
case DDS_R16F:
fNumChannels = 1;
fBytesPerPixel = 2; // 16-bits
break;
case DDS_G16R16F:
fNumChannels = 2;
fBytesPerPixel = 4; // 32-bits
break;
case DDS_A16B16G16R16F:
fNumChannels = 4;
fBytesPerPixel = 8; // 64-bits
break;
// IEEE 32 bit formats
case DDS_R32F:
fNumChannels = 1;
fBytesPerPixel = 4; // 32-bits
break;
case DDS_G32R32F:
fNumChannels = 2;
fBytesPerPixel = 8; // 64-bits
break;
case DDS_A32B32G32R32F:
fNumChannels = 4;
fBytesPerPixel = 16; // 128-bits
break;
// Anything else is not supported
default:
supportedFormat = false;
break;
}
if (!supportedFormat)
{
close();
return MS::kFailure;
}
}
else
{
close();
return MS::kFailure;
}
// Return image information based on the header
//
if (info)
{
//printf("** Opened DDS: %s (w=%d,h=%d, fmt=%d)\n",
// filename.asChar(),
// fWidth,
// fHeight,
// fHeader.fFormat.fPixelFormat);
// Only read in 1 image for now
info->numberOfImages( 1 );
info->width( fWidth );
info->height( fHeight );
info->channels( fNumChannels );
info->hasAlpha( fNumChannels == 4 );
// Mip maps not handled for now
info->hasMipMaps( false );
}
}
return MS::kSuccess;
}
//
// DESCRIPTION:
MStatus ddsFloatReader::load( MImage& image, unsigned int imageNumber)
{
MStatus loaded = MS::kFailure;
// Check for half type, since we need to perform
// some special bit conversions to get a float back.
//
bool isHalfType = false;
switch (fHeader.fFormat.fPixelFormat)
{
case DDS_R16F:
isHalfType = true;
break;
case DDS_G16R16F:
isHalfType = true;
break;
case DDS_A16B16G16R16F:
isHalfType = true;
break;
default:
break;
}
// Create the output buffer
//
image.create( fWidth, fHeight, fNumChannels, MImage::kFloat);
float* outputBuffer = image.floatPixels();
// Create the input buffer to read one scanline at a time.
//
float *inputFloatBuffer = NULL; // For IEEE float
unsigned short *inputIntBuffer = NULL; // For half-float = 16-bits per channel
if (!isHalfType)
{
inputFloatBuffer = new float[fWidth * fNumChannels];
}
else
inputIntBuffer= new unsigned short[fWidth * fNumChannels];
// Can't get sufficient memory so fail and close up the
// file.
if ((!inputFloatBuffer && !inputIntBuffer) || !outputBuffer)
{
if (inputFloatBuffer)
{
delete [] inputFloatBuffer;
inputFloatBuffer = NULL;
}
if (inputIntBuffer)
{
delete [] inputIntBuffer;
inputIntBuffer = NULL;
}
close();
return MS::kFailure;
}
// Transfer to output buffer.
//
if (isHalfType)
{
unsigned int x,y;
float *outPtr = outputBuffer;
// Do a scanline at a time. From top-to-bottom
// so that scan lines are flipped for Maya's usage.
//
outPtr += (fHeight-1) * (fWidth * fNumChannels);
for (y=0; y<fHeight; y++)
{
// Read in one scan line of data
unsigned int numBytes = fWidth * fBytesPerPixel;
fread( inputIntBuffer, 1, numBytes, fInputFile);
unsigned short *inPtr = inputIntBuffer;
for (x=0; x<fWidth * fNumChannels; x++)
{
float outValue = halfToFloat(*inPtr);
*(outPtr + x) = outValue;
inPtr++; // Jump 2 bytes at a time = 1 channel
}
outPtr -= (fWidth * fNumChannels);
}
loaded = MS::kSuccess;
}
// IEEE 32-bit float
else
{
unsigned int x,y;
float *outPtr = outputBuffer;
// Do a scanline at a time. From top-to-bottom
// so that scan lines are flipped for Maya's usage.
//
outPtr += (fHeight-1) * (fWidth * fNumChannels);
for (y=0; y<fHeight; y++)
{
// Read in one scan line of data
unsigned int numBytes = fWidth * fBytesPerPixel;
fread( inputFloatBuffer, 1, numBytes, fInputFile);
#if defined(OSMac_)
// Need to swap bytes on Power PC (Mac)
unsigned int *bPtr = (unsigned int *)inputFloatBuffer;
unsigned int b;
for (b=0; b<numBytes / 4; b++)
{
swap_endian( bPtr );
bPtr++;
}
#endif
float *inPtr = inputFloatBuffer;
for (x=0; x<fWidth * fNumChannels; x++)
{
*(outPtr + x) = *inPtr;
inPtr++;
}
outPtr -= (fWidth * fNumChannels);
}
loaded = MS::kSuccess;
}
// Cleanup memory and close the file
if (inputFloatBuffer)
{
delete [] inputFloatBuffer;
inputFloatBuffer = NULL;
}
if (inputIntBuffer)
{
delete [] inputFloatBuffer;
inputFloatBuffer = NULL;
}
close();
return loaded;
}
MStatus initializePlugin( MObject obj )
{
MFnPlugin plugin( obj, PLUGIN_COMPANY, "8.0", "Any" );
MStringArray extensions;
extensions.append( "dds");
CHECK_MSTATUS( plugin.registerImageFile(
kImageFormatName,
ddsFloatReader::creator,
extensions));
return MS::kSuccess;
}
// DESCRIPTION:
MStatus uninitializePlugin( MObject obj )
{
MFnPlugin plugin( obj );
CHECK_MSTATUS( plugin.deregisterImageFile( kImageFormatName ) );
return MS::kSuccess;
}