I have created a simple OpenGL application.
When zooming into a textured quad, the texture becomes pixelated instead of blurry. I would guess that is due to missing mipmaps?
I create my texture like this:
glGenTextures(1, &mTexture);
glBindTexture(GL_TEXTURE_2D, mTexture);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
And I update it from a PBO like this:
glBindTexture(GL_TEXTURE_2D, mTexture);
glBindBuffer(GL_PIXEL_UNPACK_BUFFER_ARB, mPboIds[mPboIndex]);
glTexSubImage2D(
GL_TEXTURE_2D,
0, 0, 0,
frame->GetWidth(),
frame->GetHeight(),
GL_RGB,
GL_UNSIGNED_BYTE,
0);
I thought that GL_TEXTURE_MAG_FILTER and GL_TEXTURE_MIN_FILTER would tell OpenGL to generate the mipmaps. Ain't that the case?
How can I enable mipmap generation for my texture?
The magnification filter is used when you increase the zoom on a texture, and can have two values:
GL_NEAREST - Returns the value of the texture element that is nearest (in Manhattan distance) to the specified texture coordinates.
GL_LINEAR - Returns the weighted average of the texture elements that are closest to the specified texture coordinates. These can include items wrapped or repeated from other parts of a texture, depending on the values of GL_TEXTURE_WRAP_S and GL_TEXTURE_WRAP_T, and on the exact mapping.
In your case, you are use the wrong magnification filter. It must be GL_LINEAR to minimize the pixelated effect.
MipMaps, for the other hand is used when you want to decrease the zoom and want to get a smooth transition when the texture start to become too far away from the viewer. To get more information about MipMaps, you can look at glTexParameter manual pages, at section GL_TEXTURE_MIN_FILTER, and how to generate in glGenerateMipmap manual page.
Related
I am rendering a huge 3D cube array, that sometimes counts thousands of cubes aligned right next to one another. I am rendering a jpg texture to the cubes, which is just a simple color with a black border around the frame.
The problem:
The array is huge, and the distant parts of the array get kind of mixed into one another, so to say. In other words, the borders in the distant cubes sometimes completely disappear, sometimes they form an arbitrary wavey line together with other neighboring borders. All in all, the scene looks kind of messy because all the fine details (hard borders between the neighboring cubes) are lost/melted together. After searching for the solution online, I understand that the problem might be in my choice of texture filtering options.
This is how the problem actually looks like in OpenGL:
This is how the current code for loading texture and setting texture parameters looks like:
glGenTextures(1, &texture3);
glBindTexture(GL_TEXTURE_2D, texture3);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
// set texture filtering parameters
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
//load image:
data = stbi_load("resources/textures/gray_border.jpg", &width, &height, &nrChannels, 0);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, width, height, 0, GL_RGB, GL_UNSIGNED_BYTE, data);
glGenerateMipmap(GL_TEXTURE_2D);
By now, I have tried playing with changing different parameters to the function glGenerateMipmap() and altering between the parameters in the glTexParameteri() function, but none did work by now.
If you want to enable Mip Mapping, then you have to use one of the minifying functions like GL_NEAREST_MIPMAP_NEAREST, GL_LINEAR_MIPMAP_NEAREST, GL_NEAREST_MIPMAP_LINEAR or GL_LINEAR_MIPMAP_LINEAR, see glTexParameter and Texture - Mip Maps:
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR_MIPMAP_LINEAR);
A further improvement can be gained by Anisotropic filtering, which is provides by the extension ARB_texture_filter_anisotropic and is a core feature since OpenGL 4.6.
e.g.
glTextureParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_ANISOTROPY, 16);
See Sampler Object - Anisotropic filtering
I'm having some trouble with a render-to-texture operation. I create a FBO, attach a texture, render to it, and everything's fine, but when I try to change the attached texture, either to 0 or to a new handle, by calling glFramebufferTexture2DEXT again, the texture I had attached becomes blank (all pixel values reset to (0, 0, 0, 0)).
