Perception and Preference Analysis of Fashion Colors: Solid Color Shirts
Round 1
Reviewer 1 Report
The presented article analyzes the results of psychophysical experiment based on solid color shirts. It studies the perception and preference on fashion colors, including the aesthetic differences and similarities between male and female. Still there are some questions to clarify:
1. How it was ensured that computer screen did not distort colours?
2. Were experts trained before experiments for such type of evaluation?
3. What was the duration between image displays?
Author Response
Response to Reviewer 1 Comments
Point 1: How it was ensured that computer screen did not distort colours?
Response 1: In the experiment, this study can use two Apple 27-inch iMacs (model: MD095CH/A) as the display. The two monitors were under the same indoor environment and light source. During the experiment, other interference factors are avoided and the same brightness and display calibration are set for the two monitors as far as possible to mitigate color distortion.
Point 2: Were experts trained before experiments for such type of evaluation?
Response 2: No, because these 34 Chinese observers already have basic knowledge about color or design.
Point 3: What was the duration between image displays?
Response 3: During the test, an image was displayed for 10 seconds to avoid interference with the next image.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The article tries to bring too many concepts together and in doing so does not build a strong case for the research. There are weaknesses, such as the suggestion of large amounts of inventory held by brands, this is no longer the case due to fast fashion models and short production runs. There is no consideration of the fashionability aspect of any of the colours tested. At times, the terms clothing preference and colour preference are inter-changed. As the study is not concerned with style why is the term clothing preference used? Plain and splendid are not robust measurement variables. There is no critical analysis of the related work mentioned in section 2. Are you trying to develop a model as suggested on lines 158-160? If so, more substantial research is required. The theoretical underpinning mentioned on line 186 is not discussed, why is this body of work important? On line 187 you claim to 'discover factors affecting fashion colour perception and preference, and the specific mechanism'; but you do not actually attempt to do this. Why are you simulating colours? Surely this is no different to measuring the preference of samples. I suspect that you are simulating for the purpose of 'aiding clothing simulation technology' (lines 191/192) but we are now on page five and there has been no previous mention of this - this point goes back to you not developing a strong case for this type of research experimentation. Some of the figure numbers do not match the text. How were the clusters actually developed? Line 199 'the dark' - this is not a colour! Why add a further 22 colours and how did you decide which colours to add? There appears to be no systematic approach to the research. On line 222 you mention previous studies but offer no discussion, how is this work meaningful to this study? The measurement terminology is rather loose: 'dislike extremely', dislike very much' - why not use on a scale of 1 to 7? The number of participants is too few to consider age and gender differences. I assume that no-one was tested for colour blindness! Easy-to-match is an obvious response as shirts are worn with other items. You make assumptions on lines 322-333 regarding an individual's knowledge on colour matching. Colour does not necessarily make a person look slim, other factors are involved such as style, cut, fit, surrounding colours; therefore people are making assumptions about the function of colour. You then mention texture perception - this has not been considered previously - yet this is a critical aspect of clothing design and one that could influence colour preference, though your research misses this entirely as you are using simulations. Line 359 - blue is a known colour preference but not necessarily any blue. You need to develop a much tighter reasoning for the work you have conducted and a more substantial number of particpants are required to be able to substantiate any claims.
Author Response
Response to Reviewer 2 Comments
Point 1: As the study is not concerned with style why is the term clothing preference used? Plain and splendid are not robust measurement variables.
Response 1: Color, as an important fashion feature, can make great influences on the consumers’ clothing preference and purchase intention and that was modified in the revised version.
We have already use “Plain: Gaudy” to replace “Plain: Splendid”. Additionally, “Plain: Gaudy” [26] sometimes is replaced by “Plain: Splendid” [32] or “Plain: Luxury” [48] or “Plain: Gallant” [22]
Point 2: There is no critical analysis of the related work mentioned in section 2. Are you trying to develop a model as suggested on lines 158-160? If so, more substantial research is required.
Response 2: We have already added more details about the Hierarchical “Feed-Forward” Model.
