ICA-PAAG. Short Guides 1 Title: The Digital Image Archive Author: David Iglésias i Franch. Centre de Recerca i Difusió de la Imatge (CRDI)
ICA Photographic and Audiovisual Archives Group (PAAG) Direction: Joan Boadas i Raset Coordination: David Iglésias i Franch Author: David Iglésias i Franch. Centre de Recerca i Difusió de la Imatge (CRDI) Publication: April 2014 2
The Digital Image Archive 0. Introduction At the end of the twentieth century, chemical photography was supplanted by digital technology, which meant the creation of a new landscape. For photography, this new setting brought with it major changes from a technological, social and cultural point of view. For archives, the care of the digital image represents not only a technological, but also a methodological challenge. For the archivist, it is fundamental to understand the object under his or her care, the digital image, and to know how to adapt it to the demands of its environment. This guide aims to establish the foundations for moving forward in this process. 1. Basic concepts Raster or bitmap image An image formed by a matrix of pixels, each with a particular value and a specific location within that matrix. Pixel is the contraction of the term picture element. It is the smallest element of a raster image with its own representation. It is a mathematical concept, but has its equivalence in the analogue reproduction by devices. Spatial resolution The number of samples that make up the image. When the reference is given in inches, we talk about ppi (pixels per inch), dpi (dots per inch) or lpi (lines per inch). In photography, the concept of resolution refers to the ability of an image to represent detail, a result of the processing of the spatial frequencies in the capture process. The calculation of detail is achieved by the interpretation of the MTF, a concept which applies equally to chemical and digital photography. The MTF is the calculation of the spacial frequency response of an imaging system and is explained by the contrast in relation to the low frequencies, based on the idea that high frequencies mean greater detail. Bit depth Number of bits used in representing a single pixel of the image (1 bit, 8 bit, 16 bit). The moment the electrical charge is transformed into numerical values, via the analoguedigital converter (ADC) in an operation known as quantization, is when the bit-depth, i.e. the number of bits that represent each pixel, is decided. The greater number of bits, the more likely the tonal values of the image will be represented, but also the larger the volume of the final file. Thus, we can say that bit depth determines the tones that can be represented between the two extremes of minimum and maximum brightness. For example, 1-bit only allows the representation of a black and white image; 4-bit allows the representation of up to 16 tones, a number which is insufficient to display a continuous tone and it is what leads to the appearance of bands on the image. It is from 8-bit and above that the representation of the image presents a correct gradation of tones. We should always think in terms of 16-bit for images intended for permanent preservation and especially for those on which further processing will be done, such as - 3 -
images that will be published. Colour The generic representation of colour is produced by encoding the resulting values of the demosaicing process. The RGB model is an additive colour system, which are those that are formed by the sum of the various primary colours: red, green and blue. It is used in monitors, projectors, cameras, and scanners. This mode can be transposed into CMY, a colour system formed by the subtraction of a particular light by an object. The primary colours subtracted are cyan, magenta and yellow, complementary to the primary additive colours. It is used in printing, where black (the key colour) is added, in what is known as a CMYK model. For digital colour, the key concept is colour management which consists of a set of tasks for the equalisation in the interpretation of colour in order to give specific meaning to the RGB and CMYK numbers so that they maintain the consistency of the original colour. For this a colour profile was created, a reference space of RGB or CMYK values with the values represented in the colour models independent of output devices, such as CIE XYZ and CIE Lab models. It can represent a specific device or an abstract colour space. The most popular abstract colour space is srgb, a space with a limited colour range designed for the web and therefore to be displayed on the screen, and the Adobe RGB (1998), designed to reproduce in CMYK but from RGB values, with a range that can represent 50% of the CIELab space. For devices, the reference standards are the ICC profiles. These are standard profiles created by the International Color Consortium intended for the description of ranges of scanners, digital cameras, monitors, and printers. Image quality The concepts explained above are fundamental to understanding the digital image but in order to analyse the digital image in qualitative terms, we need to introduce other concepts that can help to focus our attention on the quality of the