Drawing & Cognition: Eye-hand co-ordination in portrait drawing

Eye-hand Coordination in Portrait Drawing:
a case study

1. Abstract 2. Introduction 3. Method
4. Observations 5. Conclusions 6. Footnotes 7. References

John Tchalenko, Camberwell College of Arts, Peckham Road, London SE5 8UF, UK

1. Abstract

We investigated with the help of an eyetracker, movement sensor and close-up video filming a well-known painter, Humphrey Ocean, drawing portraits (Miall & Tchalenko, 2001). In the present study we report on a further study of this data concentrating on the painter’s eye-hand coordination. We observed that, in general, his eye closely followed the drawing hand, with fixations on, or very near, the line being drawn. But there were also frequent exceptions to this behaviour when the artist’s eye departed from the drawing hand to fixate other parts of the drawing or turned to the model. Examples are presented for each of these cases, illustrating the process of visual memory fading and refreshing, and the possible action of a motor memory component in the drawing method of this painter.

2. Introduction

Humphrey Ocean is an artist of international repute with work in major UK and world collections. His portraits are drawn or painted from life and are realistic, and his pen or pencil drawings are made with lines only and no toning. An overriding concern with him is precision, of which he says: "I think detail is precision. Detail means where the line lands, and if it lands a millimetre to the right, or a millimetre to the left, it changes the weight, in some way, of the shape that it is describing. So when that line lands, you just want it to land in the right position…" On constantly glancing back at the model he remarks: "There is a point where the image starts breaking up. If I draw your eye, I know there is a point where I start making up or assuming something – and then it’s beginning to kind of shatter. So you have to look back and see what the next bit is." These considerations, as well as a long working experience between us, made Humphrey the ideal candidate for our observations on eye-hand coordination.

When drawing a portrait, Humphrey Ocean, as nearly all painters who work from life, transcribes, detail by detail, a visual perception of the model onto the paper. Consequently, during the making of the picture, the painter’s gaze goes from model to paper and back again many hundreds or thousands of times. In the periods when the painter is looking at the paper and drawing, he, or she, proceeds from a visual memory of the detail just perceived on the model. After a short time, usually one or a few seconds, the painter looks back at the model to refresh this memory of the detail being drawn, or look at the next detail to be drawn, and then returns to the paper, and so forth as the cycle is repeated.

For both portraits examined here, Portrait of Nick and Luke 2, the artist entered a regular gaze rhythm from the outset and maintained it throughout the session with only minor local variations. While drawing, he referred back to the model about 12 times/minute, usually with single fixations lasting about 1s (Figure 1). These rates and durations were also recorded for other portraits and seem to characterise Humphrey Ocean’s way of working, with the exception of quick (1 or 2 minute) drawings (Miall & Tchalenko, 2001).

Figure 1: Portrait of Nick: The first two minutes as recorded by the eyetracker
Click image to enlarge.
Back

3. Method

The eyetracker (the AlphaBio Eyeputer) is a specialised video camera system mounted on a headset and coupled to a computer. The infrared sensitive camera records a close-up image of the eye while an image processing board within the computer calculates the position of the centre of the pupil. The headset also includes a second camera (the ‘scene camera’) filming the scene in front of the head. Coordinates of the each point in space that the artist is viewing can be calculated in real time, and displayed either as a cursor superimposed on the scene camera image, or digitally stored on the computer for detailed analysis.

For this project we combined these recordings with those from a hand movement sensor (Polhemus Fastrak motion analysis system), so that we could also follow the movement of the artist’s pencil. This device records the 3D position of a lightweight marker, attached to the artist’s hand or pencil, with the data also digitally stored on the computer for later analysis. While these eye and hand movements were being recorded, the drawing’s progress was filmed continuously with a close-up video camera.

The temporal pattern of eye movements can be derived from the digital data, as in Figure 1, or from a frame-to-frame analysis of the scene camera’s recordings, as in Figures 2, 3 and 5. The latter method, although restricted to a resolution of 1/25th s, has the advantage of making evident the locations in space of fixations and hand positions, data which can then be used in conjunction with the close-up video recordings.