Among them, the first real mathematical model to bridge the gap between low-level statistical features and aesthetic emotions aroused by visual textures is proposed by Thumfart and colleagues [23]. Using the results of a psychological experiment, they modeled the relationship between computational texture features and their aesthetic properties. In contrast to previous approaches, this layered model provides insight into the hierarchical relationships involved in the aesthetic experience of texture properties. The structure of the hierarchical feed-forward model of aesthetic texture perception consists of three layers: Affective layer (How can the texture be described), Judgment layer (What the object says about itself) and Emotional layer (What do I feel when interacting with the texture) [23].
Point 3: The theoretical underpinning mentioned on line 186 is not discussed, why is this body of work important? On line 187 you claim to 'discover factors affecting fashion colour perception and preference, and the specific mechanism'; but you do not actually attempt to do this.
Response 3: We have already changed this sentence into: Using existing research findings of aesthetic perception models [37,38] and colour emotion [26,29,39-44] as a theoretical basis, this study aims to discover the specific mechanism of fashion colour perception and preference.
Point 4: Why are you simulating colours? Surely this is no different to measuring the preference of samples. I suspect that you are simulating for the purpose of 'aiding clothing simulation technology' (lines 191/192) but we are now on page five and there has been no previous mention of this - this point goes back to you not developing a strong case for this type of research experimentation.
Response 4: We just investigate the color preference of 3D clothing with the help of existing 3D clothing simulation software, like CLO 3D software. So, we do not focus on the aiding clothing simulation technology, just use the existing 3D clothing simulation software.
Point 5: Some of the figure numbers do not match the text.
Response 5: We have already corrected them in the revised version.
Point 6: How were the clusters actually developed? Line 199 'the dark' - this is not a colour! Why add a further 22 colours and how did you decide which colours to add? There appears to be no systematic approach to the research.
Response 6: We collect 480 colors from solid shirts of different clothing brands (shown in Figure 8(a)) and use Mean Shift clustering algorithm [47] to classify them into 19 clusters in CIELAB color space (shown in Figure 8(b)). Nineteen cluster centers and their corresponding numbers of clusters are shown in Figure 8(c), and the white is the most popular color, followed by the black and red. The 41 colors are selected to cover a reasonably wide range of lightness (L*), chroma (C*) and hue (hab) in CIELAB color space, and meanwhile to cover the clustering centers of existing solid shirts’ colors.
Point 7: On line 222 you mention previous studies but offer no discussion, how is this work meaningful to this study? The measurement terminology is rather loose: 'dislike extremely', dislike very much' - why not use on a scale of 1 to 7?
Response 7: We collect multitude of adjective pairs related to solid color shirts, and 11 adjective pairs are selected on the basis of a questionnaire survey conducted by 10 participants, each asked to select 10 adjectives appropriate to describe visual impressions of solid color shirts. Among them, 9 adjectives, including “Cold: Warm” , “Heavy: Light” “Passive: Active”[24,25], “Dirty: Clean”[24], “Tense: Relaxed”[24], “Plain: Gaudy”[26], “Traditional : Modern”, “Masculine : Feminine”[24], “Dislike: Like”[24], are from existing studies. Besides, another 2 scales “Slim-Look: Fat-Look” and “Hard-to-Match: Easy-to-Match” are from the participants’ lists which have the high frequent.
Additionally, “Cold-Warm” [22,23] , “Plain: Gaudy” [26], “Traditional : Modern”[22,32] sometimes are replaced by “Cool: Warm” [24-26,32], “Plain: Splendid” [32] or “Plain: Luxury”[48] or “Plain: Gallant”[22], “Classic: Modern”[24] or “Old-fashioned: Modern”[23] or “Ancient: Modern”[22] respectively.
During the main experiment, each word pair was presented using a seven-step, forced choice scale to measure the observer response. Taking “Dislike: Like” as an example, the 7 steps included “Dislike Extremely”, “Dislike”, “Dislike Somewhat”, “Neither Like nor Dislike”, “Like Somewhat”, “Like”, “Like Extremely”. A certain number of female and male observers are asked to fill in the same 7-step mandatory scale with choice questions to assess the 41 shirts.