pixels. We have mentioned resolution in reference to the ability to represent detail, but the most important attribute of the image, and the most decisive in the evaluation of the observer, is the tone. Its evaluation is complex but an objective assessment can be made based on the knowledge of the dynamic range, deviations of light and the OECF (Opto-Electronic Conversion Function), which relates the optical densities of the original with the numerical values of the digitised image. Along with tone, colour is the other most important qualitative factor, since the human eye is an excellent judge of colour There are other important concepts related to image quality. The dynamic range which is the ratio of the tonal values included between the brightest and darkest point of a scene that can be captured by a scanner or a camera. The white balance which consists in tuning the white point of the camera depending on the type of light. The colour of the light is determined by the colour temperature, which depends on the wavelengths that make up a particular light (light can have a predominance of red, green, yellow or blue, according to the predominant wave length). The noise consists of the appearance of random fluctuations which are undesirable in an image. It is inevitable when receiving the signal from an electronic device. This can be caused by the sensitivity of the system, the exposure time and temperature. To measure the noise of an electronic device, the signal received is compared with the noise obtained, known as SNR (Signal to Noise Ratio). Finally, artefacts are the set of alterations of the image which compromise the level of quality in whatever manner. Noise is an artefact, as is also the aliasing that occurs when the sampling rate is insufficient for the spatial frequency of the digitised image and is manifested in the presence of information that doesn t exist in the analogue original (typically pixels of colour) in the outlines of elements that form the image. - 4 -
2. Graphic formats Formats are the structures that give sense to the bits that make up a particular digital resource and, consequently, allow its reproduction in a particular technological environment. Two main types can be identified: raster (bitmap) formats, represented through a matrix of pixels, and vector formats, which represent objects through geometric figures. The former are those that relate to photography. There are different structures for raster formats, but generally they always have a header containing metadata, and the set of bitmap data, i.e., the information for each pixel forming the image. The level of metadata can vary greatly depending on the format, but in all cases there is a minimum of information that allows the version of the format to be identified and the image to be reproduced in particular environments. Formats are a key element for the digital file since they contain both the values of the image and the values for its reproduction. For this reason, understanding the logic of their structure, the level of information associated with them, the benefits offered by the software and the characteristics which favour their preservation, is essential. The commitment to safeguarding digital images obliges us to gain considerable knowledge about the object of our work and, accordingly, graphic formats occupy a decisive position. Below, we explain those formats which have a major role in the archival field: TIFF, JPEG and JPEG2000. TIFF Format created by the Aldus Corporation in 1986, in an attempt to overcome the lack of a standard for the image. In 1987 it passed into the hands of Adobe which created new versions: version 4.0 in 1987, 5.0 in 1988, and 6.0 in 1992. Over the years, the format has established itself as the de-facto standard and has added multiple features (extensions). The bid to create a standard RAW format led to the emergence of DNG, and the need to overcome the 4 GB barrier resulted in the BigTiff. But despite the usefulness of the new formats, the benchmark version remains that from 1992. Most notable characteristics: Representation of bitonal, gray-scale, palette-colour and true-colour images. The use of a compression system is optional. Accepts different systems, such as the LZW, RLE and JPEG. Compatible with high depth, 16-bit. It is designed to be extensible and to evolve according to new requirements that might arise. It allows unlimited inclusion of private or specific information (metadata). Versions 5 and 6 are compatible with most applications. The structure of this format has the distinctive feature of containing the directories where the image tags (metadata) are stored. Metadata. The mandatory elements of the TIFF baseline for images in grayscale or true-colour apply only to 11-12 of the total 32, not counting those corresponding to extensions. TIFF is a de-facto standard, the TIFF / IT is an ISO 12639:2004 standard and the TIFF / EP is an ISO 12234-2 standard. The technical specifications are published at: http://partners.adobe.com/public/developer/en/tiff/tiff6.pdf - 5 -
