PDA

View Full Version : Hardware noise() support - any news?



StefanG
09-18-2005, 07:42 AM
Does anyone have any information (which can be shared) concerning when the GLSL noise() functions will start returning something else than zero with anything else than 3DLabs cards?

Isn't this a long awaited feature? What is the problem? Patent issues? (I know Ken Perlin has a patent pending on one specific hardware implementation of Perlin noise.) Or is there simply a lack of demand? Software shaders for RenderMan use noise() calls all over the place, so I would expect it to be just as useful for hardware shaders.

Imagine what we could do if a noise() call would be just as quick as a texture fetch. It can be done, there's no doubt about that. It is actually quite a lot simpler than a texture fetch, as there are no memory reads or cache issues involved.

Obli
09-21-2005, 08:45 AM
I believe it'll need more artist tools before noise gets really requested to be in hardware. Last time I posted something on this, no one really cared about noise.
As soon as the bandwith/performance gap gets tragic, someone will surely have to put it in silicon and start saving.

Humus
09-21-2005, 07:01 PM
Personally I find using a texture for noise much more useful, with mipmapping and everything.

StefanG
09-27-2005, 08:42 AM
Personally I find using a texture for noise much more useful, with mipmapping and everything. Textures can certainly get some jobs done well, but animated and/or 3D noise becomes difficult to do, 4D noise is impossible and there will be directional artefacts in a stored image noise texture that is magnified. There are many good reasons why procedural noise is used so extensively in offline rendering.