watercolors    past paintings   notes on composition   bio/reviews   contact  







The following is about a visual property of pictures that has been forgotten in the 20th century. If the reader would like to do the study described below, please write me for the images.


We say that we are look at a painting when we mean that we are looking into a painting. It is as if we were looking into a world either realistic or abstract created by the painter. The eye travels rapidly from form to form to make sense out of the self-contained image inside this delimited space. Pictorial composition is an artistic and art critical term that is concerned with the arrangement of forms, colors and tonalities so that the view creates feelings of unity, harmony and balance. Artists have another way of looking at pictures which is almost never discussed unless an actual picture is being viewed and so understanding its existence remains largely hidden outside of art studios. This is the view of a painting as a visual object. When a picture is visually successful, the artist or viewer can only say something to the effect of “that’s it” or “you’ve got it.”  There is no descriptive word or metaphoric expression for this quality. As noted by Burgin:  “We know very well what good composition is—art schools know how to teach it—but not why it is; 'scientific’ accounts of pictorial composition tend merely to reiterate what it is under a variety of  different descriptions (eg, those of Gestalt psychology)1


The first historical description of this second way of looking at a picture and the concept of pictorial composition associated with it was provided by Roger de Piles. a French painter, diplomat and art critic at the end of the 17th Century2.  He described an effect which was immediate and intense with the feeling that one could see the whole painting as a unified image. Pictorial balance, harmony and unity were inherent in this effect which he felt was of primary importance as the object of the painter. He did not view the traditional balancing of forms in the painting or the need for these forms to lead the viewer to discover the subject as important3.


De Piles was unable to  give a precise description of this “overall effect” or a method for achieving it. Nonetheless his books were very well known to artists throughout the18th century.  In the 19th century his views are reflected by a well known statement of Eugene Delacroix writing in the 1830's: "There is a kind of emotion which is entirely particular to painting: nothing (in a work of literature) gives an idea of it. There is an impression which results from the arrangement of colours, of lights, of shadows, etc. This is what one might call the music of the picture...you find yourself at too great a distance from the painting to know what it represents; and you are often caught by that magic accord.4"


The author has created computer images that exhibited this effect. The type of balance used to create this effect was determined, and a study was done proving that paintings which exhibited this balance were recognized by some subjects as distinctly different from other identical paintings that did not exhibit this effect. Comparing paintings which are slightly different with one another does not prove that there is anything distinctive about them and the majority of subjects cannot recognize the effect described above. Therefore, a technique was devised to compare identical paintings differing only by this state of balance. 


The description of how the author created this picture is based on the concept that a picture is an object.5 It is a subjective conceptual view of picture making which does not start from empirical thinking and is probably only appreciated by artists who create pictures. Previously, the author had in the course of creating paintings noticed that bilateral tonal balance was necessary to achieve a balanced painting, based on a gray scale and not color. In addition the top half had to be somewhat darker that the bottom half. The nature of this balance was assumed to be a center-of-mass type of balance because that seemed logical. All previous work has been from a gestalt psychology point of view and has discussed balance as a summation of forms or tones analogous to a center-of-mass type of calculation. Good composition was said to place the center of gravity at the geometric center at or along a major axis. 6,7,8,9,


When the images created by the author and unbalanced variants were analyzed in Matlab, their center-of-mass varied equally from their geometric center and did not distinguish in any way the images exhibiting the particular artistic effect under consideration. After prolonged analysis the balanced images did correlate well with the average luminance of each quadrant of the painting as described above. The upper left and right quadrants had to be balanced as well as the lower left and right quadrants.  It was further noticed that the top bottom luminance ratio when viewed at eye level was between 0.90 and 0.95. This is the luminance balance criteria. 10


 Any mathematical determination of balance including a center-of-mass concept of balance presumes that a picture is being seen as an object. The center-of-mass approach is probably related to how the eye moves around a picture when looking at forms in the picture. Hirsch & Mjolsness, 1992 found that a center-of-mass computation best explained the ability of the eye to notice slight differences in changes in the positions of dots in a non-delineated area. The center-of-mass concept emphasizes the center from which all calculations must be made. Paintings by definition are delineated areas, and the quadrant luminance balance approach utilizes the frame as its point of reference. 11


Objective and procedure


The objective is to prove that a painting created using the luminance balance criteria is indeed special and can be distinguished from the same painting that does not meet this criteria.12a A study was devised to compare identical paintings based on the idea that the balance was that of a painting as a physical object.  Thus the study set out to comparing two physical objects, one balanced and the other unbalanced, both displaying the same visual painting. Based on experience, it was thought that there were a limited number of people who could distinguish between the two.


