Matching-blur downsampled images

This page contains images that were artificially blurred and then downsampled using a standard approach (with a Lanczos low-pass filter), and using our blur-matching model. In the later case, the images were pre-blurred according to the model predictions for the known blur factor (uniform across an image) prior to downsampling. These examples are inteded to test whether the blur-matching model gives plausible results also for the images not included in the experiment. It also illustrates the difference between standard and appearance preserving downsampling.

image: cyclist, blur=0.027948	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: cyclist, blur=0.046148	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: cyclist, blur=0.067913	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: cyclist, blur=0.093939	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: cyclist, blur=0.125061	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: cyclist, blur=0.162278	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: cyclist, blur=0.206782	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: cyclist, blur=0.26	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: coast, blur=0.027948	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: coast, blur=0.046148	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: coast, blur=0.067913	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: coast, blur=0.093939	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: coast, blur=0.125061	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: coast, blur=0.162278	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: coast, blur=0.206782	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: coast, blur=0.26	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: desk, blur=0.027948	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: desk, blur=0.046148	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: desk, blur=0.067913	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: desk, blur=0.093939	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: desk, blur=0.125061	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: desk, blur=0.162278	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: desk, blur=0.206782	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: desk, blur=0.26	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: lab, blur=0.027948	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: lab, blur=0.046148	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: lab, blur=0.067913	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: lab, blur=0.093939	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: lab, blur=0.125061	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: lab, blur=0.162278	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: lab, blur=0.206782	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching
image: lab, blur=0.26	x2 naive	x4 naive	x8 naive
	x2 blur-matching	x4 blur-matching	x8 blur-matching