JPEG The technical specifications of the JPEG define it as "a family of algorithms for encoding-decoding for continuous tone images with a dataflow architecture to maintain and represent the compressed data. Therefore, this definition makes it clear to us that JPEG is not a proper format but a compression system for continuous-tone images. But under the common name of JPEG we find the formats developed by the Joint Photographic Experts Group (JPEG) to facilitate the flow of JPEG data; the exchange within a wide range of technological platforms; and applications with the aim of being highly compatible with all kinds of applications and environments. These formats include JPEG_EXIF, JFIF, SPIFF and JTIP. Most notable characteristics: - Representation of gray-scale and true-colour images. - In the standard compression it transforms RGB values into YCbCr. In the case of gray-scale it uses only the Y component. - There are four modes: sequential JPEG (Baseline), progressive JPEG, hierarchical JPEG, lossless JPEG (uncompressed). - Compression is based on the DCT and entails losses of information. Applications allow the applied compression ratio to be determined, which can range from 12:1 to 100:1. - It is not compatible with high depth, 16-bit, with the exception of JPEG with lossless compression. - The metadata which the JFIF format includes is minimal and nothing in comparison to the high performance of the JPEG-EXIF. The technical metadata is included in the APP0 frameworks. Additional frameworks are ignored by applications that don t recognise them and, therefore, those applications can also interpret the format and display the image. This circumstance makes it a highly compatible format. - ISO 10918-1 Standard. - The technical specifications are published at http://www.jpeg.org/public/jfif.pdf JPEG 2000 The format was developed by the Joint Photographic Experts Group (JPEG) in order to overcome the weaknesses of the JPEG format and bring new features to the image files. The first draft of the first specification was published in 1999 and, since then, the 12 parts which constitute the JPEG2000 family, characterised by its encoding, have been developed. The values that the format possesses have caused some heritage institutions to back its adoption to the detriment of other formats with a longer tradition in the heritage field, especially TIFF. Its implementation is slow and it doesn t have the predominance that it has in other fields, such as in medical images. Although it is compatible with most image editing software, it has compatibility problems with Web browsers. Most notable characteristics: - Representation of bitonal, gray-scale, palette-colour and true-colour images. - Wavelet compression (DWT - Discrete Wavelet Transform), with or without losses (optional). - Compatible with high depth, 16-bit. - It allows the storage of multiple resolutions in a single file. - It allows the creation of different resolutions in specific areas of the image, a function much appreciated for medical images. - 6 -
- Processed slowly due to its complexity. For this reason it has not been adopted en masse by the digital camera industry. - In low and medium resolution, the quality is superior to JPEG. - At high resolutions it avoids the effect of blogs, but on the other hand loses definition. - The standard comprises 12 parts. The parts of the format applicable to the photographic image are: JP2_FF, JPEG 2000 Part 1; JPX_FF, JPEG 2000 Part 2; JPM_FF, JPEG 2000 Part 6 - The structure of the JPEG 2000 family is always based on the JP2 (Baseline) which is extensible. The basic component of the structure is known as the box which contains different types of fields at the same time. The boxes may contain image data or metadata. - There is a minimum of required metadata, such as basic image data and the information relating to colour. However, it's an extensible format that can accommodate metadata in XML and different schemas such as the IPTC and the Dublin Core, amongst others. - ISO / IEC 15444-1 Standard. - In principle, JP2 is free of patents, although the situation is not entirely clear. - The technical specifications are published online, but they have to be paid for. However there is a version of the draft specifications from the year 2000 viewable at: http://www.jpeg.org/public/fcd15444-1.pdf 3. Metadata Metadata is: "structured information that describes, explains, locates, or at least facilitates recovery, use, or management of a source of information. Metadata is often defined as data about data or information about information"(niso, 2001). Therefore, metadata is not limited to the traditional descriptive fields of the archive but has a wider scope. When we speak about metadata, we also speak about the technical information which electronic files automatically store in order to operate in a particular computing environment and, also, in order to document aspects that may be of interest when editing of image such, as capture values. Below we explain some of the most significant standards for the digital image: EXIF, IPTC and XMP. Although not expanded on in this guide, for technical metadata it is essential to consult the dictionary of metadata for the still image, the ANSI / NISO Z39.87 EXIF The EXIF (Exchangeable Image File Format for Digital Still Cameras) format is a standard format