SDL_image to textured quad

Started by
29 comments, last by icecubeflower 15 years, 9 months ago
Hey I've a got 2D game going on. I used SDL and also OpenGL. For loading images I always use .TGA's and I load them myself with code I mostly took from NeHe's site. It's like this: I declare this data type:

struct TextureImage
{
   GLubyte *imageData;
   GLuint bpp;
   GLuint width;
   GLuint height;
   GLuint texID;
};

Then I declare some variable:

TextureImage *whatever;

Then I declare this function:

bool LoadTGA(TextureImage *texture, string &filename)
{
   GLubyte TGAheader[12]={0,0,2,0,0,0,0,0,0,0,0,0};
   GLubyte TGAcompare[12];
   GLubyte header[6];
   GLuint bytesPerPixel;
   GLuint imageSize;
   GLuint temp;
   GLuint type=GL_RGBA;

   FILE *file = fopen(filename.c_str(), "rb");

   if(file==NULL||
      fread(TGAcompare,1,sizeof(TGAcompare),file)!=sizeof(TGAcompare)||
      memcmp(TGAheader,TGAcompare,sizeof(TGAheader))!=0||
      fread(header,1,sizeof(header),file)!=sizeof(header))
   {
      if(file!=NULL)
         fclose(file);
      return false;
   }
   texture->width  = header[1] * 256 + header[0];
   texture->height = header[3] * 256 + header[2];

   if(texture->width<=0||texture->height<=0||(header[4]!=24 && header[4]!=32))
   {
      fclose(file);
      return false;
   }

   texture->bpp=header[4];
   bytesPerPixel=texture->bpp/8;
   imageSize=texture->width*texture->height*bytesPerPixel;

   texture->imageData = (GLubyte *)malloc(imageSize);  //Ah, malloc, C's answer to "new".  So when is this memory freed?
                                                       //you have to free it yourself NeHe has a Deinitialize function

   if(texture->imageData==NULL||fread(texture->imageData, 1, imageSize, file)!=imageSize)
   {
      if(texture->imageData!=NULL)
      {
         free(texture->imageData);  //well, I guess this is how you free it
         texture->imageData = NULL;
      }
      fclose(file);
      return false;
   }

   //for a fun time comment this out
   for(GLuint i=0; i<(unsigned int)imageSize; i+=bytesPerPixel)
   {
      temp=texture->imageData;
      texture->imageData = texture->imageData;
      texture-&gt;imageData = temp;
   }

   fclose (file);

   <span class="cpp-comment">// Build A Texture From The Data</span>
   glGenTextures(<span class="cpp-number">1</span>, &amp;texture[<span class="cpp-number">0</span>].texID);  <span class="cpp-comment">//texture is a pointer to a TextureImage</span>
                                         <span class="cpp-comment">//why not just texture.texID?  why &amp;texture[0]??  wtf</span>
                                         <span class="cpp-comment">//the the address of a pointer?  the first element of it?  it has no elements, wtf?</span>
                                         <span class="cpp-comment">//wtf could [0] possibly mean?</span>
   <span class="cpp-comment">//oh I remember, you use -&gt; for pointers instead of . like above</span>
   <span class="cpp-comment">//so here he wants to use . so maybe ampersand on a pointer "converts" it back to a regular data type?</span>
   <span class="cpp-comment">//I have no idea what's going on</span>

   glBindTexture(GL_TEXTURE_2D, texture[<span class="cpp-number">0</span>].texID);  <span class="cpp-comment">//And here it needs no ampersand.  Why?  I don't know.</span>
   glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
   glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);

   <span class="cpp-keyword">if</span>(texture[<span class="cpp-number">0</span>].bpp==<span class="cpp-number">24</span>)
   {
      type=GL_RGB;
   }

