and I wasn't even there

Comments (View)
Comments (View)
Comments (View)

"Distance viewing": A visual analysis of Ernst Gombrich’s - The Story of Art


Ernst Gombrich’s The Story of Art has proven to be the most successful history of art book ever written. Today buyers of the book, now in it’s 16th edition, are greeted with a red sticker informing them that 7 million copies already exist. The book is a survey of art from prehistoric cave paintings up to the early 1980’s, the last painting in Gombrich’s story is Hockney’s My Mother, Bradford, Yorkshire, 1982. Given its huge audience it might be argued that the artworks in it are amongst the world’s best known.

 In terms of this article I assume a broad view of the book, something like this – The artworks selected by Gombrich form a canonical list that represents his particular view of the history of art. His version is a history that focuses on art made in Europe and it charts the development of the classical tradition in image making, sculpture and architecture. More generally it is a history of illusionistic representation in art. Also it is the selection of one man and must therefore represent to some extent his subjective taste in art.

As the title suggests the book is a narrative – it proceeds in roughly sequential steps from old to new, each previous example relating to the next. In this way a sense of artistic development is revealed. In the case of The Story of Art there are 411 steps, or 411 artworks that Gombrich selected to tell his story. Helpfully, Gombrich chose to write only about artworks he could illustrate  

This is quite obviously a well-known and effective method of presentation; the reader can chart specific themes or description in the text alongside the images and easily compare images next or close to each other in the sequence. Whilst it has many advantages for close study of individual or small groups of images, the book format also comes with set of disadvantages, particularly in the case of a project like The Story of Art. It’s an odd feature of survey books of this type that they set out to present the entire history of a subject and yet can only be accessed on a micro, page-by-page level. We can’t actually see the story of art only get an impression of it through connecting individual events. This has all sorts of implications for how we understand time in a narrative like Gombrich’s and counterintuitively when you read The Story of Art it can be difficult to grasp the bigger picture.

It seems to me that this is precisely the problem that Lev Manovich’s ImagePlot software seeks to address. Through computational analysis of images and visualization techniques, millions of images can be looked at together thus resolving the problem of only seeing a tiny slice of the whole when reading a book like The Story of Art. Manovich has coined the term ‘style space’ to refer to the process of ImagePlot. He defines it simply as,

 A style space is a projection of quantified visual properties of images into a 2D plane.

So the goal is visualize every image from Gombrich’s The Story of Art, to understand its ‘style space’, in order to explore certain patterns in the art historical canon and to test out prior assumptions we have about the book. ImagePlot uses image properties like hue, saturation and brightness to analyze images. Alone these properties produce interesting patterns but when combined with other information about the images such as, periods, iconography and medium we can say that ImagePlot provides an alternate and powerful method for addressing images in the history of art.


The Data.

411 digital facsimiles made from the images in Ernst Gombrich’s The Story of Art

The images are high-resolution scans not scanned by me. 

The data is not 100% accurate and there will undoubtedly be a few mistakes. This is mainly because organizing this data so it would work with ImagePlot was a very labor intensive process and took many hours and I didn’t want to get bogged down in perfecting it at the expense of doing the actual visualization work. 

The Story of Art visualized

Gombrich - The Story of Art - All Images; x-sequential order in book;  y-brightness

Full size image of above plot 

This image shows all 411 images compared with their average brightness. The plot does not take into the account actual date of the images only that the images on the left are the earliest and those on the left are the latest. It is not possible to conclude much from this plot and I have included it only to highlight the weaknesses and problems with the data itself. The images in the top of the graph are the brightest but this is because they are three-dimensional objects, like sculpture, and the brightness comes form the white background not the actual object. It is therefore not possible to include images with white backgrounds when working with image color properties. Also, to a large extent the reason these ‘bright’ images appear on the left of the plot is that no paintings survive from ancient Greece. This is a difficulty that seems to crop up with ImagePlot and I have written about it previously, namely you have to be sure you are comparing like with like. To avoid this problem I have only used two-dimensional artworks from the 411 when using the computational analysis tools.

However that doesn’t mean other plots can’t be made that make use of all 411 images in Gombrich’s canon. ImagePlot is versatile enough so that you can add your own data manually, independent of any automatic analysis.

Gombrich - The Story of Art - All Images; x-date;  y-location

Full size image of above plot

For this plot I assigned a number to each picture depending on where it was made.

Just compiling this information shows that Gombrich included art from 31 different countries. A simple bar graph can be made that shows the quantity artworks produced in each country.

