Writing an array to a BMP file

[Wawski]

First of all, being new to this forum I would like to say hello to everyone :)

now my problem; I'm trying to write the contents of an array to a BMP file.
I've google'd, wiki'ed, and searched for two days now and I can't get this figured out.

I know about the BITMAPFILEHEADER, BITMAPINFOHEADER, RGBQUAD, and BYTE structures and the data members contained inside of them and what they represent, I just don't know how in the hell to use them!

I got the fstream working and I can write to txt files. That's about as far as I've managed to get myself..

Can someone teach me how to get these freakin' structs set up and written so I can output the array data?? I just want the most basic way possible to write a BMP (apparently 24-bit bmp files are easiest because you're writing straight colour data to each pixel)

Any help would be appreciated, and links to sites with tutorials might cause me to /wrists

Thanks a lot for your time!

[helios]

Pfft. bmps are so easy you don't even need structues. You can just write the headers yourself:
All integers are unsigned and little endian (http://en.wikipedia.org/wiki/Endianness )

Bytes 0-1: Magic word. It contains the letters 'B' and 'M'.
Bytes 2-5: Double word. File size.
Bytes 6-10: No idea. Leave this zeroed.
Bytes 11-14: Double word. Start of the bitmap.
Bytes 15-18: Double word. Size of the header. Normally set to 40. The header is considered to start at offset 15.
Bytes 19-22: Double word. Width.
Bytes 23-26: Double word. Height.
Bytes 27-28: Word. Planes. Must be 1.
Bytes 29-30: Word. Bits per pixel. You'll need this as 24.
Bytes 31-34: Double word. Compression. Leave it as zero.
Bytes 35-38: Double word. Compressed size. Since there's no compression, this value will be the same as file size.
Bytes 39-42: Double word. Horizontal resolution. See below.
Bytes 43-46: Double word. Vertical resolution. Pixels per meter, according to Wikipedia. The values seem to vary, but 2835 should do for both.
Bytes 47-50: Double word. Number of colors in the palette. Leave as zero.
Bytes 51-54: Double word. "Important colors". Leave as zero.
Bytes 55-EOF: Bitmap.

In a 24-bit bitmap, every pixel channel uses exactly 8 bits and, being a little endian file, they appear in BGR order.
Rows appear from top to bottom (in a 640x480 image, the first pixel in the file is 0;479, then the next row is 478, etc.).
If the last byte in a row isn't at file offset divisible by 4, padding pixels must be added until this condition is met. For example:

//file is an std::vector<unsigned char>
while (file.size()%4)
	file.push_back(0);

This pretty much covers bmps.

[Wawski]

so lets say I have;

float grid[Xmax][Ymax];

How would I write that into the file? That's the problem I'm having, I'm not sure how to input the byte data you listed up there.

ofstream myfile;
myfile.open("test_BMP.bmp");

myfile << // what goes here and how

myfile.close();

would it be something like;

myfile << "BM";
DWORD fileSize = sizeof(grid) // +sizeof(header)?
myfile << filesize;
myfile << 0000;
DWORD startOfBitmap = //no idea 

I think with that example it shows where I'm stuck with this lol

Thanks for the quick reply!

[monia]

I think you will get the reply in this site http://www.experts-exchange.com/Programming/Languages/CPP/Q_23364336.html

[helios]

Well, the thing is that you can't use << for output of binary data. You need to write all the data you have to an array and then write() it to the file in one sweep.

float isn't a suitable data type for pixel data. Maybe unsigned long, but the ideal is a struct pixel24{unsigned char r;unsigned char g;unsigned char b;}, which is portable to all platforms and doesn't waste space.

Example of how to write the data to an array:

data[0]='B';
data[1]='B';
writeLittleDWORD(data+2,305419896);
writeLittleDWORD(data+6,0);
writeLittleDWORD(data+11,55);
writeLittleDWORD(data+15,40);
writeLittleDWORD(data+19,640);
writeLittleWORD(data+27,1);
writeLittleWORD(data+29,24);
//... 

writeLittleDWORD() is a function you'll have to define that will automatize writing binary data to an array. Its prototype can be something like void writeLittleDWORD(unsigned char a,unsigned long b). When b==305419896 (0x12345678), for example, it will perform the following operations (not the actual code, of course):
a[0]=120; //0x78
a[1]=86; //0x56
a[2]=52; //0x34
a[3]=18; //0x12

0x, by the way, is the notation used for hexadecimal numbers.

[Wawski]

I ended up just writing a .raw file, open in photoshop, and re-save as a proper file :p

[helios]

...

I thought you wanted to write a program to see how it was done. You should have told me you just wanted to convert data. That would have saved me a lot of time and typing.

[Wawski]

no no, I totally appreciate what you took the time to write, and I did try my best to understand it. I just couldn't get it to work properly so I just reverted to a method that I knew. After all it did net me roughly what I was looking for.
I don't want you to think that I just ignored your advice, because I did take the time to go through it it detail, line by line, especially since you're the only person that replied to this thread with real tangible advice and suggestions!

[helios]

Even if you couldn't make it work, make sure you understand the concept of endianness. It's basic for some advanced topics in programming.