   <span class="cpp-comment">//Here using . instead of -&gt; so maybe [0] takes care of it?  Then why the ampersand above?</span>
   glTexImage2D(GL_TEXTURE_2D, <span class="cpp-number">0</span>, type, texture[<span class="cpp-number">0</span>].width, texture[<span class="cpp-number">0</span>].height,<span class="cpp-number">0</span>, type, GL_UNSIGNED_BYTE, texture[<span class="cpp-number">0</span>].imageData);

   <span class="cpp-keyword">return</span> <span class="cpp-keyword">true</span>;
}

</pre></div><!–ENDSCRIPT–>

So now I pass *whatever to that function along with the name of a .tga file and I can bind *whatever whenver I want and slap &#111;n some quads.

<!–STARTSCRIPT–><!–source lang="cpp"–><div class="source"><pre>
glBindTexture(GL_TEXTURE_2D, whatever.texID);
glBegin(GL_POLYGON);
   glTexCoord2f(<span class="cpp-number">0</span>.<span class="cpp-number">0</span>,<span class="cpp-number">1</span>.<span class="cpp-number">0</span>);	glVertex3f(<span class="cpp-number">0</span>,<span class="cpp-number">768</span>,<span class="cpp-number">0</span>.<span class="cpp-number">0</span>);
   glTexCoord2f(<span class="cpp-number">1</span>.<span class="cpp-number">0</span>,<span class="cpp-number">1</span>.<span class="cpp-number">0</span>);	glVertex3f(<span class="cpp-number">1024</span>,<span class="cpp-number">768</span>,<span class="cpp-number">0</span>.<span class="cpp-number">0</span>);
   glTexCoord2f(<span class="cpp-number">1</span>.<span class="cpp-number">0</span>,<span class="cpp-number">0</span>.<span class="cpp-number">0</span>);	glVertex3f(<span class="cpp-number">1024</span>,<span class="cpp-number">0</span>,<span class="cpp-number">0</span>.<span class="cpp-number">0</span>);
   glTexCoord2f(<span class="cpp-number">0</span>.<span class="cpp-number">0</span>,<span class="cpp-number">0</span>.<span class="cpp-number">0</span>);	glVertex3f(<span class="cpp-number">0</span>,<span class="cpp-number">0</span>,<span class="cpp-number">0</span>.<span class="cpp-number">0</span>);
glEnd();

</pre></div><!–ENDSCRIPT–>

But TGA's always take up a few trillion terrabytes and no &#111;ne's used them since Fred Flintstone so I want to start loading .PNG's instead.

I want to use SDL_image loader and using SDL_image loader for an SDL game is easy enough except I want it to load the PNG into a data type that I can bind to an OpenGL texture like I'm doing above with .tga's.

I cannot find how to do this.  I don't think it would take more than a page of code, I figured there would be something out there I could just block copy and figure out later but I can't find anything.  I don't know why, lots of people have probably loaded .png's with SDL_image and textured OpenGL quads with them.
Advertisement
Enjoy all the love of many different formats the SDL_IMage loads

bool TextureLoader::Load(std::string& filename){	SDL_Surface* tex = IMG_Load(filename.c_str());	if(!tex)            return false;                           //setup your texture parameters however you like hereglBindTexture(...);glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB8, tex->w, tex->h, 0, GL_RGB, GL_UNSIGNED_BYTE, tex->pixels);	SDL_FreeSurface(tex);	return true;}


[Edited by - MARS_999 on July 9, 2008 1:30:19 AM]
I don't really get it. What is this TextureLoader class? Am I supposed to write it myself and add imageWidth and all those variables to it and add the Load(string) function?

It looks to me like all it does is return true. tex is freed at the end, I don't see what the result is. I need something that I can bind and then map to a quad. Sorry I'm probably just not understanding it, I don't understand graphics very well.
Yeah its from a class, but here is a complete code, look above.
Quote:Original post by icecubeflower
I don't really get it. What is this TextureLoader class? Am I supposed to write it myself and add imageWidth and all those variables to it and add the Load(string) function?

