Amgueddfa Cymru — National Museum Wales

Home

And they're here: for the first time, we have figures for a year of e-book sales, supplied directly by publishers. It's still far from the whole picture, as not all e-book figures are available. But we now have a much better idea of what the book-buying landscape looks like in the UK.

The figure that stands out is that e-book sales are now up to 13%-14% of all book sales. However, as their prices are cheaper, that's only 6%-7% of revenue. Print book sales are down again, by 3.4% on 2011, as are average prices.

The e-book market is still dominated by fiction, and those e-book figures track the print figures. That is, if a book sells well in print, it also does well in e-book. The stand-out example is a particularly, shall we say, shady trilogy, whose e-book sales are about 36% of the print sales. Could the success of the e-book version of these titles lie, I wonder, in the fact that no-one can see what you're reading on your Kindle...?

So, it's mixed news: more books were bought in 2012, but because more of them were e-books, publishers made less money. Good news for reading, less so for publishing.

Meanwhile, here at Amgueddfa Cymru our journey into 'e' continues...

 

With thanks to The Bookseller for the sales figures.

Graham Davies, Online Curator, Amgeddfa Cymru - National Museum Wales

Earlier this year I was approached by the Keeper of Art regarding a recent acquisition of two oil paintings dating from 1700. The paintings in question - two large panoramic paintings of Margam House - were purchased with help from the Heritage Lottery Fund and the Art Fund, and as the funding included some form of digital exploration, I was asked to produce an interactive to compliment the gallery display (that would also be available to tour with the paintings).

On closer inspection of the paintings, lot of small, intricate detail became noticeable.  As some details were very small - and would benefit from additional interpretation - some sort of zoomable image would offer the visitors with the best form of digital exploration.

In previous gallery interactive developments, the New Media Department have used the iPad2 platform, but as the newer iPad3 has a retina display - a screen resolution and pixel density so high that a person is unable to discern the individual pixels at a normal viewing distance - this platform seemed ideal to explore small detail in high resolution images. As for the software, we decided to use an adaptation of a previously developed interactive that allowed the user to move around a high-resolution image to predetermined hotspots (to explain certain details of the painting).

I then went about locating numerous details on the canvas that would be interesting and informative to the viewer. I was keen to pick things out that would not be contained on the gallery labels, thus ensuring no duplication and offering an enhanced visitor experience.

Resolution issues

After obtaining high-resolution images of the two paintings I was able to upload these to our Content Management System (amgeuddfacms), to allow them to be served from our website server. It was at this point that a strange technical inconsistency occurred....

Viewing 3800?pixel wide image on a standard Firefox browser on my desktop screen rendered the image sharp and clear. However, serving the exact same image into the iPad3 web browser rendered a fuzzy pixelated image. Given that the same image was being served to both screens it seemed that the different web browsers were handling the image in a different way.

A little bit of research on the internet revealed the possible problem: The default web browser on an iPad (Safari), runs WebKit which only seems to serve images up to a maximum size (somewhere around 1024 pixels wide). The iPad browser seemed to be downsampling the original 3800 wide image to 1024 wide, before then upscaling this 1024 wide image back up to 3800, causing the image to render at almost 400% it's downsampled 1024 pixel size.

The problem was how to get around this. After some researching on the internet for similar problems, it seemed that there was no conclusive solution. This was mainly due to the logical assumption that it’s bad practice to serve huge images to a website (especially on a mobile device such as an iPad, where web content was readily downloaded via mobile networks), so the advice was always to use small images. Of course we wanted very large images, so this didn’t help!

One solution that Chris Owen, our Web Manager came up with was to serve two halves of the image separately and automatically stitch them back together again after they loaded on the web page  - thus the page would load two smaller images. Technically this gave a good result, but cutting the image in half was not enough. We therefore generated a script that sliced the image into 500 by 500 pixels (totaling 64 separate images), and stitched them all back together again once they were loaded into the browser.

The outcome was a high-resolution image (made up of smaller individual images) that renders sharply even when zoomed right in. This gets around the issue of the Safari web browser on an iPad automatically scaling down large images.

Research suggests that this may be a first in application development of this sort, especially one developed wholly in HTML5.

Gesture enhanced interactive

Once this high resolution was served up onto the new iPad3, in high resolution quality, it became clear that it would make more sense to make this a ‘gesture enhanced’ (pinch to zoom) interactive in addition to interpreting predetermined parts of the painting.

