Designing Color Symmetry in Stigmergic Art

Round 1

Reviewer 1 Report

In this paper, the author provided an algorithm for transforming color (called color symmetry in this paper) and combined it with stigmergy. In general, this paper is quite straightforward though I do have some complaints.

The term "color symmetry" here refers to a transformation of color in a color space. However, this term often refers to a distribution of color across image space. In particular, the distribution of color in one part of the image is a reflection of the other part across a symmetry axis (Morales & Pashler, 1999; Wu & Chen, 2014; Gheorghiu, Kingdom, Remkes, Li, & Rainville, 2016). The definition of color symmetry at the beginning of Abstract is consistent with that in the literature. However, the examples in Figure 3-7 are not. This can cause great confusion for readers.

The algorithm for color transformation given in this paper includes rotation and reflection. All equations and examples given are color transformation in a two-dimensional space. However, color is three-dimensional (hue, saturation, and luminance). The author seemed to take an arbitrary slice (HSV?) of this three-dimensional space to generate his works. A discussion on how this algorithm can be used to determine the color of a pixel in the three-dimensional space would be very helpful. 

In Introduction, the author wrote “ … the design of color symmetry discussed in this paper also gives a reinterpretation of the color study shown in Fig. 1a.”. I did not see such discussion anywhere in the paper. The author generated some Kandinsky-inspired images in Figure 3. This is very far from “re-interpretation” of Kandinsky’s work. In particular, Figure 3 looks nothing like  Kandinsky’s work. 

Author Response

Thanks for your comments on my paper. My responses are as follows.

The term "color symmetry" here refers to a transformation of color in a color space. However, this term often refers to a distribution of color across image space. In particular, the distribution of color in one part of the image is a reflection of the other part across a symmetry axis (Morales & Pashler, 1999; Wu & Chen, 2014; Gheorghiu, Kingdom, Remkes, Li, & Rainville, 2016). The definition of color symmetry at the beginning of Abstract is consistent with that in the literature. However, the examples in Figure 3-7 are not. This can cause great confusion for readers.

The works you mentioned deals with the relationships between color and symmetry perception. My work deals with applying the mathematical concept of color symmetry to generative art. Although there are some relationships between these two topics, my focus is on design, not on perception. Also, I think the examples in Fig. 3-7 are consistent with the basic definitions of color symmetry, but the images in Fig. 6 and 7 consist of multiple symmetry axes and/or rotation centers, while the most basic color symmetries (Fig. 4, but also the examples used in studying color and symmetry perception) only have a single symmetry axis or rotation center.  Nevertheless, I have added some additional examples (new Fig. 6) which show more clearly such simple color symmetries.     

The algorithm for color transformation given in this paper includes rotation and reflection. All equations and examples given are color transformation in a two-dimensional space. However, color is three-dimensional (hue, saturation, and luminance). The author seemed to take an arbitrary slice (HSV?) of this three-dimensional space to generate his works. A discussion on how this algorithm can be used to determine the color of a pixel in the three-dimensional space would be very helpful.

As I mentioned on p. 4 of the original manuscript  “We assume that the image is composed of n × m pixels and each pixel p i has a color c i in a HSV color space: c i = ( h i , s i , v i ) . In the following discussion (and also in the experiments reported later) the focus is on the hue h i while the saturation s i and the value (brightness) v i remain constant, but of course the same method can be applied to the whole HSV color space. We now define the color permutation associated with the symmetry group by a map f ( h i ), which is a mapping of the hue component of the HSV color space onto itself.”

Thus, the reviewer is right that only the H component is mapped. An application to the whole HSV could be achieved by also mapping the S and V component. However, I noticed that the maps varying the hue create a substantial variety of color shifts in the sections affected by color symmetry, particularly for the color maps involving trigonometric functions. Thus, fine-scaled and strong contrasts in color lead to substantial color distortions, which is an effect I found interesting the works of generative art I created. Moreover, if at least the pixels' saturation or brightness are preserved, the ``color symmetric'' sections of the image resembles the original image, but we also have a kind of alienation or distancing effect in terms of color, which is again an artistic effect I find interesting. Thus, my focus on hue. I have added some remarks in the revised version of the manuscript.

In Introduction, the author wrote “ … the design of color symmetry discussed in this paper also gives a reinterpretation of the color study shown in Fig. 1a.”. I did not see such discussion anywhere in the paper. The author generated some Kandinsky-inspired images in Figure 3. This is very far from “re-interpretation” of Kandinsky’s work. In particular, Figure 3 looks nothing like  Kandinsky’s work. 

I disagree with the reviewer that Figure 3 looks nothing like Kandinsky’s work as I think the basic layout is exactly the same. However, I also admit that this might be a question of artistic taste. I thus have rewritten the sentence mentioned.

Reviewer 2 Report

The authors propose to incorporate colour simmetry in a stigmergic art system. For that purpose, they model a color permutation scheme of an object according to its simmetry group, and then apply it this scheme to image generation using a stigmergic algorithm. 

The approach yelds some interesting results, namelly, in my opinion, in the pure abstract realm. (Figurative transformations are not as convincing since they sometimes resemble naive special effects.) 

The model of stigmergic nest construction appear to be somewhat limited in for the objectives of this study (altough, on the other hand, its simplicity may help to understand the effects of the color permutation scheme). The author could try different algorithms. 

The proposed future work points to relevant issues.

(Image 4 analysis seems a litlle bit confusing. In line 186, I wonder if the authors mean «In the upper right image».)

Author Response

Thanks for your comments on my paper.

The authors propose to incorporate colour simmetry in a stigmergic art system. For that purpose, they model a color permutation scheme of an object according to its simmetry group, and then apply it this scheme to image generation using a stigmergic algorithm. 

The approach yelds some interesting results, namelly, in my opinion, in the pure abstract realm. (Figurative transformations are not as convincing since they sometimes resemble naive special effects.) 

The model of stigmergic nest construction appear to be somewhat limited in for the objectives of this study (altough, on the other hand, its simplicity may help to understand the effects of the color permutation scheme). The author could try different algorithms. 

The proposed future work points to relevant issues.

(Image 4 analysis seems a litlle bit confusing. In line 186, I wonder if the authors mean «In the upper right image».)

Thanks for spotting this. I have corrected the sentence.