It looks to me like all it does is return true. tex is freed at the end, I don't see what the result is. I need something that I can bind and then map to a quad. Sorry I'm probably just not understanding it, I don't understand graphics very well.


When you call glTexImage2d, it will actually copy the data over to the video card, not just bind it. Once you done this it isn't necessary to keep the image loaded. In his snippet, he only shows how to use SDL's image library capabilities. You can use your knowledge on OpenGL's texture ability with his code to -texture a quad-. He also didn't show exactly how the params are passes as it isn't necessary, you can come up with it.
Why do I need to make a TextureLoader class? Why do I need a class? Why can't I just make a function like this:

void Load(SDL_Surface* tex, std::string& filename){   tex = IMG_Load(filename.c_str());   if(!tex)      return false;                              //setup your texture parameters however you like here   return true;}


And then after I use that function I have my tex variable to work with so I can go like this back in my main program:

   glBindTexture(...);   glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB8, tex->w, tex->h, 0, GL_RGB, GL_UNSIGNED_BYTE, tex->pixels);   SDL_FreeSurface(tex);


Really I'm not even sure why I even need that glTexImage2D function. Usually I just bind a texture and then draw a quad and I'm done. Unless I need glTexImage2D back in my Load function?
The function you are questioning is what dumps the image into video memory! This way the video card can use its own memory rather than the slow transfer of ram -> video ram.

Here is a snippet of my texture class:
#include "Texture.h"#include <string>#include <sdl/sdl_image.h>#include <sdl/sdl_opengl.h>Texture::Texture(const char *filename){	// Load Texture into Memory	SDL_Surface *image = IMG_Load(filename);	if(!image)		throw std::pair<const char*, const char*>("Texture::Texture", "Error Loading Image from Disk");			// Create Video-Card Texture	glGenTextures(1, &tex_id);	glBindTexture(GL_TEXTURE_2D, tex_id);	glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, image->w, image->h, 0, GL_RGBA, GL_UNSIGNED_BYTE, image->pixels);	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);	SDL_FreeSurface(image);}void Texture::Bind(){	glBindTexture(GL_TEXTURE_2D, tex_id);}


Using something like this makes it a little neater (at least in my opinion) when binding different textures at different instances. Also the destructor can take care of any texture releasing, etc, if needed to be done in real time. =)
I think I almost get it. That *image you declared is used and then freed on the spot. Before it's freed the image is loaded into video memory via that function I dared to question. So that function call goes out of scope. Now anytime I draw a quad that texture is going to be mapped to it until I call Texture::Texture(const char *filename) again. Am I right?
For small projects, keep all of the texture objects in video memory. Bind which ever texture you need to use by calling the Bind method in the class, that way the next textured raster being rendered it will use the newly bound texture without having to do any IO (still in video memory). Basically the less IO you do, the faster... Much faster. :)

One way I have been doing this is creating a texture manager, where it will load images from a directory, than the user can request a reference of the texture object for rendering. Once the manager goes out of scope all the resources are freed.
Sorry man, I don't really understand what you just said.

I don't really understand how the textures you have loaded into video memory are ever accessed again. When I did it with the LoadTGA(TextureImage *texture, string &filename) function that I listed in my first post *texture was NOT freed at the end like your *image is. It's up to me to make sure it's freed later.

That way whenever I load a new map and new monsters I assign a TextureImage* to every new TGA file by sending it into my LoadTGA() function along with the name of the file. Then my Draw() function executes continuously but as it draws the map and hero and monsters it just binds those TextureImage*'s over and over.

Your *image is freed at the end of the function. It is bound so it can be texture mapped right then. But then if something else is bound then it seems to me that the image is lost.

I just don't see how you can free the image from memory right after you bind it. The Draw() function uses that image every cycle. Why would you free it?

This topic is closed to new replies.

Advertisement