4. Observations

Line E (Portrait of Nick)
This line was drawn about 8 minutes into the 5-hour portrait. It depicts the back edge of the model’s upper eyelid of the right eye. The only other marks preceding this line on the paper were two eyelashes of the same eye (Figure 2). The line is about 5 cm long, concave downwards with a changing curvature. It was drawn left-to-right with the pencil stopping twice on the paper’s surface while the eye went back to look at the model – hence the three sections E1-E2-E3 (1).

Figure 2: Line E, Portrait of Nick
Click image to enlarge.
Back

Figure 2 provides the temporal sequence of eye and hand movements, and summarises the corresponding values measured. Using averaged values, one finds that the artist drew a line 1.7 cm in length in 2.16 s before refreshing his visual memory for 1.19 s. Fixations on the paper during drawing were stable and located near the start and slightly above the section being drawn. Prior to E1, Humphrey practised the section by moving his pencil without drawing several times over the future trace, and towards the end of E3, his eye looked away from the line to points about 5 and 10 cm to the right and below his pencil. He finally returned to E1-E2-E3 as his hand came to rest near his lap.

Line N (Luke 2 Portrait)
Luke 2 was a quick pencil drawing lasting 11.5 minutes, and the line examined here which depicts the nose’s outline, was started 3.5 minutes from the start (Figure 3). Although slightly more complex than the previous (the pencil left the paper momentarily to start N2 and practised before N4) we still observe the same draw – pause and refresh rhythm as before. The averaged refresh rate of 2.10 s after a line of 1.6 cm is nearly identical to the previous case. Fixations on the drawing are similarly stable or, for the longer line N3, show one adjustment saccade from the start to the end of that section. Nowhere do fixations follow the pencil tip in smooth pursuit.

Figure 3: Line N, Luke2 Portrait
Click image to enlarge.
Back

The lines studied above are of a particular category: single lines that define the shape and volume of the portrait’s main features: eyes, nose, ears, etc. Their trace is unpredictable, in the sense that knowing the starting point is insufficient information for drawing them because angles and curvatures change along their lengths. Furthermore, as they appear, on the whole, in the early stages of a portrait, there are no previous lines or marks to act as reference. To achieve a precise rendering of such lines the artist must refer back to the model every time his image memory fades.

Lines H1-H6 and H9-H12 (Portrait of Nick)
T hese two lines, the last drawn in this portrait, represent the outer contour of the hair on the model’s right side. They are also the longest single lines of the picture: 20 and 16 cm respectively (Figure 4). The first, of gentle uniform curvature, was drawn in two consecutive strokes, H1-H4 and H5-H6, with the hand resting at the lap in between. H5-H6 was then reinforced to become H7-H8. The second, H9-H12, is straight at the top and gradually curving at its lower end. For both these lines, the relation between eye and hand is more complex than for E and N (Figures 5 and 6).

Figure 4: H lines, Portrait of Nick
Click image to enlarge.
Back

Figure 5: H lines, Portrait of Nick, graph for figure 4
Click image to enlarge.
Back

Figure 6: Eye and hand positions during drawing of H lines, Portrait of Nick
Click image to enlarge.
Back

H1-H4
Prior to drawing, several long fixations on the model (average duration 2.35s) alternated with shorter fixations on the paper near the future trace of the line, during which time the hand practised without marking the paper(2). Drawing started at point 1 and the pencil was then tracked to 2 and 3 with two small saccades. The hand then continued to 4 but the eye left the trace for three further fixations elsewhere on the picture.

H5-H6
This is a continuation of the previous line starting with a small overlap. Drawing was preceded by fixations on the picture and model, then the eye went to point 5 as the pencil started but, instead of following the line, further fixations proceeded elsewhere on picture and included one fixation on the model. The latter corresponded to a short pause of the hand about 3 cm before point 6, similar to the refresh fixations seen in the E and N lines, although for H5-H6 the eye did not return to the line when drawing resumed.

H7-H8
Preceded by several fixations on the model lasting a total of 3.32 s, this line reinforced very precisely the previous line. At the start of drawing, the eye coincided with the pencil at point 7, but left almost immediately for fixations elsewhere on the picture while the hand drew the line.

H9-H12
A single short fixation on the model preceded the drawing which started at point 9. The pencil was tracked with a saccade to 10, then started leaving the line at 11 for three fixations elsewhere on the picture while the hand finished the line at 12.

