Artists do have a bias. They put in an average of 26% too much clouds in the sky compared to meteorological observations.
For the period 1850 to 1967, average observed cloudiness and painted cloudiness were subjected to the torture of statistical regression:
Observed Cloudiness (%) = 0.5*Painted Cloudiness (%) +26% (r^2 = 0.90)
So when oil-on-canvas indicates 70% cloud cover, read 44% cloud cover and you will be close to what the weather station observer would record. The Y-axis on the right hand side of the illustration above has the bias removed.
The observed-in-the-modern (post 1910) increase in cloudiness in North American and Europe is around 17%. The small red indicator in the illustration above illustrates a 17% increase in cloudiness. The change is consistent with the artist's impressions and expressions. If these artists are indeed faithful weather artists, then Earth's experience is one of real meaningful climate changes and we have been having them for hundreds of years. The brush-strokes tell us that there was a major increase in cloudiness from 1400 AD to 1600 AD, a decline in cloudiness to 1800 AD and increase and then a rise in cloudiness to 1967. In an earlier CED we found that all of the terrestrial LTER sites had teen-full percentages of cloudiness increases during the 20th century.