This means that the user can now fully explore the entire image, zooming right into any part of the image, whilst being able to read interpretive labels embedded within the image.

Colour Accuracy

The next problem to solve was the one of colour accuracy. Due to the original paintings being very dark, most images that we had to play with were lightened in order to see the detail. This lightening caused the colours to be untrue to the original, something that would be noticeable once screen and canvas were next to each other in the gallery.

A quick phone call to our photography department affirmed that they had high-resolution master TIFF files available that were ‘colour correct’, i.e. the colours in the digital capture were exactly as they were in the painting.

These colour correct images turned out to be strikingly different to the ones that we had previously been playing with, the increased sharpness causing even more detail to become apparent, even figures appeared that weren’t noticeable on the previous images.

The application will be installed alongside the paintings of Margam House in the Art in Wales 1500-1700 gallery at National Museum Cardiff in October 2012.

Bringing it all together: Art in Wales 1550-1700 

A parallel development to this in-gallery interactive is a website interactive exploring a major portrait of the builder of Margam House – Sir Thomas Mansel with his wife, Jayne (hung alongside the Margam paintings in Art in Wales 1550-1700 gallery). Again, certain parts of the painting can be explored in high resolution through interpretive labels embedded in the image. This further compliments an existing interactive, exploring another major portrait in the gallery – Katheryn of Berain, the Mother of Wales.

It is hoped to extend this program to include all the items in the gallery, thus forming a holistic digital interpretation of Art in Wales from 1550-1700, available both within the gallery and through the website.

VADU: Visual Audio Display Unit

My last blog entry was back in April 2009, so this is a hesitant return.

Background

In July 2011 the refurbished Contemporary & Modern Art galleries were re-opened at the National Museum Cardiff, the following VADU would be included in the initial exhibition. The specification was mainly to showcase video shorts: recorded, interviewed and edited by a few of my colleagues (Art and New Media), the videos would have subtitles and also there would be a visitor comments page.

The iPad 2 was about to be launched when I started work on the VADU. Magic was in the air, queues were forming.

Note: it has been used for two exhibitions, so far:

  1. Contemporary & Modern Art (July 2011 – January 2012)
  2. The Queen: Art and Image (January 2012 – April 2012)

Project Ethos

Although there is an 'App Store', it seemed like overkill to write and release an application simply for four machines in an art gallery - I used 'Kiosk Pro' instead, an application which basically removed all the usual iPad functionality and locked it down to a fullscreen Safari browser.

In terms of creating the backend mechanisms for providing/collecting content, one shouldn't reinvent the wheel. Therefore, I used the trusty Museum‘s web CMS (Amgueddfa CMS) to control things - the Amgueddfa CMS controls the museum websites and intranet for all seven nation museums. Built in-house over several years, it has evolved with the requirements of the organisation. It's built with open source PHP, MySQL and Javascript. Succinct and effective, even if we say so.

In regards to the actual design aspect of the VADU interface: the only two constraints were museum brand guidelines and a particular colour had to be used to unify it with the surrounding new gallery signage.

Around May 2011, Braun's designer Dieter Rams design ideas entered my world (I can't repeat his mantra here, but only because it is copyrighted). Anyhow, I tried to create an interface that was simple, intuitive, consistent, and didn't distract the user from the actual content:

  • Only two colours
  • No drop-shadows or gradients
  • 10px borders
  • I did allow myself one curvy corner (bottom-right), one guilty pleasure
  • User navigation: horizontal gestures only
  • Page transitions: vertical only

Building Blocks

If you haven't already looked at the photograph 1 and 1b, I would take a look now; so we have a shared reference images in mind.

The top section displays the title of the exhibition, followed by page buttons (Art, Comments and What's On). All exciting stuff - the tip of an arrow indicates which page the user is on (hopefully in a subtle fashion).

The language button starts the most convoluted process on the VADU, in terms of animating a page change. I didn't want the change to be instant, instead I gave it a more graceful flow. The actual result of changing the language only swaps the domain name from English to Welsh or visa-versa (museumwales.ac.uk, amgueddfacymru.ac.uk), but the time it takes to do that is over four seconds. It should become clearer later, if you watch the video below.

I used the Javascript Mootools framework, it has served us well - lightweight and fast. I chained the transitions and effects to create the smooth transitional actions throughout the VADU. It was also necessary to chain things because I would describe the Safari browser on the iPad as OK, in regard to the power of Javascript, rather than excellent if used on a desktop (there is reduced access to the normal power you find in apps wrapped up in objective C; no WebGL access).