With the last three lines, all sections below ear level were made while the eye was looking elsewhere, indicating that the painter did not need to fixate the lines as they were being drawn or after they had been finished(3). The H lines are probably more predictable than E and N in the sense that their curvatures, balancing the hairs’ tensile strength and the action of gravity, are more "flowing" and uniform. Furthermore, the H lines can be situated with respect to several previously drawn lines representing that part of the face. To achieve their precise rendering, the artist can make use of these elements in addition to his visual memory of the detail as seen on the model.

5. Conclusions

For the painter, the drawing process is task controlled – i.e. governed by the need to advance the portrait. The two types of lines we selected for analysis inform us on different aspects of this process.

The E and N lines illustrate the simplest cases of a fading visual memory needing to be refreshed. The painter’s eye fixates on the line as it is being drawn, and on the model while he is updating his memory. He can draw about 1.6cm in about 2 seconds before needing to refresh. Compared to results obtained in other areas of visual perception and imagery, this mental image maintenance time is exceptionally long. Kosslyn (1994) suggests that "the activity in the visual buffer should not persist longer than about a quarter of a second". However, in the painter’s case, the information that needs to be held in memory to continue an existing line is essentially restricted to a single element - the line’s angle. Furthermore, seeing the line as it is being drawn materialises the mental image and may aid in its maintenance (4).

The H lines denote a more complex eye-hand coordination procedure. The painter can draw up to 16cm before referring back to the model and he only needs to fixate the starting point of the line being drawn. During most of the drawing, his fixations centre on other parts of the drawing not directly related to that line (5). This would allow for perception of the "picture so far" to become an additional input into the drawing process, for example by situating that line with respect to others previously drawn. In a simpler eye and hand task also consisting of seeing and reproducing a model (in this case, a pattern of coloured blocks), Ballard et al (1992) observed a similar dissociation between eye fixations and hand position, the two only coming together at key points of the task. Their tests focused more on decision making and did not involve the sort of fine motor tuning required in drawing a portrait. Another observation on the H line shows how complex this tuning can be. It concerns the retracing of an existing line without the aid of vision. Whereas drawing without looking could be explained by motor commands based on a visual mental image, retracing a 10 cm line perfectly about 10 seconds after the original, and without ever having seen the original, suggests the presence of a motor memory component to the drawing process. More work with experts in drawing is required to elucidate this aspect.

These two quite different types of lines occur in their simplest forms during the first and last stages of a portrait respectively. In between, all the other lines exhibit a varying mixture of their characteristics, i.e. the rhythm and timings of draw – pause – refresh cycles, and the influence, as work progresses, of the emerging picture itself. Portrait drawing, at least for this painter, is a complex combination of a fading memory image of the model and an increasing presence of the emerging picture.

6. Footnotes

1. This line is also referred to in Miall & Tchalenko, 2001 as Line 5,6,7. Back to text

2. The hand’s movements clearly indicated that Humphrey was situating the future line with respect to other lines already drawn: right side of the head, ear, neck and a few strands of hair. Back to text

3. Another particularity about these H lines is the speed at which they were drawn. The pencil was moving at an average rate of 2.8 cm/s, exactly three times faster than in the previous E and N lines. Back to text

4. It is also possible that some ‘chunking’ may be taking place, i.e. the painter, subconsciously, divides the line he is seeing on the model into smaller sections, and then attends to these one at a time. However, even in this case, as the subdivision is determined by the painter’s experience of what he can hold in memory before needing to refresh, the dominant mechanism is still one of draw – pause and refresh. Back to text

5. In some cases, the drawing hand is still within the peripheral field, and it could be argued that the line is still seen. The painter, however, holds his pencil in such a way that most of the space before and after the actual pencil tip is hidden from his view, and thus very difficult to see accurately when focusing elsewhere. In most cases, fixations are too far away from the pencil to be of any use. Back to text

7. References

Ballard, D.H., Hayhoe, M.M. Li, F. and Whitehead, S.D., 1992, Hand-eye coordination during sequential tasks. Phil. Trans. Roy. Soc. Lond. B, 337, 331-339.

Kosslyn, S.M., 1994, Image and Brain, MIT Press, pp 516.

Miall, R.C. and Tchalenko, J.S., 2001, The Painter’s Eye Movements, Leonardo (Vol. 34, Issue 1, pp. 35-40).