How to convert jpg image to raw 32 bit float

[russianbear]

Hello.
I am trying to convert a jpg image to raw format, specifically RGBA 32 bit float.
I am using Magick++ btw...

Here is what I am doing so far...

Code: Select all

int _tmain(int argc, _TCHAR* argv[]) {

   Geometry g(800, 533);
   Image theI;
   try {
      theI.read(g, "./bluejay.jpg");
      theI.depth(32);
      theI.write("./bluejay32int.procd.rgba");
   }//end try block
   catch(Exception& ex) {
      cout << "imagicktstr,error," << ex.what() << endl;
   }//end catch block
   return 0;
}//end main

This code compiles and successfully outputs a raw image; however, it is still integer!

How do represent the pixels in "float" format????
I noticed that using the command line interface you can use the convert and define options...but I don't see the corresponding APIs for Magick++.

I've been looking at the Image.write method, specifically where I can specify a pixel buffer as the fifth paramter and the type...but I'm still not sure if that is proper way to do this.

Any sort of help would be much appreciated!

[russianbear]

Okay! I got it working, so I'm answering my own question.
My original bluejay.jpg file that I am reading in is 1,666KB big and I know that if I do this right, my file should be 4x as big (RGBA).
So....

Code: Select all

int _tmain(int argc, _TCHAR* argv[]) {
   Geometry g(800, 533);
   Image theI;
   FILE* fh;
   float *floatPixelMap;
   try {
      theI.read(g, "./bluejay.jpg");
      int height = theI.rows();
      int width = theI.columns();
      int size = height * width * 4;

      PixelPacket* myMap = theI.getPixels(0, 0, width, height);
      floatPixelMap = new float[size];

      for (int counter = 0, k=0; counter < size; counter++, ++k) {
         floatPixelMap[counter++]  = (float)myMap[k].red      / 255.0f;
         floatPixelMap[counter++]  = (float)myMap[k].green    / 255.0f;
         floatPixelMap[counter++]  = (float)myMap[k].blue     / 255.0f;
         floatPixelMap[counter]    = (float)myMap[k].opacity  / 255.0f;
      }//end for loop
      fh = fopen("./bluejay32float.procd.rgba", "wb");
      fwrite((void *)floatPixelMap, sizeof(float), size, fh);
   }//end try block

   catch(Exception& ex) {
      cout << "imagicktstr,error," << ex.what() << endl;
   }//end catch block
   return 0;
}//end main

This works...bluejay32float.procd.rgba is 6,663KB big
Just in case anybody needs to do something like this, I don't want them to have to waste an entire day like I did.