the settings for 2x downsampling about kernelgan HOT 6 OPEN

Yes
we generated x2 kernels with those parameters and downscaled the image twice or generated a x4 kernel analytically and downscaled once by 4 (2 identical possible ways)

from kernelgan.

also the parameters for 4x downsampling

from kernelgan.

The dataset was generated with some randomness.
You can either take what we generated or generate them yourself but it will not be identical.

from kernelgan.

The dataset was generated with some randomness.
You can either take what we generated or generate them yourself but it will not be identical.

Thanks. Could you tell me the values of the following parameters you used to generate 2x/4x testing images?

scale_factor = np.array([4, 4]) # choose scale-factor
avg_sf = np.mean(scale_factor) # this is calculated so that min_var and max_var will be more intutitive
min_var = 0.175 * avg_sf # variance of the gaussian kernel will be sampled between min_var and max_var
max_var = 2.5 * avg_sf
k_size = np.array([21, 21]) # size of the kernel, should have room for the gaussian
noise_level = 0.4 # this option allows deviation from just a gaussian, by adding multiplicative noise noise

from kernelgan.

See section 5.2 in the paper.
It includes all details, parameters and links

from kernelgan.

Section 5.2 doesn't specify the scale factors. Did you use the same settings (e.g. kernel width, rotation and noise) for both scale factor 2 and 4?

from kernelgan.