When looking at a framed picture, we know that it is simply a framed picture. However if a picture is an object, the eye sees the frame as part of the object. With respect to the test situation it was observed that when viewing a picture on an LCD monitor, a balanced picture with a white frame on a black ground could be seen as balanced or unbalanced depending on whether the totality of the picture and frame met the luminance balance criteria.


The basic setup was that a balanced framed picture is compared with an unbalanced framed picture. The internal picture is identical but the white frame of the balanced framed picture is modified with black bars on one or two sides. When the two are viewed sequentially on a black background, the black bars blends into the background in such a way that the position of the central image remains constant and only the frame changes causing the image – the image plus frame – to be balanced or unbalanced. .


Using Photoshop a balanced 7.3 x 10 inch internal image is surrounded by a white frame creating a balanced 8.5 x 11 inch picture. Part of the white frame was replaced with black bars to create a second image. These are images 1,2; 9,10;13,14; 15,16.  (See figure I)  Image 5 and 6 are different. A 7.3 x 10 inch image was enclosed with an irregular white border. The image was modified so as to render the irregularly framed image balanced according to the luminance criteria. White or black bars were then added so that the imaged border was 8.5 by 11 inches.  The central image was identical in both cases. A third type of image comparison was created in which the central 7.3 x 10” image was unbalanced according to luminance balance criteria and placed in a symmetric 8.5 x 11” frame creating a combined image that was unbalanced. This was compared to the same framed image rendered asymmetrical and similarly unbalanced by black bars. It is extremely difficult for subjects to see a difference between two states of luminance imbalance so that a perceptive person sensitive to luminance balance should see these two images as the same. Images pairs 3,4; 7,8; 11,12. These were the control images


figure 1 copy.jpg

Figure 1 


       Eleven subjects viewed the 8.5 x 11” images on a color calibrated Sony monitor SDM-HS53 at a distance of 24 inches. The windows native photo program “Photos” was used for viewing the images which provided the black background for viewing the images. An additional forty seven subjects viewed the images on a black ipad using a color correcting application ColorTrue maintaining the same effect of a constant picture and a changing border. These images were 14.6 x 19.1 cm or somewhat less on the iPad. 12


The subjects were shown the first pair and told that the central image was identical and that the only difference was the white frame. They were told that this would be the same for all subsequent pairs and that some of the pairs of central images would be indistinguishable to everyone while some pairs might appear different to some even though they were in all cases identical. They were then shown the images in sequence and asked if they saw a difference in the central image of each pair. If they saw a difference, they were asked if they could explain. They could switch back and forth within the pair as much as they wanted. The analogy of the monophonic and stereophonic recordings was explained to those subjects who asked how they could see a difference in identical pictures. Subjects were drawn from people associated in any way with a New York City art school. Many of the subjects were painters and sculptors but some were workers or family of students.


The subjects were asked if they perceived any difference in the images and not a difference in the quality of the images because the latter is a question of taste. Comparing two images, even identical ones, inevitably lead to subjective questions of taste such as which image is better or preferable to another. Many subjects seemed to view the study as a measure of visual competence.


The luminance values of the pictures’ quadrants are listed in table 1 as if they were seen without the black borders. This is the order in which they were shown.



Upper Left quadrant

Upper right quadrant

Lower right quadrant

Lower Left quadrant

Luminous Ratio upper half to low half 10

Image 1






Image 2





0.928 balanced

Image 3






Image 4






Image 5






Image 6





0.929 balanced

Image 7






Image 8






Image 9






Image 10





0.928 balanced

Image 11






Image 12






Image 13






Image 14





0.923  balanced

Image 15





0.924  balanced

Image 16






Image 17






Image 18






Image 19






Image 20







Table 1

As a point of reference, the actual picture in image 1 without the white border has luminance values of URL 122.21, ULQ 122.27, LLQ 131.76, LRQ 131.89 and a tonal ration of 0.928. Luminance values are between 1 and 255 on an RGB scale as determined by photoshop.10



The number of correct responses identifying whether there was a difference or not is listed in table 2.


Correct Answers











No. of responders












Table 2

Assuming a probability of a correct answer to be 0.5, the probability of getting these results is <0.0006


This is from a population including many painters. It is probable that people with this ability or desire are attracted to painting due to this sensitivity. There are also a large number in this mostly artistic population who cannot see the effect as presented in computer images. (see below)


When a difference was perceived, the subject was asked to describe it. Differences were seen in pairs that were both balanced and unbalanced. That is in pairs where the writer thought there could be an observed difference and in control pairs. The most commonly described difference was that colors were more intense or of a different hue. Other differences observed were that one was more vibrant or had more depth. In some, objects were seen as larger, closer or farther away.


