-- blur 8: I calculated the (2*n+1)*(2*n+1) filter using Gaussian formula, where σ=1. But it didn't blur much.

-- sharp (cannot remember the parameter): I extrapolated with the blur image using 2.0 as parameter. But due to the above reason, it doesn't sharpen much.

-- contrast 2.0

-- noise 0.5

-- quantize 2

-- random Dither, ordered Dither, Floyd Steinberg Dither: They're all quantized to 2 bits.
I don't know why my dithering are all brighter than what the sample program generates. I know I'm doing right, but still wondering why they're different.

-- extract channel 0

-- saturation -0.5

-- crop 450 320 -500 -200:  I considered the negative width and height values valid but of opposite direction

-- edge Detect: I only used a 3*3 filter, so it's not so obvious

-- fun: cosine, it can be used as reflection of water in animation

-- rotate 240

-- scale 0.5 1.5

-- art:  emboss, using bump map+contrast(4)+blur(2)  It's quite trivial, but I like the effect:-)