The documentation doesn't say that this is supposed to happen, and it's a bit troublesome for me because I need to retain the information in this texture. Does anyone know why this is happening, and how to prevent or work around it?
Just ran into a nearly the same situation. The end solution was when we were creating the texture we had to make sure to set up the TexParameters.
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
According to https://www.khronos.org/registry/OpenGL-Refpages/gl4/html/glTexParameter.xhtml, GL_TEXTURE_MIN_FILTER defaults to GL_NEAREST_MIPMAP_LINEAR but we weren't mipmapping. I am not sure why that would cause the texture to appear to be erased, but that is what we did to fix it.
I noticed that when I'm loading a texture, it might change the current drawing color, depending on the texture's color. For example after executing
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, info.biWidth,
info.biHeight, 0, GL_RGB, GL_UNSIGNED_BYTE,bitmap);
glTexParameterf(GL_TEXTURE_2D,
GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexParameterf(GL_TEXTURE_2D,
GL_TEXTURE_MAG_FILTER, GL_LINEAR);
all consecutive polygons drawn to the screen will have a color depending on the texture image loaded.
Is that standard? I didn't find this behavior documented.
Yes, it's how it works, remember that GL is a state machine, so you left the texture bound (and probably enabled), so when drawing it used the first pixel (assuming you didn't provide any texture coordinates) to color the primitive.
To solve it, disable texturing when you don't want texturing, you can do it with:
glDisable(GL_TEXTURE_2D);
I am creating a simple opengl application which obviously includes some 3d-objects and textures. My problem is however that artifacts appear on every texture. These come in the form of triangles along the edges.
I have noticed that it disappears as soon as I move the view-point closer to texture it renders perfectly. Therefore I have a suspicion that it has something to do, either with the mipmapping or the z-buffer.
Please note that all texture-coordinates are loaded from a .3ds-file and all of them are verified to be within the range of 0-1.
Here are a picture of my problem:
Picture 1
The textures are loaded like this:
//Texture parameters
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR_MIPMAP_NEAREST);
glTexEnvf(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE);
//Define the 2d texture
glTexImage2D(GL_TEXTURE_2D, 0, 4, width, height, 0, GL_RGBA, GL_UNSIGNED_BYTE, array);
//Create 2d mipmaps
gluBuild2DMipmaps(GL_TEXTURE_2D, 4, width, height, GL_RGBA, GL_UNSIGNED_BYTE, array);
When I was programming using DirectX, the near plane / far plane distance ratio caused artifacts in the edges.
In my case, if near plane was 1 unit away from 'camera' and far plane was 10000 units away, the ratio is 1/10000 and it created problems. If i set the near plane to 10 or 100, the ratio becomes bigger. It solved the jagged edges problem.
I don't know if/how it is applicable in OpenGL, but you might want to check it out.
I'm trying to capture the screen to a texture with a lower resolution than the screen itself (to render back onto the screen and create a blur/bloom effect), and it doesn't work quite well. I understand mipmaps can be used to do this, but I just can't get the correct sequence of commands to work.
My current code:
width=1024;
height=1024;
glGenTextures(1, &texture);
glEnable(GL_TEXTURE_2D);
glBindTexture(GL_TEXTURE_2D, texture);
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glEnable (GL_BLEND);
glBlendFunc (GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
glTexEnvf(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_MODULATE);
glHint(GL_PERSPECTIVE_CORRECTION_HINT, GL_FASTEST);
glTexParameterf( GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP );
glCopyTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, 0, 0, width, height, 0);
// code for rendering the screen back on goes here
You can't capture and downfilter in one go. You have to capture the full screen to a larger texture first, then mipmaps should be created if auto-create mipmaps are enabled, then you can render one using that texture again, makig sure you adjust the mipmap level suitably.
However, that will look ugly, as the auto mipmapgeneration usually uses a box filter.
What I'd do is to set up some FBOs (Frame Buffer Objects) and GLSL shaders instead. That gives you finer control over all steps:
create the original image in a texture
apply some nice gaussian low-pass filtering
blend the filtering with the original image to the frame buffer