Since deciding on the Hudson River Valley School of art, a dataset that is both consistent in theme but varying in compositions and use of art characteristics (e.g. color, contrast), I have been exploring more analyses. If the dataset is varied enough, I believe it makes sense to consider multiple characteristics, rather than just one.
As far as color and contrast goes, I will be going ahead with the Photoshop Histogram Analysis. In my previous blog posts, I have detailed how it provides information on the Red, Green, and Blue channels of the image. I will look at varying levels of these channels throughout the artworks (using mean, range, etc.), and at the end of the survey, ask the subject which of the three colors they prefer the most. Here, results will look at a possible correlation between color preference and artwork preference. For contrast, many current measures of contrast focus on comparing the darkest and brighest points of an image and focus on extremes. Because I want a measure of average contrast throughout the work, RMS (Root mean square) contrast seems to be the most relevant, as it looks at the average values of pixels in comparison to the mean. In addition, I could also compare the representation of the colors channels with one another to look at a measure of color contrast.
While I have mentioned both color and contrast before, the primary new characteristic that I am considering is subject composition. As aforementioned, these Hudson River Valley School artworks are similar in theme; they depict mountains and vegetation, bodies of water, and the sky. These three to four components are consistent throughout most of these works. However, while the presence of these components is consistent throughout these artworks, the usage and representation of them may not be consistent, and seem to be varied. There are several aspects to consider here, including area ratio, continuity, and dividing or border lines. For area ratio, we can do a simple comparison of how much of the painting is water versus how much of the painting is sky. Compositional differences here could surely affect one's perception of these artworks (but we must consider and ask whether the subject simply prefers one over another, e.g. sky over mountains). We can also look at the continuity of these areas; is there one connected region, multiple regions, or is one region fragmented by another component? For example, there may be two small ponds (bodies of water) in a work, or a large mountain in the middle of the artwork may divide the sky into two areas. To look at the areas, I can just use Photoshop to measure the area taken up by each component. Alternatively, I could use a grid system and look at the representation around each point to suggest a specific probability of representation. This would also allow me to specify and analyze in which quadrant of the painting certain components are most common. Lastly, we can also look at the line that divides or borders two of these regions. For example, the line that divides the sky and bodies of water is very typically the horizon line. The vertical position of this line in relation to the painting's height can be noted, and may signify how much vegetation/water versus sky/mountains is depicted (another question: is there actually a relationship here?). However, we also have lines that divide the mountains/vegatation and the bodies of water. These lines are likely not horizonal and straight like the horizon line, and may have a non-zero slope or be irregular. However, by looking at the general slope and horizontal and vertical position of this line, in addition to the other area-related charactertistics, we can better find significant and relevant compositional differences between these artworks of similar theme.