FidelityFX Blur
FidelityFX Blur demonstrates a single pass Gaussian blur effect which is functionally the same as a standard two pass separable Gaussian blur (but more performant).
For details on the techniques that underpin the FidelityFX Blur effect you can refer to the respective technique documentation .
Requirements
-
Windows
-
DirectX(R)12
-
Vulkan(R)
UI elements
The UI elements included in the sample enable switching between different Gaussian kernel weights, different kernel sizes, and floating point precision. Also, the UI provides a comparison mode to compare the FidelityFX Blur technique to standard blur techniques. The table below summarises the UI elements and what they control within the sample.
Element name |
Value |
Description |
---|---|---|
Algorithm |
|
Switches the currently used algorithm. |
Gaussian Kernel Sigma |
|
Switches the sigma used for generating the Gaussian kernel. |
Kernel Size |
|
Switches the size of the blur kernel. |
Floating Point Math |
|
Switches between FP32 and FP16 implementations of the technique. |
Display the difference |
|
Enables/Disables the comparison mode. |
Comparison mode settings |
|
Same options as above, but for the comparison mode filter. |
Diff Factor |
|
Scale to apply to the comparison difference for display. |
Comparison mode
The comparison mode is enabled by selecting the Display the difference between two algorithms
checkbox, displays the difference between the two selected blur algorithms.
The image generated by the algorithm selected in the Compare Algorithm
dropdown and its corresponding kernel sigma, size, and math settings, are subtracted from the image generated by the algorithm, and corresponding settings, selected in the Algorithm
dropdown. In order to better visualize very tiny differences, the Diff Factor
slider specifies a value that the difference between the two images is multiplied by before it is displayed on screen.
Sample controls and configurations
For sample controls, configuration, and FidelityFX Cauldron Framework UI element details, see Running the samples .