An additional smaller study was done with two subjects who had observed the effect in the first study and the author. Images with a tonal balance ratio from 0.843 to 1.024 were compared. These images differed from the first group by having a simple gray bar above or below the picture as in figure 3. The luminosity of the bar within the border and thus the total vertical pictorial luminance balance could be easily controlled. It was determined that a balanced image viewed parallel to the eye had a luminance ratio of between 0.90 and 0.95, the top half with respect to the lower half. The effect degrades slowly until it ceases to become noticeable. The exact cut off point depends on the sensitivity of the subject and is arbitrary for any given person.



Figure 3




Pictures meeting the luminance balance criteria art to visually coherent. The criteria explain the effect noted by artists that a change from one viewing situation to another can create or destroy the effect. Changing the temperature of the light or the position of the work will change these ratios. Viewing a painting on a desk flat before the viewer will be entirely different from viewing it in a vertical position at eye level as will viewing it vertically above or below eye level if the illumination is constant. This is not normally noticed because it is difficult to distinguish between different degrees of two relative incoherent states.


There are some exceptions to the luminance balance criteria in paintings. Paintings composed of many different forms that cannot be condensed into a smaller number of larger forms can force the eye to focus on them and not the entire picture especially when seen up close.  Paintings composed of geometric forms when created according the luminance balance criteria do not allow for freedom of eye movements. They do seem to allow to some minor degree a global view of the painting. The eye movement effects of coherent paintings will be discussed below.


The present study was done with pictures that strictly meet the luminance balance criteria in computer images. It is rare to observe the pure phenomena in a museum. Good partial coherence is very evident but impossible to define at present. While the study indicates that a relative minority might observe perfect coherence under the conditions of this study, there is no reason to believe that larger numbers of people cannot appreciate it or good partial coherence. It is a very important aspect of the aesthetic effect of paintings.


          One can speak of an incoherent image or an image balanced in such a way as to make it relatively unpleasant to look at. About these images it has been said that the eye has difficulty visually entering the picture or that the eye can too visually easily leave the picture. As Burgin has noted:


 Composition' (and indeed the interminable discourse about composition - formalist criticism) is therefore a means of prolonging the imaginary force, the real power to please, of the photograph, and it may be in this that it has survived so long, within a variety of  rationalisations, as a criterion of value in visual art generally. 1


 With respect to incoherence, one cannot avoid questions of taste. In the modern period a markedly incoherent painting seems to be at times an aesthetic intention. In the same way that other classical ideas of the beautiful have been rejected by modernism, this aspect of composition has also been rejected. A coherent painting can easily be rejected as just pretty unless endowed with other qualities.


Proposed Neuropsychological explanation of findings: Pictures stimulate in a special selective manner the magnocellular visual pathway


The first mode of looking at a painting or looking in a painting at the various forms can be called the saccadic mode: the eye jumps from form to form. This is the normal way of looking at the world when it is not moving, and it is the way images are viewed in all the studies concerning a center of mass effect.  The second mode in which the painting is seen as an object involves a different mode of vision.


The second mode has several characteristics as described by the earliest of recorded investigators Roger de Pile and Eugene Delacroix. It is seen at first glance, and the painting is appreciated without seeing the details. The gaze seems to take in the whole picture, and the eye can move easily around the image. This evokes the feelings of balance, unity, and harmony.


Almost all visual information is simultaneously sent to various vision centers in the brain through two different visual streams: the parvocellular pathway and the magnocellular pathway. The first is for high level object resolution, precise location, texture and color. This is the traditional way of viewing pictorial composition.  The type of visual information processing with a balanced picture corresponds to the rapid responses of the magnocellular stream. This pathway is very sensitive to luminance change and not to color. It has low spatial frequency so that forms cannot be delineated precisely, and it is highly sensitivity to motion. Moving scenes and objects seen in peripheral vision are processed first through the magnocellular pathway.12 When a picture achieves a particular state of tonal balance, it would seem that this non-moving object stimulates the magnocellular pathway in a particular way such that some people who are sensitive to this effect notice that it is distinctly different in some difficult to describe manner.


Paintings are not found in nature, and there is no reason that the eye would look at them any differently than any other object unless they did in some way relate to an important experience. A bilateral tonally balanced circumscribed form with the upper half somewhat darker might be interpreted as a schematic face. The fast magnocellular stream directly connects to facial recognition centers including centers evaluating the emotional content of facial expressions. This high level evaluation would direct further investigation by lower level vision centers.  Thus a scene could be scanned using predominantly peripheral vision with attention directed to an embedded face or movement.13,14,15, If a painting were seen on first approximation to be a face, it would induce the feeling that we want to investigate that particular picture rather than just move on.