Point 8: The number of participants is too few to consider age and gender differences. I assume that no-one was tested for colour blindness! Easy-to-match is an obvious response as shirts are worn with other items. You make assumptions on lines 322-333 regarding an individual's knowledge on colour matching. Colour does not necessarily make a person look slim, other factors are involved such as style, cut, fit, surrounding colours; therefore people are making assumptions about the function of colour.
Response 8: All of the participants were tested for color blindness and were color-trained design students. They have been trained in color for at least a year so this is an expert experiment. This is not a small number of people, because the questionnaire requires everyone to do more than 2.5 hours of experiments, including the rest due to aesthetic fatigue. As for the two scales "slim-look: fat-look" and "hard-to-match: easy-to-match", we have answered in the last question.
Point 9: You then mention texture perception - this has not been considered previously - yet this is a critical aspect of clothing design and one that could influence colour preference, though your research misses this entirely as you are using simulations.
Response 9: The hierarchical feed-forward theory of aesthetic perception [22,23] takes both image texture and color into consideration. This paper just focusses on the colors of solid color shirts and dose not consider the clothing style, texture, and materials, which is our research limitations.
Point 10: 10. Line 359 - blue is a known colour preference but not necessarily any blue.
Response 10: The blue color whose hab is about 270 is more popular compared with other colors when their chroma C* and lightness L* are same.
Author Response File: Author Response.docx
Reviewer 3 Report
General comments
This is an interesting and potentially valuable contribution to the understanding of consumer preferences.
The article presents the results with a focus on the statistics. Although this is understandable from a methodological perspective there is less regard for readers from a fashion design or development perspective who need more interpretation of these statistical results. The article would benefit from additional analysis that discusses the application and implications of the findings. More specifically, the results have not been discussed in relation to consumer preference, i.e. how can these results be applied in practical term? E.g. can you probe this notion of easy to match? What did the participants understand by this and is it a concept that is subjective or can it be objectively tested? Were there certain colours that were found to be easy/hard to match? E.g. If you introduce a system that relies on statistics and averages, is there not a danger that it will discourage experimental use of colour and that use of colour will become risk aversive and uniform or in other words there will be a lack of choice for the consumer? Particularly in relation to this notion of hard to match colours.
It is also important to note the global nature of fashion and that colour preference is culturally influenced. These tests therefore may have different results in different geographical regions.
Specific comments
Line 42/43 please correct the grammar and sentiment. ‘fast fashion is becoming a killer of environment’
Line 51 – correct - ‘this study believes’ – a study cannot believe.
Line 63 – explain how environmentally fuel is relevant to dyeing and finishing
Line 63-64 – clarify and strengthen this question I.e. make a more robust statement about the relationship between the aim of the study and its potential to decrease environmental pollution
Line 79 correct the grammar– ‘ever permanently changing trends’ remove permanently
Lines 85-87 sentence starting Due to the size – This seems to be about the need to store large volumes of stock rather than hoarding it. correct this by using appropriate terminology.
Line 90-92 – This sentence starting ‘consumer clothing preference research’ can be deleted because it is grammatically incorrect and is actually saying the same thing as the following much clearer sentence starting ‘by understanding’,
Line 96 – please change – ‘from the above point of view’ to a sentence that sums this point of view up.
Line 100 – clarify what is meant by – ‘it is ubiquitous and a source of information’
Line 103 – correct grammar ‘make great influences’
Line 113 – delete ‘on the other hand’ it is used incorrectly
Lines 113-114 Is it possible to fully ‘know’ what consumer’s preferences for colour are. Surely these preferences are subject to change. Please modify this statement to reflect the complexity of consumer preference
Lines 123 – 124 correct the grammar
Experimental details
Please explain how the colours were collected form the shirts. i.e. were the physical shirts used or images of shirts?
Page 207 – match figure numbers in text and caption
Line 199 – clarify what is meant by ‘the dark’ should it be dark red?
Line 235 – explain why all the participants had a basic knowledge of colour and design. Does this mean they were experts or not experts? Please clarify as this is confusing.
