添加粒子系统后 glTexSubImage2d 不起作用（每个粒子上都有纹理）

发布于 2024-11-27 03:47:29 字数 1769 浏览 0 评论 0原文

我们很高兴使用 glTexSubImage2d 每隔几帧更新一次纹理，该纹理在我们的 GL 初始化中已使用 glTexImage2d 初始化。添加每个粒子纹理本身的粒子系统后，我们显示 glTexSubImage2d 纹理的四边形不会显示。

粒子的纹理是 PNG，因此我们使用 SDL 将 PNG 加载到 SDL Surface，然后使用 glTexImage2d 将 PNG 绑定到纹理。

如果我们将四边形的 glTexSubImage2d 调用更改为 glTexImage2d 调用，则会显示纹理，但这效率极低，并且至少将帧速率减少一半，因此宁愿使用 glTexSubImage2d （就像以前一样）。

有谁知道为什么我们现在不能使用 glTexSubImage2d 吗？

下面是纹理初始化和绑定的相关代码片段：

加载粒子纹理

//Load smoke texture
SDL_Surface *surface;   
SDL_Surface *alpha_image;

if( (surface = IMG_Load("smoke_particle.png")))
{
    SDL_PixelFormat *pixf = SDL_GetVideoSurface()->format;
    alpha_image = SDL_CreateRGBSurface( SDL_SWSURFACE, surface->w, surface->h, 32, pixf->Bmask, pixf->Gmask, pixf->Rmask, pixf->Amask );

    SDL_SetAlpha(surface,0,0);
    SDL_BlitSurface( surface, NULL, alpha_image, NULL );

    glGenTextures( 1, &texture );
    glBindTexture( GL_TEXTURE_2D, texture );

    glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR );
    glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR );

    glTexImage2D( GL_TEXTURE_2D, 0, GL_RGBA, surface->w, surface->h, 0, 
                GL_RGBA, GL_UNSIGNED_BYTE, surface->pixels );
}

设置四边形的纹理：

glEnable(GL_TEXTURE_2D);
glGenTextures(1, &texVid);

glBindTexture(GL_TEXTURE_2D, texVid);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexImage2D(GL_TEXTURE_2D, 0, 3, VIDEO_WIDTH, VIDEO_HEIGHT, 0, GL_RGB, GL_UNSIGNED_BYTE, NULL);

glPixelStorei(GL_UNPACK_ALIGNMENT, 2);

initGL() 中的唯一调用是启用GL_TEXTURE_2D、GL_BLEND 设置 glBlendFunc() 以及四边形纹理的设置如上。

有什么想法吗？

原文

We were happily using glTexSubImage2d to update a texture every few frames which had been initialised with glTexImage2d in our GL initialisation. After adding in a particle system with each particle textured itself our quad showing the glTexSubImage2d texture doesn't display.

The particle's textures are PNG and so we use SDL to load the PNG to an SDL Surface and then glTexImage2d is used to bind the PNG to a texture.

If we change the quad's glTexSubImage2d call to a glTexImage2d call the texture shows but this is extremely inefficient and cuts the framerate in half at least and so would rather be using glTexSubImage2d (as it worked before).

Does anyone have any idea why we now can't use glTexSubImage2d?

Below is relevant pieces of code for the initialisation and binding of textures:

Loading in the particle texture

//Load smoke texture
SDL_Surface *surface;   
SDL_Surface *alpha_image;

if( (surface = IMG_Load("smoke_particle.png")))
{
    SDL_PixelFormat *pixf = SDL_GetVideoSurface()->format;
    alpha_image = SDL_CreateRGBSurface( SDL_SWSURFACE, surface->w, surface->h, 32, pixf->Bmask, pixf->Gmask, pixf->Rmask, pixf->Amask );

    SDL_SetAlpha(surface,0,0);
    SDL_BlitSurface( surface, NULL, alpha_image, NULL );

    glGenTextures( 1, &texture );
    glBindTexture( GL_TEXTURE_2D, texture );

    glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR );
    glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR );

    glTexImage2D( GL_TEXTURE_2D, 0, GL_RGBA, surface->w, surface->h, 0, 
                GL_RGBA, GL_UNSIGNED_BYTE, surface->pixels );
}

Setting up the quad's texture:

glEnable(GL_TEXTURE_2D);
glGenTextures(1, &texVid);

glBindTexture(GL_TEXTURE_2D, texVid);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexImage2D(GL_TEXTURE_2D, 0, 3, VIDEO_WIDTH, VIDEO_HEIGHT, 0, GL_RGB, GL_UNSIGNED_BYTE, NULL);

glPixelStorei(GL_UNPACK_ALIGNMENT, 2);

The only calls in initGL() are the enabling of GL_TEXTURE_2D, GL_BLEND setting up glBlendFunc() and the setting up of the quad's texture as above.

Any ideas?

分享到QQ

分享到微博