Jump to content
  • Advertisement
Sign in to follow this  
MadsGustaf

OpenGL The most efficient way of loading alot of bitmaps into memory?

This topic is 3795 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

This is how i currently load my 2d textures. As i am rather new to the OpenGL api, this seems as a pretty straightforward approach, but it also seems terribly inefficient, as it takes several seconds to load all my bitmaps.
[source langg = "cpp"]
int LoadGLTextures()
{
 int status = FALSE;
 AUX_RGBImageRec* TextureImage[textures];
 
 for(int i = 0; i < textures; i++)
 {
  memset(TextureImage,0,sizeof(void *)*textures);
  switch(i)
  {
  // load the appropriate image into the Tex. image array.
  }
 status = TRUE;
   glGenTextures(1,&texture);
  
   glBindTexture(GL_TEXTURE_2D, texture);

		glTexParameteri(GL_TEXTURE_2D,GL_TEXTURE_MAG_FILTER,GL_LINEAR);
		glTexParameteri(GL_TEXTURE_2D,GL_TEXTURE_MIN_FILTER,GL_LINEAR_MIPMAP_NEAREST); 
		gluBuild2DMipmaps(GL_TEXTURE_2D, 3, TextureImage->sizeX, TextureImage->sizeY, GL_RGB, GL_UNSIGNED_BYTE, TextureImage->data);
		
if (TextureImage)									// If Texture Exists
	{
		if (TextureImage->data)							// If Texture Image Exists
		{
			free(TextureImage->data);					// Free The Texture Image Memory
		}

		free(TextureImage);								// Free The Image Structure
	} 
}//for loop

Share this post


Link to post
Share on other sites
Advertisement
How many bitmaps are you loading, and how big are they?
It would be useful to see how you are actually loading the bitmap data from the file, but the OpenGL code you are using is how it should be.

Share this post


Link to post
Share on other sites
I couldn't help but notice the following in your code.

AUX_RGBImageRec* TextureImage[textures];
memset(TextureImage,0,sizeof(void *)*textures);


Given the usage, it seems quite obvious that your intent was to set all the pointers in the array to the null pointer value. Of course, this is not what your code does. Your code changes the binary representation of all the pointers in the array to the binary representation of the integer zero, which may or may not be the same as that of the null pointer value.

In short, you are setting the pointers in the TextureImage array to a certain value, and then hoping that the value which you used also happens to be the null pointer value:
// Set pointers to a value which I hope is "null"
memset(TextureImage,0,sizeof(void*)*textures);


A shorter version, which also happens to do what you intended, is:
AUX_RGBImageRec* TextureImage[textures] = {};

Share this post


Link to post
Share on other sites
I assume you use the glaux library since there is a AUX_RGBImageRec in your code. I suggest to dump this library since it is deprecated. There are other image loading libraries with which you can load OpenGL textures. I for instance use SDL_image, convert the loaded bitmap to the appropriate OpenGL format and upload it. There is also DevIL and probably many others.

Using such image loading libraries may still not the most optimal way to load the images since you first have to convert them to the right type and bpp.

The most optimal way appart from not loading the images at all is to lay out the data in exactly the same way the API expects it and store this in a file. This means there will not be any parsing/conversion required on load. There might be endianness issues though.

And another good idea it to combine the above with compressed textures, as they are smaller and can thus be loaded faster. They are also faster when rendering.

Share this post


Link to post
Share on other sites
Quote:
Original post by Trenki
And another good idea it to combine the above with compressed textures, as they are smaller and can thus be loaded faster. They are also faster when rendering.


I'm pretty sure that this is untrue. If anything, compressed textures will take longer to load, as the data will need to be uncompressed before they are loaded into memory. They also won't be rendered any faster, as the data is always stored the same way in memory, regardless of how they are stored on disk.

Unfortunately, there's not a whole lot you can do to vastly increase the speed of loading files. However, here are a few suggestions:

- Do you have lots of very small images? The cost of opening and closing each file, especially when there's very little to read, can quickly add up. Putting all of the image data into fewer files, and then only using what you need when you need it (say, by using glTexCoord) can reduce this cost.

- Do you really need all of your images immediately? Its unlikely, considering how long it takes so long to load them all. It would probably be better to just load them when they are needed. Unless you need to load in a *very* large image, or need to load in many images, taking the hit while the program is running through the main loop probably isn't going to hurt you much. It'll happen pretty rarely, so its usually not a problem.

Share this post


Link to post
Share on other sites
1) Textures can be compressed using hardware supported compression (ie DTX, etc). These don't need to be decompressed and are smaller/faster to load.

2) General compression (ie zlib) can help load times if file IO is a bottleneck. Profile your file loading times vs smaller file+decompression. Heavy compression will definitely take longer, but you can get a nice size reduction with relatively minimal compression which can be a win as file IO is all sorts of slow.

Share this post


Link to post
Share on other sites
Quote:
Original post by derickdong
Quote:
Original post by Trenki
And another good idea it to combine the above with compressed textures, as they are smaller and can thus be loaded faster. They are also faster when rendering.


I'm pretty sure that this is untrue. If anything, compressed textures will take longer to load, as the data will need to be uncompressed before they are loaded into memory. They also won't be rendered any faster, as the data is always stored the same way in memory, regardless of how they are stored on disk.


You are wrong on this one. Compressed textures (with DXT or other compression formats) take up less space on disk and you can upload them with the glCompressedTexImage commands. They will NOT get expanded internally so they will use less texture memory as well. Compressed textures are a hardware feature and will be faster since less data has to be read from the texture memory to generate pixel colors.

I mean, when compressed textures would get expanded to normal textures internally then there would not be any point to have them in the first place.

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

We are the game development community.

Whether you are an indie, hobbyist, AAA developer, or just trying to learn, GameDev.net is the place for you to learn, share, and connect with the games industry. Learn more About Us or sign up!

Sign me up!