Lines 235 – 236 – There are a number of grammatical errors – please correct
Figures – having three figures in one is difficult for the reader. Can these be separated into three separate figures.
Line 256 – correct grammar – ‘females are more selective’ is a preferable term
Lines 269 – 271 – please indicate if this is making a comparison between age and gender preferences. This is not clearly stated and so is confusing for the reader.
Line 275 – can you be sure for this limited study that younger people are more sensitive to colour? Is it merely an indication of this sensitivity
Line 278 – 290 should be moved to the methods/experiment section to explain the tools that have been used for the experiments.
Lines 290-299 The article would benefit from a much more detailed explanation of the universal models of colour emotion (Ou et al), within the literature review
Figure 8 – is very small and hard to read.
The term splendid is a strange one. Is this a translation? Please clarify if this experiment was conducted using English words.
Line 333 The hierarchical feed forward theory of aesthetic texture perception needs to be explained within the literature review. What is it? when and why did it originate and by whom?
Author Response
Response to Reviewer 3 Comments
Point 1: Line 42/43 please correct the grammar and sentiment. ‘fast fashion is becoming a killer of environment’.
Response 1: In this revision, 'fast fashion is becoming a killer of environment' has been changed to 'fast fashion is becoming a cause of environmental destruction'.
Point 2: Line 51 – correct - ‘this study believes’ – a study cannot believe.
Response 2: In this revision, ‘this study believes’ has been changed to ' this study find '.
Point 3: Line 63 – explain how environmentally fuel is relevant to dyeing and finishing.
Response 3: In the previous writing, the dye was written as fuel, this paragraph has been deleted.
Point 4: Line 63-64 – clarify and strengthen this question I.e. make a more robust statement about the relationship between the aim of the study and its potential to decrease environmental pollution
Response 4: This paragraph is not well written, and more evidence may be needed. Now this paragraph has been deleted, and we will try to continue to discuss this topic in future studies.
Point 5: Line 79 correct the grammar– ‘ever permanently changing trends’ remove permanently
Response 5: In this revision, 'permanently' has been removed.
Point 6: Lines 85-87 sentence starting Due to the size – This seems to be about the need to store large volumes of stock rather than hoarding it. correct this by using appropriate terminology.
Response 6: In this revision, this sentence has been changed to 'Due to the size, style, color, seasonality and other factors of clothing products, a large number of products are often produced to meet the needs of customers, resulting in inventory. '
Point 7: Line 90-92 – This sentence starting ‘consumer clothing preference research’ can be deleted because it is grammatically incorrect and is actually saying the same thing as the following much clearer sentence starting ‘by understanding’
Response 7: In this revision, this sentence has been deleted.
Point 8: Line 96 – please change – ‘from the above point of view’ to a sentence that sums this point of view up.
Response 8: The three paragraphs from line 96 to line 110 have been reexpressed:
‘As an important fashion feature, color can influence consumers' clothing preference and purchase intention. Color is light carried on wavelengths absorbed by the eyes that the brain converts into colors that we see [9]. Colors can have a powerful psychological effect and evoke emotions and therefore it can change our behavior too, which is also supported by science, as color addresses one of our basic neurological needs for stimulation [10]. For consumers, the visual appearance and color is placed above other factors when shopping (1% sound / smell, 6% texture, 93% visual appearance) [11]. The consumers make up their minds within 90 seconds of their initial interactions with either people or products and about 62~90% of the assessment is based on colors alone [9]. For retailers, shopping is the art of persuasion. Though there are many factors that influence how and what consumers buy. 85% of shoppers place color as a primary reason for why they buy a particular product [11]. However, a great deal is decided by visual cues, the strongest and most persuasive color.’
Point 9: Line 100 – clarify what is meant by – ‘it is ubiquitous and a source of information’
Response 9: We have already deleted it in the revised version.
Point 10: Line 103 – correct grammar ‘make great influences’
Response 10: In this revision, this sentence has been changed to ' As an important fashion feature, color can influence consumers' clothing preference and purchase intention. '
Point 11: Line 113 – delete ‘on the other hand’ it is used incorrectly
Response 11: We have already deleted it.
