And also compute number of required mipmap levels down to 1x1 texture
size

New demo showcasing mipmapping with texture2DArray.

If drivers expose that extension, we can use it to avoid going through
allocating a buffer objec...

- Increase/decrease photon map resolution with arrow up/down
- Increase/decrease number of photo...

The extension ARB_sampler_objects (core in OpenGL 3.3) introduced
"Sampler Objects" to decouple ...

Make it clear that the "data" argument can be null, in which case the
storage of the buffer obje...

refactoring: No need to toggle depth test all the time. That was some
left-over from another dem...

The reason for that 1282 error actually was that we need to use
HALF_FLOAT when doing pixel unpa...

The same as from the clearTexSubImage function above.

Also: fixed tabs/spaces formatting in pre...

The "data" parameter for those two functions can be null. In this case
the buffer is cleared wit...

I wanted to clear the texture without using a host-allocated ByteBuffer.
OpenGL 4.4 has support ...

The shader can specify, via the "binding" layout qualifier, the binding
point of an image unifor...

First, use getProgramResourceIndex to obtain the "resource index" of the
SSBO binding that is us...

First, use getProgramResourceIndex to obtain the "resource index" of the
SSBO binding that is us...

First, use getProgramResourceIndex to obtain the "resource index" of the
SSBO binding that is us...

First, thanks to Ioannis for his help and final solution to this
problem!

There are certain dri...

First, thanks to Ioannis for his help and final solution to this
problem!

There are certain dri...

Was allocating twice the memory needed for the boxes. Now the fix
ensures it to render twice as ...

Use symbolic constants for cube face indices

Also added distance attenuation like the other demos do

This demo uses a cube map array texture to hold a "photon map" for each
of the boxes in the scen...

- modelViewMatrix was actually just a viewMatrix as it did not represent
a transformation from m...

Use FloatBuffer views to fill floats.
+ we cannot forget flip(), because not needed
+ putFloat()...

Only build the VBO for one single unit box and an additional buffer for
the per-instance data (b...

Only build the VBO for one single unit box and an additional buffer for
the per-instance data (b...

Only build the VBO for one single unit box and an additional buffer for
the per-instance data (b...

- Use glTexStorage for 4.3 programs to allocate texture storage without
optionally initializing ...

We are actually dealing with world position and normal here, not "view"
position/normal.

This demo uses a Shader Storage Buffer Object to dynamically specify the
boxes of the scene from...

Also offset the view position a bit towards the camera to avoid
artifacts

- use world position and normal output from rasterizer (makes things
easier)
- currently, only s...

Still open points:
- Activate FBO to render viewPosition and viewNormal
- Duplicate Compute Shad...

The idea of using hybrid rasterization + ray tracing approach is not
new. It allows to use the r...

There was an issue when the tangent-space for randomDiskPoint
computation changed orientation/sw...

We don't need the temporary float[] anymore. We can instead directly
work on the Matrix4f elements.

We compute the random values outside of those functions now, and inject
them into the every func...

The javaDocLink property now returns the qualified javadoc link and replaces the dsaLink property.

So no need to set them to zero in the constructor.

Check if setting is already set.
Use new system property org.lwjgl.opengl.maxVersion
Comments an...

handling of frame resizing (when minimizing/restoring window no
framebuffer reset needs to be don...

Also increased performance by precomputing random values once and
reusing it all around a shader ...

This demo has been contributed by Kai Burjack (@httpdigest). It comes in two versions; the first...

When defined, functions above the specified version will not be loaded, even if they are exposed...

Added many new extensions, improved documentation, added documentation to CLCapabilities flags.

The alternative method had a flipped check and in some cases we were outputting multiple checks ...

This change makes Map functions consistent with Data/SubData funtions. In practice it is pointle...