I made an attempt at a LUT color correction effect, but then I remembered I wasn't in ReShade, and couldn't load directly from the shader itself...
Which means to continue I'll have to be able to compile the code by myself, which any previous attempt failed.
Onwards...
---------- Post added at 18:50 ---------- Previous post was at 18:38 ----------
Also, regarding the optimization of the box blur;
The way to do it is by first populating all the samples from the 1D kernel (let's say 9 for a radius of 4 (from -radius to +radius with a center pixel), and for each pixel of the row, deleting the left most pixel from the kernel and adding the new right-most pixel.
But that means that the color array is persistent throughout the execution of the shader on an image; and there's no way of having a global variable that persists through the shader's lifetime; that means there's no way of being able to void the kernel size's influence on the shader complexity.
Am I missing something?
Which means to continue I'll have to be able to compile the code by myself, which any previous attempt failed.
Onwards...
---------- Post added at 18:50 ---------- Previous post was at 18:38 ----------
Also, regarding the optimization of the box blur;
The way to do it is by first populating all the samples from the 1D kernel (let's say 9 for a radius of 4 (from -radius to +radius with a center pixel), and for each pixel of the row, deleting the left most pixel from the kernel and adding the new right-most pixel.
But that means that the color array is persistent throughout the execution of the shader on an image; and there's no way of having a global variable that persists through the shader's lifetime; that means there's no way of being able to void the kernel size's influence on the shader complexity.
Am I missing something?