The continued stimulation of the magnocellular pathway which is contrary to what the eye would expect of a non-moving object also induces the feeling that the eye can take everything in at a glance. This evokes a feeling of pictorial unity while permitting the eye to smoothly roam through the picture. It can evoke a feeling of surprise, a visceral experience, which is at the core of an aesthetic experience. Surprise, whatever the cause, indicates that that the experience has a particular quality that we cannot relate to other than using another similar surprise as metaphor. Like all “surprises” an aesthetic surprise cannot be repeated at will. and that one has to learn to recognize that one is experiencing it or, conversely, not feeling it when other emotional and intellectual factors are present.



1 V. Burgin p. 150


2  de Piles 1708


3  Puttfarken 1985 p. 40


4  ibid p. 121,


5  The author conceptualized the image as a rectangular balloon with gray dark and light areas or forms imprinted on it. Light areas raised the corresponding side while dark areas lower it. Using Photoshop, corrections could be made in the image until at the point of perfect balance, the colored balloon gives the impression of being suspended against the wall without twisting or falling. Artists sometimes use the technique of watching where the eye travels in the painting to indicate the “trouble spots.” Whether or not these are trouble spots, observation of eye movements is an important aspect of analyzing a painting. The author uses eye movements to determine how difficult it is to move from one place to another within the painting. In other words, the subjective difficulty of a saccadic movement was used or inversely the ease of moving smoothly through the painting. A balanced painting gives the feeling of allowing the eye to slide through it. Binocular vision is necessary to observe this.


6  A well known description of this can be found in Poore 1903 chapter 3.


7  The subject was investigated by Arnheim, 1974, 1982. He viewed pictorial forms as having visual force effects on the viewer analogous to magnetic fields. Every painting had a center of balance of these”field effects.” (1974 p. 20)  He insisted that “in a  balanced composition all such factors as shape, direction, and location are mutually determined in such a way that no change seems possible, and the whole assumes the character of ‘necessity’ in all its parts. An unbalanced composition looks accidental, transitory, and therefore invalid. “ 


8 The most interesting studies in so far as they reveal how people conceive of image balancing were done by Locher et al 1998  who observed students of industrial design engineering as they created images composed of geometric shapes. They noted that the students' compositions were balanced around the center corresponding to the geometric center.  Locher et al 2001  in a similar study with advanced students in an Academy of Fine Arts observed that they did likewise


9  A good review of the studies concerned with pictorial composition as a balance of mass effect can be fnhjnound in Gershoni 2011


10 Luminance of each quadrant was determined by Photoshop. The equation is 255-(luminance of upper half))/SUM(255-(luminance of lower half)) where the luminance of an RGB image is measured between 1 and 255.  The monitor and the iPad images were balanced using an X-rite ColorMunki color calibrator


11  Hirsch & Mjolsness, 1992


12 Purves 2001 p.275

13 Kveraga 2007 p.13232

14 Tobimatsu 2012

15 Hietanen 1992




Burgin V Thinking Photography MacMillan Press London 1982

Arnheim R. Art and Visual Perception  Berkeley, University of California Press 1974


De Piles, Roger Cours de peinture par principes ( The Principles of Painting) , Jacques Estienne 1708


Gershoni  S Hochstein S “Measuring pictorial balance perception at first glance using Japanese Calligraphy” 2011  i-Perception Volume 2 pp. 508-527


Hietanen J K Perrett D I Oram M W Nemspm P J Dittrich W H  1992 The effects of lighting conditions on responses of cells selective for face views in the macaque temporal cortex  Exp Brain Res 89:157-171 (1992)


Hirsch J & Mjolsness E, 1992 “Center of mass computation”  Vision Research 1992  Vol. 32:2, pp. 335-346


Kveraga K, Boshyan J, Bar M 2007 Magnocellular Projections as the Trigger of Top-Down Facilitation in Recognition   The Journal of Neuroscience, 27(48):13232–13240


Locher, PJ  Wagemans J 1993 Effects of element type and spatial grouping on symmetry detection Perception 22:565-587


Locher PJ, Stappers PJ, Overbeeke K (1998) The role of balance as an organizing design principle underlying adults' compositional strategies for creating visual displays  Acta Psychologica 99(2):141–161


Locher, PJ , Cornelis E, Wagemans J Stappers P. 2001 Artists’ use of compositional

balance for creating visual displays. Empirical Studies of the Arts, 19: 213-227.


Poore Henry Rankin  1903  Pictorial Composition and the Critical Judgment of Pictures New York


Purves D, Augustine GJ, Fitzpatrick D, et al., editors. Neuroscience Third Edition, Sunderland, MA, Sinauer Associates 2004


Puttfarken, Thomas  Roger de Piles' Theory of Art, New Haven, Yale University Press 1985


4300 words