What colours are our towns?

The urban questions that architectural projects are constrained by are not only formal in nature. The colour of the chosen materials must also guarantee harmony between a new construction and its environment. The choice of colours is constrained by regulations that are sometimes very strict, and yet these regulations depend on tools that are largely open to subjective assessment. It sometimes seems easy to decide which colours should be dominant in a given location, but how can colours be classified in accordance with quantified criteria? Luan Nguyen, a young architectural engineer at the University of Liege has laid the foundations for a standardized and objective method(1) for characterizing the dominant color of a house, street, neighbourhood or city. Apart from compliance with regulations, a tool such as this one opens up a lot of possibilities for better understanding the extent of colour trends in an urban environment.

What is the dominant color of towns? This is the question addressed by Luan Nguyen, who is due to defend his thesis in a few months’ time under the supervision of Jacques Teller, at LEMA (Local Environment Management and Analysis), in the ArGEnCo department of the Faculty of Applied Sciences at ULg. In his search for answers, the young architectural engineer published the results of an efficient, quantified and standardized method which promises a palette of varied applications.

Having studied fine art for a period, in the hope of becoming an illustrator, it was no surprise that Luan Nguyen should take an interest in questions related to colour during the course of his studies as an architectural engineer. His end-of-study thesis had already focussed on this subject by dealing with architectural constructions but had not looked at the problem from a wider urban perspective. He adopted a more global approach to the question based on his professional life outside the University. As the head of a firm of architects which he himself founded, Luan Nguyen became aware that, contrary to the old adage, personal tastes and colours were indeed open to discussion.  

Regulations on subjective criteria

The construction of any dwelling is governed by a series of urban constraints the aim of which is to guarantee architectural harmony between a new construction and the environment. This set of regulations, which can be constraining to a greater or lesser degree, depending on the town or neighbourhood in question, very often imposes a specific colour palette, or a single colour. “The drawback of this is the indecision related to the characterization of this colour. When a new brick needs to be chosen before the construction of a new building, a specific shade of colour is imposed, such as red, brown or light grey. The terms used are quite vague, flowery and subject to personal interpretation. When we appear before the municipal officer, he may well consider that the brick we are showing him is a brownish colour while it may appear red to us”, says Luan Nguyen. This is further complicated by the fact that the colour of a brick will not be the same inside (due to the artificial lighting of the community house), as outside, and will vary even more according to the changing light at different times of the day. Finally, judging the exactness of the colour of an entire wall based on a single brick can only ever be approximate.

In Wallonia, regulatory statutes in relation to colour remain very vague and rudimentary. By taking an objective approach to the problem, the researcher observed that these imprecisions were problematic on a Europe-wide scale. The tool most often used is still a chromatic chart which is arranged according to colour samples from a street, city or country and which is subject to a visual qualitative examination. The use of a colorimeter that yields very precise colour coordinates, can only do so for a very small surface of a facade and does not precisely identify the “average” or dominant colour of a wall.

“In reaction to this observation, I began by asking myself a first question which would help me to apply my research on a city-wide scale. How might one suggest a method for the objective characterization of colour”? It is certainly not about offering a tool to assist municipal regulations. The challenge is a much more fundamental one. In terms of architecture, how can we go from an approach based on qualitative appreciation to a quantitative appreciation of colour?

Continued research at LEMA

Luan Nguyen therefore began his thesis at LEMA, which had already suggested tools for characterizing buildings and urban planning. These tools would enable him to adapt his approach to an already existing quantitative framework. Up to that point it had always been a question not of colours but of forms. This meant that LEMA was able to call upon important expertise which, in particular, made it possible to objectively characterize the formal impact of the annexes to the courthouse in Liege on their environment.

Another determining legacy was the characterization by the Wallonia government of ten types of urban areas based on the shape of buildings. The research of Luan Nguyen focuses on four specific urban areas:  historic centre, periurban housing districts, working-class neighbourhoods and commercial zones. The categories grouped together eighteen representative fragments of the Liege urban area, a sample on which the young researcher decided to apply his methodology. “The reason behind the differentiation of these neighbourhoods was to make it possible to determine the dominant colours for each area and to create a representative sample for the city of Liege”.

From prospecting to the establishment of a single colour

In the first instance, some 2,000 photos of buildings were taken by students in standardized conditions. “The light and therefore the colour observed are not the same at midday or towards the end of the afternoon”, continues the researcher. “Another example, if a building is exposed to the direct light of the sun while one part of the wall remains shaded, its bricks will reveal two main colours that are different from each other”. The photographs were all taken in the spring, at midday, beneath a covered sky, with the systematic presence of the same neutral grey chart for a consistent white balance. The same camera was used for all the photographs and was calibrated to the same aperture and exposure settings etc. The colours of the 2,000 buildings could then be assessed having been taken under uniform conditions.

“We needed to decide what we wanted to study in these photos. Made up of stone, wood, roofing and panelling, a facade presents various materials of different colours. We were developing a global approach which would combine all these elements. But we first focussed on the brick which was the principal material and therefore the dominant colour. A second stage of the procedure consisted of cropping the rest of the photo from this principal material. We then used an algorithm to rebalance the colour of the brick based on our white balance. We finally had a dominant colour for a building”. This colour fragment was then characterized by means of a graphic tool in the HSL chromatic system (Hue, Saturation, Luminosity, which characterizes a colour according to its hue, saturation and luminosity).

