Question about butteraugli's threshold value

[NanyangYe]

Dear all:

There is something we are not certain about. We test butteraugli on this image.
Original:

Compressed JPEG compression quality 90

According to butteraugli's result, the distmap's value is above good threshold and even bad threshold, so the difference can be visible. But according to our experiment, human cannot observe the difference at all. And the butteraugli' results keep the same even when the quality reached 98. Is there any problems with our understanding. Many thanks.

[jyrkialakuijala]

How do you conduct the human rating experiment?

I see some difference in the darker areas where the pattern disappears into the black. There is higher saturation and the pattern disappears more quickly into the dark in the lower images. Also, the lower image has some ringing in the background pattern.

[NanyangYe]

@jyrkialakuijala The images were compared in a side-by-side experiment on a calibrated
sRGB display with the peak luminance of 220 cd/m^2, dark room. The
viewing distance was controlled so that the angular resolution was 60
pixels per visual degree.

[mqudsi]

The issue with the samples you posted isn't jpeg compression artifacts; there's a very noticeable difference in the gamma or the color range.