Thread: image format

I know I just asked eariler about bmps and said that I din't want to use SDL, well png's were annoying me because of the lack of documentation on the function I needed so I started sdl. I want to load a heightmap, the picture I want to use is a tga. I said screw that and converted it to bmp. I loaded the code from NeHe using the SDL_LoadBMP function(it works on their picture), but it seg faults on my picture. I even wrote my own bmp and cut the image from the tga and then pasted it on my bmp. gimp said its thing and did the conversion. No avail. Here is the code that doesn't work on my image, but on Nehe's it does

Code:

int tl(void){
        int status=0;
        SDL_Surface *TextureImage[1];
        if((TextureImage[0]=SDL_LoadBMP("/home/michael/temp/lesson-6/lesson06/da
ta/nehe.bmp"))){/*their picture*/
                status=1;
                glGenTextures(1,&texture[0]);
                glBindTexture(GL_TEXTURE_2D,texture[0]);
                glTexImage2D(GL_TEXTURE_2D,0,3,TextureImage[0]->w,TextureImage[0
]->h,0,GL_BGR,GL_UNSIGNED_BYTE,TextureImage[0]->pixels);
                glTexParameteri(GL_TEXTURE_2D,GL_TEXTURE_MIN_FILTER,GL_LINEAR);
                glTexParameteri(GL_TEXTURE_2D,GL_TEXTURE_MAG_FILTER,GL_LINEAR);
        }
        if(TextureImage[0])
        SDL_FreeSurface(TextureImage[0]);
        return status;
}

I feel like a beginner, where hello world doesn't work. If anything does anyone here have a heightmap that is a bmp?
[edit]If i edit out the glTexImage2D function it doesn't seg, but of course no texture. What would cause it?

i hate file formats, why are they so complicated.

the formats are usually complicated in the interest of being expandable and adding features people might want to use. for example the TGA format is fairly easy , but in the latter versions at the end of file there are fields of bytes dedicated to the identification of the maker of the TGA image(i.e. copyright info) and other somewhat useful features. By far if you want a simple image format BMP is it. only has two headers and then you only have to remember that in order to use a bmp it has to be DWORD padded(padded so that each line is DWORD divisible).

Also I gather from the link you gave that an instance of SDL_Surface is automatically configured to have the right format/pixel pointers? Also , what exactly is your problem with the code beyond the fact that it apparently doesnt work?

There is no need to store 3 color values per pixel in a heightmap. You only need 1 value so store them as 8-bit RAW files. Heck you could make up your own RAW file format where the first two DWORDs of information would be the height and width followed by the RAW height values.

If your height values are bytes then this would make for some very small files and very fast load times. Also it would make smoothing and bilinear filtering of the heights a lot faster. And note that your heights should be BYTEs not WORDs or DWORDs. Your perspective projection will bring out the slight differences in height, plus you can scale the heights into WORDs or DWORDs later.


I store all my heightmaps as RAW images, including the ones I used for my polygon terrain and my voxel terrain.
BMPs are only useful for the textures and/or color maps if you are using one. Even using 8-bit BMPs is a pain in the arse although not really hard...just a pain. Simplify it and move to RAW.


You want to store your heights as grey scale images in planar mode. Convert your height maps into 8-bit grey-scales and save as a planar RAW file with no header information. Some programs allow you to specify empty bytes prior to the image information so you can insert your own header.