Gombrich - The Story of Art - All Images; x-country;  y-number of images

Full size image of above plot

As expected the Story of Art is intensely Eurocentric with Italy (105) and France (80) responsible for the lions share of the art. Whilst Gombrich included a reasonably broad range of Asian countries, (although given the size of Asia we might conclude that the range isn’t actually that large) he only choose 1 or 2 examples from each country. Egypt is the only country to figure from Africa (although he did include a West African sculpture in his discussion on Picasso which is not represented in this graph). The USA does make some impact but the rest of the Americas are largely non-existent.

Going back to the plot we can see other geographic features of the canon. Obviously the most striking feature is that the majority of the plot contains no artworks at all. Even though humans were expressing themselves visually up to 15,000 years ago, for Gombrich, the history of art doesn’t begin in earnest until 12,000 years later. On one level this is hardly surprising given the lack of objects to survive over such a long period. But that is not to say that nothing has survived and the plot shows quite neatly just how much of human history is not included in the standard version of the ‘history of art’.

From Ancient Greece onwards there is a sharp rise in the number of artworks and where they were made. The plot is deceptive because the sharp rise only encompasses 31 countries but never the less if we use the assumption that The Story of Art is a history a certain kind of artistic tradition then we can see clearly those ideas spreading rapidly from 1st century onwards.

In order to get a clearer picture of this development I have removed the prehistoric cave paintings from the data and then re-plotted. In the 2nd plot I have annotated this graph to highlight some of areas of interest.

Gombrich - The Story of Art - All Images except cave paintings; x-date;  y-location

Full size image of above plot

The plot shows all artworks in The Story of Art from 2750 BC up to 1982. The date of each work is on the X-axis and the country it was produced in on Y-axis. The higher up a country appears on the Y-axis the later it occurs in the book. Unsurprisingly Greece is low down and the first major concentration of art appears in -700 BC to -100BC. I have also marked on other major periods and where they appear on the plot.

The Story of Art certainly conforms to our generally accepted view of artistic accomplishment; indeed the book is probably one of the main reasons for establishing this tradition. There are high concentrations of artworks around Ancient Greece and Rome, (although perhaps fewer around Ancient Rome than might be expected), the Italian renaissance, 17th century Dutch art and the 2nd half of the 19th century in France.    

What the plot does show however, which might not be immediately obvious from reading the book, is way artistic movements and periods become shorter over time. Reading the plot from left to right each major period gets progressively shorter with the impressionist period being barley visible. Over time, change in art appears to happen faster and faster.

As mentioned, the The Story of Art is incredibly Eurocentric, but whilst commentators on the book are keen to point this out, they have much less to say about the nature of this Eurocentricity. Firstly the influence of Ancient of Greece did not occur in Greece and Gombrich has not included any art made in Greece after approximately 100 BC. The main point however is that Gombrich seems to have conceived The Story of Art as a development towards geographic diversity with the last 200 years of art being the most diverse. So whilst generally the book focuses on the famous artistic periods, Italian Renaissance etc, there is also a subtext that sees tradition progressively reach more and more counties as time goes on.     

As noted earlier, I considered it only possible to run computational analysis on images from The Story of Art that were 2 dimensional. This plot contains all paintings, except for the prehistoric cave paintings, and plots the date on the X-axis and average brightness on the Y-axis.

Gombrich - The Story of Art - All paintings except cave paintings; x-date;  y-brightness

Full size image of above plot

The plot shows painted and 2 dimensional artworks to be sparse from the time of Ancient Greece up to the start of the 15th century. Gombrich of course was primarily a Renaissance scholar so it’s unsurprising that the amount of paintings in The Story of Art increase dramatically at the beginning of the Quattrocento. In this view of the history of painting the artists of the 11-13th centuries don’t really feature. Instead painting is seen as ‘beginning’ with Giotto’s advances in illusionistic figure modeling and the subsequent development of single point perspective and the visualization supports the argument that The Story of Art is also the story of illusionism.

The paintings can be analyzed in more detail by removing much of period prior to year 1400. This visualization plots paintings made from the year 1000 onwards with the X-axis being the date and Y-axis average brightness.

Gombrich - The Story of Art - All paintings 1000 AD onwards; x-date;  y-brightness

Full size image of above plot

The first thing to note is the gap between 1700-1765. Whilst it is easy to predict a proliferation of paintings in the 14th century it is a genuine revelation, to me anyway, that the early 18th century was a fallow period for painting. It’s possible to say therefore that for Gombrich there were 4 general periods of 2 dimensional artworks. 1st from around 1900 BC to 1300 in which there was very little activity, 2nd from 1300 to 1700 where there was intense activity, 3rd from 1700-1765 where again the canon is all but empty of paintings and 4th from 1765-1982 which saw painting return to its increasing numbers. Further to this the last period is not as concentrated with paintings as the period from 1300-1700. As well as being Eurocentric The Story of Art is also Renaissance-centric.

Moving on to the brightness variation I have annotated the above plot, highlighting certain artistic periods, to better understand the results.