“The HSL system offers three parameters which are very intuitive for characterizing a colour”, explains Luan Nguyen. “We could have kept the RGB system, but the coordinates, ranging from 0 to 255, do not make it possible to identify a colour. If somebody says, for example, that a colour is at 127 in the red, 62 in the green and 201 in blue, this means nothing to anyone”.  On the other hand, in the HSL system, the hue is expressed in a circle with degrees as a unit of measurement. Knowing that 0° corresponds to red and this then migrates towards yellow, green, blue, indigo etc., we can rapidly imagine the hue of a colour when we are given its corresponding degree. Saturation determines the purity of the colour and is expressed in percentages. Once again, a colour with a level of saturation close to 100% can be easily imagined as very pure Conversely, a colour with a saturation level of 10% will tend towards greyish pastel colours. While maintaining the hue and saturation level, if we see a point at 15° with a 90% saturation level, we can assume that the colour is saturated red. Luminosity, then, makes it possible to position the colour between black and white. Therefore, any colour can be represented by a point located in a cylinder, in a 3D space space based on polar coordinates. This can be done according to the degrees in the circle and their relative positions with regard to the centre in terms of saturation and based on how high or low they are in the cylinder in accordance with the light.

Individual variations in an average colour

The forty facades photographed in the rue des Anglais is a concrete example of what can be obtained by means of this method. A palette of colours which, when quantified, can be redistributed to as many points in the HSL colorimetric space. “In summary, we collected 2000 “individuals” in eighteen different neighbourhoods which were spread throughout this large cylinder and this allowed us to already observe certain trends. The distribution in a commercial zone, for example, was very different to that of the rue des Anglais yielding a palette that tended towards grey. In the case of workers’ apartment blocks, we observed a dominant reddish-brown”.

By means of this sample characterizing each of the 2,000 facades by a single colour, the study had already taken a first step, using this number-based approach, to determining the colour of towns. Each point had a value which made it possible to characterize these eighteen neighbourhoods divided into four urban zones and to already identify some trends. The second stage required the use of statistical tools to assemble these 2,000 individuals divided into samples (neighbourhoods) based on common characteristics. “In simple terms, it involved the arbitrary creation of four large colorimetric types. A concept which belongs to another very extensive heritage in architecture, the typology that attempts to classify buildings by category. Once again, typology is historically based on shape, whereas we applied these methodologies to colour”.

Four generic, synthetic colours should therefore describe the main trends observed in the different neighbourhoods of the city of Liege. “In order to achieve this, we therefore regrouped the 2,000 points present in space in four sets, using a partitioning algorithm, while taking account of their spatial proximity in the cylinder. We then attributed to each of these sets, a point in the centre of all the individuals of a group, called a centroid, which corresponds to HSL colorimetric coordinates, a generic colour, a type for a set of facades sharing similarities in terms of colour”.

A very promising tool  

For the town of Liege, Luan Nguyen found himself with four centroids of four different colour types, making it possible, at a glance, to give the colorimetric tendencies of a neighbourhood or a street. One was a grey-beige, one red-orange, one red brick and a light grey, each of these colours could be traced back to coordinates that had been quantified in a rigorously precise way. 

“We were then able to group each of the individuals of each neighbourhood according to their type and observe the dominant colours of these neighbourhoods and to make all kinds of observations, such as the objective characterization of the colorimetric homogeneity of a set of dwellings”. The neighbourhoods in the centre, for example, such as the rue des Anglais, point to a certain homogeneity, where, conversely, the four types are strongly represented on the boulevard Frankignoul, and show a strong heterogeneity in this neighbourhood. By complying with regulations in terms of materials authorised for the construction of social housing, the workers’ neighbourhoods show two types, red brick and red orange. “Evidently, these trends and these distributions of dominant colours for a type can be qualitatively deduced by the eye of the observer. The challenge here was to create a characterization tool making it possible to objectify and quantify them according to standardized criteria”.

“In addition to making it possible to identify the trends, the tool also makes it possible to capture a certain period”, explains the researcher. “That is the beauty of this method. Colour evolves over time and the same campaign carried out ten years previously would have created different types. It is therefore now very possible to imagine launching extensive campaigns over time on the same series of neighbourhoods and provide a database allowing for all sorts of analyses and comparisons. For example, to determine the impact of degradation due to deposits of atmospheric pollution on buildings, etc.” The possibilities for extending this research are many and also include the question of regulations. This is a problem that Luan Nguyen approaches with caution. “My fear with regard to the use of such a tool to impose colorimetric standards would be the creation of regulations that are too strict, which would result in a very rigid approach to colours in towns and cities and which would in turn become a constraint in terms of their evolution. It could happen that such and such a neighbourhood would be obliged to have a set of colours between such and such a degree and to prohibit any deviation from this. This would stifle creativity. Of course, the architect must take the environment into account in terms of integration. But integration does not imply repetition of what exists already. From an urban point of view, the most valuable and progressive thing would be to create a new event by relying on a contemporary language of architecture, to suggest a dialogue between the old and the new. This dialogue, when it is pertinent, becomes interesting and creates harmony. It is a question of shape and a question of colours”.   

(1) NGUYN LN., TELLER J.,Color in the urban environment: A user-oriented protocol for chromatic characterization and the development of a parametric typology, Color Research and Application, 2016