Image Filters

Image filters are arithmetic performed on color channels. Knowing which operation sits behind each control explains why some filters are instant and others are slow, and why applying the same two filters in a different order can give a different result.

Grayscale isn't an average

Converting to grayscale by averaging the three channels equally produces a flat, disappointing image, because human vision doesn't weight the primaries equally. We are far more sensitive to green than to red, and least sensitive to blue. Proper luminance conversion therefore weights green heaviest, red second, and blue least, at roughly seventy-two, twenty-one, and seven percent respectively. The visible difference is significant: foliage and skin tones keep their tonal separation instead of collapsing toward the same gray.

Sepia works the same way, as a fixed matrix applied to each pixel that maps the channels onto a warm brown axis. Both are per-pixel operations. That explains why they run instantly regardless of image size.

Blur is the expensive one

Blur is a convolution rather than a per-pixel operation. Each output pixel is a weighted average of its neighbors, so the work grows with the radius, and a large blur genuinely costs more than every other filter combined. Gaussian blur uses a bell curve for those weights, which is what makes it look natural rather than smeared. A useful implementation detail is that a Gaussian blur can be applied as a horizontal pass followed by a vertical one, producing the same result far faster than a single two-dimensional pass.

Order changes the outcome

Filters aren't commutative. Increasing contrast and then desaturating gives a different image from desaturating and then increasing contrast, because the first operation changes the values the second one reads. Blur before sharpen isn't the same as sharpen before blur. If a combination isn't producing what you expected, reordering it's often the fix.

It's also worth knowing that adjustments applied to an already-compressed image amplify whatever artifacts that compression introduced. Raising contrast on a heavily compressed photograph makes the blocking visible. That is why serious editing should start from the highest quality original available rather than from a version that has already been through a lossy save.