Gombrich - The Story of Art - All paintings after 1000 AD; x-date;  y-brightness

Full size image of above plot

The first thing to note is the difference in brightness between paintings of the early Renaissance and the late Renaissance. As the 15th century progressed artists began to paint much darker pictures. One possible explanation for this is the increasing use of oil paint at the expense of fresco during the Renaissance but also it seems likely the increasing sophistication and popularity of techniques like chiaroscuro would have played a part.    


In addition to these very dark pictures there is small cluster of very bright pictures towards the top of the plot. These are drawings and it’s of interest, although perhaps not much of surprise given the regard Renaissance sketches and cartoons were and still are held in, that drawing only features significantly in The Story of Art is during the Renaissance.

From the 17th century onwards, (until the dawn of the 20th century) artists seem to occupy a similar brightness range from that of the late Renaissance perhaps due to the notion that these painters were thought of as having reached some unsurpassable standard, (perhaps also Gombrich’s taste tended toward darker images?) Taking a step back it’s quite remarkable how dark the tradition of European painting actually is – an impression you don’t tend to get walking around the National Gallery in London for example.

A curious feature of the plot is a small group of very dark pictures that cluster around 1640-50. These include two pictures by Rembrandt including a self portrait, two by Velazquez including Las Meninas and a Rubens self-portrait. It’s interesting that these three artists, amongst the most famous in the whole of art history, were all painting very dark pictures at this time. It’s even more striking when you consider that this small cluster also contains self-portraits by these three men – Las Meninas amongst other things is a self-portrait.    

Towards the end of the plot – the 20th century - the pictures become much brighter. Artists during this period seem to have rejected previous stylistic paradigms. This of course is also the period in which abstraction developed and certainly a feature of abstract painting is brightness. This is a pattern that Manovich and Douglas discovered previously in their 2009 article Visualizing Temporal Patterns – although they used skew data, rather than brightness for the plot. (I think skew is slightly different to brightness although I might be wrong!) It’s is also a pattern that can be seen in their work on Mondrian as he moved from figurative to abstract painting.

This plot shows the same set of 2 dimensional pictures with the date on the X-axis and average saturation on the Y-axis. 

Gombrich - The Story of Art - All paintings after 1000 AD; x-time;  y-saturation

Full size image of above plot

This plot shows some variation in pattern but it is not as pronounced as the brightness plot. In the 19th and 20th century the saturation of images decreases compared to 15th 16th and 17th centuries. As with brightness the least saturated images occur during the 20th century.

This plot also shows that some images in The Story of Art are printed in black and white and I’ll end with confirmation that the data is not perfect and will require tweaking to be strictly accurate. In this case I would want to remove any black and white images from the data set.  


 Future work

So far ImagePlot has produced some good results when applied to Gombrich’s The Story of Art. It confirms features of the canon that are expected but quite difficult to actually test by reading the book. It’s also a novel way to view the book and can be useful in understanding it’s development – particularly the occurrence of artworks over time.The idea of distance reading in computational text analysis is well known and ImagePlot makes possible what might be described as ‘distance seeing’ for images.

But this set of images is sorely lacking another set to compare it too. It would add an extra vital dimension to this work to compare The Story of Art with other art history survey books in order to analyze how the canon has changed over time and to what extent they either develop or diverge from Gombrich’s version.

Non-temporal visualizations, such as this one, which plots brightness on the X-axis and saturation on the Y-axis, are interesting on their own but I get the impression that their true potential is only realized when compared with other similar data sets.

Gombrich - The Story of Art - All images; x-date;  y-saturation

Another line of enquiry is the timelines that are provided as an appendix in The Story of Art. These are general timelines showing artists lives, historical periods and important works. I feel it would be interesting to compare these timelines to the actual occurrences of artworks in the book to see how they relate and to what extent the timelines are unrepresentative of the book as a whole. 

Also I have not specifically referred to Gombrich’s text, mainly because I haven’t read it since I was at school, but there is certainly scope for testing the text against the patterns of the visualization. In addition the data could be annotated with iconographical information, which could push the potential of ImagePlot into all sorts of new directions.  

Comments (View)

The All Seeing iPad

Comments (View)

Reclaim the Like button rehearsals

Comments (View)

My iBooks Bookshelf

I love having it all in the palm of my hand.

And remember, if you own a Kindle you don’t have any friends because who would lend a book to a friend for a maximum of 14 days after which she was forced to return it to you even if you said she could keep it for longer, and where she had to download some propriety system to read it but only if the publisher had decided to allow it to be lent in the first place? 

Comments (View)

Peering into misery

What is the museum?

A construct served to you on a handheld digital device by a global corporation that makes products irrelevant to any content it contains in which you are encouraged to blindly and meaninglessly support through a fake button? 

From Mashable

Comments (View)