The middle section displays the video in focus at the time, large screen print from the video is shown in the background - title, summary and extraneous information such as video length are shown on the information panel. In large font the word: 'Play', indicating the user can start a video. Left and right arrows also allow the user to shuffle through the videos. The information panel can be moved from the right or left of the background image - something that resulted from the fact artists don't like their work flipped [in a digital sense] i.e. if the focus of the background is to the left, the information panel can be positioned to the right by indicating such in the CMS metadata entry.

The bottom section can be dragged with a finger left and right, selecting any one of the sixteen videos. It was quite important to have draggable areas, because it is simply expected by iPad users (thus, making it intuitive). The same draggable feature is used for the what's on page (photograph 6).

If the user selects a video, the screen removes all navigation features so they are only left with the title of the video, a video time indicator (so the user knows the video is only short), the video itself in the center of the screen, the subtitles at the base of the screen and a 'back' button. The user can pause and un-pause the video by tapping of the video in the center (see photograph 3). The video's themselves use the same colour as the VADU, so it all fits together neatly.

Finally the comments page is simple too: optional name input, text input and the last three comments are displayed on the right-hand side (hopefully encouraging the user to write something). The comments are fed into the usual website comments system, approved (or not) by a staff member (photograph 5) – there has been over 2000 comments left of the gallery VADU since July 2011, which is quite a lot considering no one was forcing these people to write something.

Subtitles

The subtitles are again displayed using javascript, they are not integrated into the video. I built a subtitles tool within the Amgueddfa CMS using Javascript and Flash, which creates a XML file that could be dropped into the caption of the video. The idea being that the majority of videos need to be available to two languages and if the videos were published on the main website at anytime, a transcript could be displayed too. Hopefully the task of transcribing the video in the first place is easier because of it. There you go, a little of the behind the scenes work (photograph 7).

Past Performance

The gallery VADUs have been very reliable; once every few months one of them may freeze, but considering they are always on (one weakness of iOS software is you can‘t boot-up into a single application), that's not too bad. I darken the screen after the galleries close – simply using a whole–screen black div.

We had a brief problem when changing the local network settings in January, so I added a check before the VADU changes language to see if there is a network available (an AJAX query: onSuccess or onFailure).

Future Development

If it continues to be used, I would like to develop a local version of the VADU, providing a fallback if the network goes down, or maybe a hybrid version (storing the videos on the VADU). This would mean a update of the iOS (from 4.3 to 5.1+), but I‘m sure there would be some associated browser performance improvements.

Other major changes shouldn't be required, as the video shorts are meant to be the star of the show.

Final Word

Obviously it helps to have a pleasant environment to place the VADUs (photograph 9).

Video Demo

I've included a short demo video for posterity:

Graham Davies, Online Curator, Amgeddfa Cymru - National Museum Wales

When a member of the Art department approached me to ask if I could feature two views of the same painting online — one version covered in dirt and yellowed varnish (as the painting was when it came into the Museum), and the other version showing hidden detail and crisp colours (after being cleaned by Museum conservators) — I realised it would make a perfect interactive if you could use your mouse to virtually 'clean' the dirty canvas to reveal the clean version underneath.

Guardi's view of the Grand Canal, Venice

The painting in question is Francsesco Guardi's View of the Palazzo Loredan dell'Ambasciatore on the Grand Canal, Venice, painted around 1775-85.

Acquired by Amgueddfa Cymru in 2011, this painting is an important acquisition as Guardi's Venetian views are regarded as highly significant in the history of landscape painting.

You clean the painting

To make the most out of this dramatic before and after view, I needed to work out a way of 'virtually' cleaning the painting online by dragging a mouse over the dirty image to reveal the original details and colours previously hidden underneath the dirt and old varnish.

Reinvent the wheel?

I wanted something that allowed the mouse to act as an eraser; allowing one image to be rubbed out to reveal a secondary image underneath. A hunt around the internet brought up the required functionality already created by by Jonathan Nicol (www.f6design.com/journal).

The next step was to acquire high resolution copies of both dirty (before) and cleaned (after) digital images of the artwork from the Photography department.

