Gaussian Adjustment: Range and Speed

Since Filter Forge is procedural, even pixel-based components (like Blur, Motion Blur, Sharpen and High Pass) have their parameters set independently of pixel size and are specified as percentage of resulting size. This is done to preserve resolution independence as much as possible, regardless of components used.

The end result is if your image is, say, 600x600 pixels and Size in filter controls is set to 600, then the Radius slider value of 2 is equivalent to 2 * 600 / 100 = 12 pixels. Remember that you can always enter fractional values manually into edit box beside the control slider, which allows you to have very small blur radii.

I'm afraid that speed of blur (and associated components) won't improve in near future. This is due to sample-based resolution-independent architecture -- we don't operate with fixed-sized bitmaps internally, so we're limited in optimizations we can do.

Also, remember that all our components are calculated with at least single precision floating point, and most of them use double precision floating point to allow them to be used in geometrical effects like height mapping, refraction, and so on -- this places quite a burden on your processor[s] and memory.

Nevertheless, we will try improve speed of entire Filter Forge, as well as individual components -- just don't expect major boosts.