1/f2 Characteristics and Isotropy in the Fourier Power Spectra of Visual Art, Cartoons, Comics, Mangas, and Different Categories of Photographs

Michael Koch, Joachim Denzler, Christoph RediesPLoS 2010 · DOI

Abstract

Art images and natural scenes share a common mathematical property: their radially averaged Fourier spectral power follows a power-law (1/f² characteristics), indicating scale-invariant properties. This study extends these findings to other man-made categories, including cartoons, comics, and manga. Through 2D power spectra analysis and principal component analysis (PCA), we demonstrate that while these images share the power-law distribution of natural scenes, they exhibit significantly higher "slope isotropy"—a more uniform distribution of structural complexity across all orientations. These results suggest that man-made aesthetic stimuli are specifically crafted to evoke pleasant visual perception by mirroring natural statistics while optimizing directional uniformity.

Introduction: The Mathematical Fingerprint of Aesthetics

What defines an image as "aesthetic"? While art history often focuses on cultural context and stylistic evolution, the field of neuroaesthetics seeks universal mathematical properties that resonate with the human visual system. A cornerstone of this research is the concept of scale invariance. In natural scenes, the statistical distribution of structural detail remains constant regardless of magnification—a property represented by a 1/f² power spectrum.

Previous research confirmed that fine art often mimics this natural scaling. However, two major questions remained: does this property extend to popular media like cartoons and manga, and can we distinguish between image categories by looking beyond simple 1D averages to full 2D Fourier spectra?

Methods and Datasets

Researchers analyzed a diverse dataset of 1,500 images, including monochrome fine art, cartoons, Western comics, Japanese manga, and various control groups such as natural scenes, human faces, and scientific illustrations.

The computational pipeline involved:

  1. Fourier Analysis: Converting images into the frequency domain to quantify "energy" (power) at different spatial frequencies.
  2. 1D Power Law Fitting: Calculating the radially averaged slope to verify 1/f² characteristics.
  3. 2D PCA: Using the first 15 principal components of 2D power spectra to determine if image categories occupy distinct statistical spaces.
  4. Anisotropy Measurement: Dividing the 2D spectra into 32 sectors to measure how mean power and slope vary across different angles (e.g., horizontal vs. vertical vs. oblique).

Key Results: The Universal Aesthetic Slope

The study found that cartoons, comics, and manga strictly adhere to the 1/f² rule, much like fine art and natural scenes. This suggests that even simplified or exaggerated media are "tuned" to the same fractal-like statistics that the human visual system evolved to process efficiently.

Fourier power spectra for cartoons, comics, and manga showing 1/f² power-law characteristics.

As shown in the figure above, the spectral power for these man-made categories decreases linearly on a log-log scale as spatial frequency (cycles per image) increases. The slopes are remarkably consistent, hovering near the -2.0 ideal. In contrast, non-aesthetic categories like scientific illustrations showed shallower slopes (~ -1.5), prioritizing high-frequency edges over naturalistic distribution.

The "Aesthetic Isotropy" Hypothesis

The most significant finding of the research concerns slope isotropy. While both art and nature tend to have more "power" in cardinal orientations (horizontal and vertical lines), art is significantly more uniform in its slope across all angles.

In natural photographs, the "texture" or complexity of the image often changes depending on the direction of observation. In contrast, aesthetic man-made images exhibit higher isotropy; the artist appears to subconsciously balance structural complexity so it remains consistent regardless of orientation. This property held true even when comparing landscape paintings to landscape photographs, suggesting it is a byproduct of the creative process rather than the subject matter itself.

Conclusion

The results indicate that artists—whether drawing a superhero or painting a masterpiece—act as "statistical engineers." They subconsciously craft images that mirror the scaling of the natural world while "smoothing out" the directional biases found in raw physical reality. While these properties are common to images designed for enjoyment, whether they are sufficient to induce aesthetic pleasure remains a subject for future investigation, particularly regarding the role of phase information and higher-order statistics.

Summary updated

Abstract

Art images and natural scenes have in common that their radially averaged (1D) Fourier spectral power falls according to a power-law with increasing spatial frequency (1/f(2) characteristics), which implies that the power spectra have scale-invariant properties. In the present study, we show that other categories of man-made images, cartoons and graphic novels (comics and mangas), have similar properties. Further on, we extend our investigations to 2D power spectra. In order to determine whether the Fourier power spectra of man-made images differed from those of other categories of images (photographs of natural scenes, objects, faces and plants and scientific illustrations), we analyzed their 2D power spectra by principal component analysis. Results indicated that the first fifteen principal components allowed a partial separation of the different image categories. The differences between the image categories were studied in more detail by analyzing whether the mean power and the slope of the power gradients from low to high spatial frequencies varied across orientations in the power spectra. Mean power was generally higher in cardinal orientations both in real-world photographs and artworks, with no systematic difference between the two types of images. However, the slope of the power gradients showed a lower degree of mean variability across spectral orientations (i.e., more isotropy) in art images, cartoons and graphic novels than in photographs of comparable subject matters. Taken together, these results indicate that art images, cartoons and graphic novels possess relatively uniform 1/f(2) characteristics across all orientations. In conclusion, the man-made stimuli studied, which were presumably produced to evoke pleasant and/or enjoyable visual perception in human observers, form a subset of all images and share statistical properties in their Fourier power spectra. Whether these properties are necessary or sufficient to induce aesthetic perception remains to be investigated.
BibTeX
@inproceedings{koch2010f,
  author = {Michael Koch and Joachim Denzler and Christoph Redies},
  title = {{1/f2 Characteristics and Isotropy in the Fourier Power Spectra of Visual Art, Cartoons, Comics, Mangas, and Different Categories of Photographs}},
  booktitle = {PLoS},
  year = {2010},
  doi = {10.1371/journal.pone.0012268},
}
Figure