Precisely aligning two slightly different angled photographs of the same picture

When I opened these digital images in Photohshop it became apparent that variations in the perspective, and distance of the photographic captures resulted in two images that did not precisely match up once overlaid on top of one another.

After an hour of miniscule adjustments using the image warp feature on Photoshop using the images as separate layers within Photoshop (one set at 50% opacity), I eventually achieved a precise overlaid match.

I abandoned trying to do this at 100% view as the image was so large and the time lag in processing too great to view the results (even for my G5 at 2.44Gz and 8GB RAM). I had to settle for a 25% view that filled my Apple 32" screen)

Once I had a satisfactory matched up and aligned the 'dirty' layer on top of the 'clean' layer, I could create the two corresponding TIFF images to incorporate into the Flash file as a basis for the interactive.

After a bit of tweaking, fiddling, and constant testing, I managed to create a simple interactive, allowing you to use your mouse to erase the dirty image, revealing the clean one underneath.

Exploring the detail.

I then decided to repeat this process to create several versions, all using crops of the high resolution images to show close up details of the painting.

Areas of particular interest I choose to separate out were people rowing a goldola, the architectural detail of the buildings, and the detail of the sky and clouds where much original detail had been almost totally obscured by years of grime, dirt and previous 'touch-ups' to the painting. The clean version revealed original intricate details and brushwork.

Future applications for Museum archives and collections

I am hoping this functionality can be utilised for other online images of the collections in the future. Ideas I have at the moment are to reveal hidden under-drawings only visible under x-ray light — as in the example of Richard Wilson's Dolbadarn Castle (NMW A 72), which has been painted over a portrait of a woman, and Landscape with Banditti around a Tent (NMW A 69) which he painted over a Venetian-style reclining nude.

Additional ideas include viewing a landscape or post industrial townscape that can be erased to reveal a historical image underneath...

Ok, so we had the iPad moment. What’s changed? Lots. The iPad itself was, in truth, disappointing for publishers. Beautiful, sure, but not very helpful. It wasn’t multifunctional and it wasn’t backward compatable with much stuff either (I can’t be the only person still using OS 10.4?) But, like Apple’s previous offers, it was a gamechanger. It established the tablet as a device, despite many people, myself included, wondering if anyone really wanted Job's 'third device'. Apple then let other manufacturers come up with their own versions, the best of which is probably Samsung’s Galaxy, and quietly went home to improve their own model. Having established the tablet, and just in time to catch the secondary wave of adopters, out comes iPad 2. With improved functionality and more features (camera – two, actually), it still passes itself off as the most desirable tablet, even if it’s not necessarily the best. With iPad 2 and the iPhone, Apple has now firmly entered the mainstream consumer market. In losing the geek factor, what has it gained? Well, turnover, and profit, obviously. While Apple’s top-quality combined hardware/software model of Macs retains its market-leading position in the creative industries, the iPods, Ipads and iPhones are now thoroughly high-street, even with their top-end price tags.

However, part of this trajectory has been the strategic downplaying of the iPad’s e-reader function, which is what publishers were most  excited about. Instead, the iPad focuses on portable, sleek, seamless acces to the web and email – truly, a big iPhone, but also ready and waiting for Web 3.0.

In terms of e-readers the iPad moment just didn’t happen. This has left Amazon’s Kindle as market leader, even though it only reads Amazon’s own e-book file format (although there are rumours Amazon will soon be allowing US publishers to submit e-books in the industry-standard e-Pub format). Amazon introduced the Kindle in 2007, by 2010 in the US Amazon were selling more Kindle books than hardbacks; today Amazon sells more Kindle books than hardback and paperback put together. At the moment it’s selling 105 Kindle books for every 100 print books, and three times more Kindle books than this time last year. In the UK, where the Kindle store has only been open a year or so,  Amazon are selling twice as many Kindle books as hardbacks.

What can we learn from this? Remember, the Amazon figures only apply to their own sales, of Kindle books, which can only currently be read on a Kindle device. What’s happening across the rest of the bookselling industry? The true picture for the UK is that sales of e-books are currently 2.5% of all book purchases; interestingly, they peaked at 3% over Christmas (did you get an e-book in your stocking?!) Adult fiction is still the most popular category, at 5.4% of all purchases; men and women are buying e-books equally, and the age group 55-64 makes up over a quarter of e-book buyers.

This 2.5% seems like a tiny figure for us all to be worrying so much about, especially as the value of the sales is low – about 1.6%. I still can't wait to have a go though.