Point 12: Lines 113-114 Is it possible to fully ‘know’ what consumer’s preferences for colour are. Surely these preferences are subject to change. Please modify this statement to reflect the complexity of consumer preference.
Response 12: To know the consumers’ preference of current fashion color can help companies to reduce the stock, improve the sale and enhance the competitiveness.
Point 13: Lines 123 – 124 correct the grammar
Response 13: We have already corrected it.
Point 14: Please explain how the colours were collected form the shirts. i.e. were the physical shirts used or images of shirts?
Response 14: The shirt images are used to extract colors.
Point 15: Page 207 – match figure numbers in text and caption
Response 15: We have already matched figure numbers in text and caption in the revised version.
Point 16: Line 199 – clarify what is meant by ‘the dark’ should it be dark red?
Response 16: The black is used to replace the dark.
Point 17: Line 235 – explain why all the participants had a basic knowledge of colour and design. Does this mean they were experts or not experts? Please clarify as this is confusing.
Response 17: They are all design majors and have received color training or courses related to color during their studies. We have added this sentence in the article.
Point 18: Lines 235 – 236 – There are a number of grammatical errors – please correct
Response 18: We have already corrected it.
Point 19: Figures – having three figures in one is difficult for the reader. Can these be separated into three separate figures.
Response 19: Putting the three images together makes it easier for the reader to compare them. We used a higher definition image in this revision so that the reader can zoom in on them.
Point 20: Line 256 – correct grammar – ‘females are more selective’ is a preferable term
Response 20: We have already corrected it.
Point 21: Lines 269 – 271 – please indicate if this is making a comparison between age and gender preferences. This is not clearly stated and so is confusing for the reader.
Response 21: This should be ‘ages’, we changed it. The sentence is ‘which indicated that different ages have nearly consistent aesthetic emotions for different solid color shirts.’
Point 22: Line 275 – can you be sure for this limited study that younger people are more sensitive to colour? Is it merely an indication of this sensitivity.
Response 22: This inference is based on our experimental results, which may require more experiments to prove it, and this is the topic we will continue to discuss in future studies.
Point 23: Line 278 – 290 should be moved to the methods/experiment section to explain the tools that have been used for the experiments.
Response 23: We have already move it to methods/experiment section.
Point 24: Lines 290-299 The article would benefit from a much more detailed explanation of the universal models of colour emotion (Ou et al), within the literature review
Response 24: We have added a much more detailed explanation of the universal models of colour emotion: Ou et al and Xin et al find “heavy/light”, “passive/active” and “cold/warm” have a high correlation with L*, C* and hab of color patches in CIE LCH color space respectively [24-26].
Point 25: Figure 8 – is very small and hard to read.
Response 25: We have tried our best to make it biggest in the paper.
Point 26: The term splendid is a strange one. Is this a translation? Please clarify if this experiment was conducted using English words.
Response 26: We have already use “Plain: Gaudy” to replace “Plain: Splendid”. Additionally, “Plain: Gaudy” [26] sometimes is replaced by “Plain: Splendid” [32] or “Plain: Luxury”[48] or “Plain: Gallant”[22].
Point 27: Line 333 The hierarchical feed forward theory of aesthetic texture perception needs to be explained within the literature review. What is it? when and why did it originate and by whom?
Response 27: We have already added more details about the Hierarchical “Feed-Forward” Model.
Among them, the first real mathematical model to bridge the gap between low-level statistical features and aesthetic emotions aroused by visual textures is proposed by Thumfart and colleagues [23]. Using the results of a psychological experiment, they modeled the relationship between computational texture features and their aesthetic properties. In contrast to previous approaches, this layered model provides insight into the hierarchical relationships involved in the aesthetic experience of texture properties. The structure of the hierarchical feed-forward model of aesthetic texture perception consists of three layers: Affective layer (How can the texture be described), Judgment layer (What the object says about itself) and Emotional layer (What do I feel when interacting with the texture) [23].
Figure 2. The Structure of the hierarchical feed-forward model of aesthetic texture perception. [23]
Author Response File: Author Response.docx