created by the camera industry, specifically by JEITA (Japan Electronics and Information Technology Industries Association) and CIPA (Camera and Imaging Products Association) to facilitate the communication of graphics files in different technological environments and above all to improve the quality of printing of files originating from very different backgrounds. Although it is a graphic format, its main value lies in the set of capture metadata that it incorporates. It works on the basis of two formats: JPEG compressed images and TIFF images without compression. In both cases, the EXIF describes a set of TIFF tags, according to the format described in version 6.0, and for that information relating to the camera which was taken into account in the TIFF, it includes metadata in its own directory, differentiated from the TIFF metadata directory and also from the GPS metadata directory. - 7 -
Currently almost all digital cameras work with this format. The objective of Exif is to store and structure the technical metadata from the image capture taken from the configuration parameters used by the camera. This metadata will be useful for any subsequent processing of data, such as for viewing, editing on screen or printing. This metadata is essential for the optimisation of the image by whatever software is used. IPTC The International Press Telecommunications Council (IPTC) has had a prominent role in the push for the creation of metadata documents linked to image files. In 1990, they defined a schema, known as the IIM (Information Interchange Model), a wrapper or container format created for the transmission of news in text and images. From IIM, IPTC headings were created which were adopted and included by Adobe in Photoshop with the Image Resource Block technology. Subsequently in 2001, this technology was replaced by XMP and two new schemes were created for photography: first the IPTC Core (2004) and then the IPTC Extension (2007), both exclusively with XMP. The IPTC Core schema allows the management of blocks of information relating to contact print, the content, the representation of the image and the copyright. The IPTC Core associates metadata with the DC, Photoshop and Adobe Rights Management and IPTC IIM schemas. In fact the relationship with the IPTC IIM schema means that it contains most of the original IIM metadata. It must be borne in mind that the IIM metadata is generated in the journalistic field and that much of this is not relevant to the photographic archive. The IPTC Extension resulted from the requirement of professional photographers with the need to manage certain information regarding the flow of business, such as information relating to rights management, to include metadata which was in addition to that included in the initial scheme. It has a lower level of compatibility level compared to the Core, but it works with more specialised data. XMP XMP (Adobe s Extensible Metadata Platform) is a standard for creating, processing, and exchanging metadata. It offers a labelling technology that allows the creation of new metadata and for this to be inserted into the same file. This is XML data, stored using a subset of the W3C Resource Description Framework (RDF). This is especially interesting for the computer industry since the software and the devices can include their own information on the same files. It is also of interest to archives because to gives them the possibility of including their own, properly encoded metadata in the XMP containers. In addition, this metadata is embedded in the files for the TIFF, JPEG, JPEG2000, DNG, PSD and PNG formats. In other formats, a separate metadata file is created. XMP defines four main metadata blocks, the DC and three of its own, plus the specialised blocks: Adobe PDF, Photoshop, Camera Raw and Exif. The inclusion of the DC in the main properties makes it a very useful technology for communication between different platforms. The main contribution of the XMP is its extensibility and the possibility to encode this additional metadata based on standards. It may include metadata created by the archives themselves, which does not necessarily have to be part of any codified standard. In these three standards, EXIF, IPTC and XMP, it is very important to point out that they are encoded in RDF standards. This schema establishes relationships for description and the key in these relationships is the existence of a Namespace. Namespaces are URIs (Uniform Resource Identifiers), resource identifiers, which have their own name. The encoding of interpretable metadata for machines is linked entirely - 8 -
to the concept of the Semantic Web and therefore has a great value in enhancing interoperability. 4. Preservation Digital preservation consists in carrying out a series of tasks so that the digital image continues to be accessible in the future. Involved in this process is the preservation of object materials, the technological tools, and technicians with responsibilities for the matter. The problem of preservation lies in the interpretation of the image, and for this reason we must begin by identifying the causes which hinder this interpretation. Lukas Rosenthaler (Focal Encyclopedia, 2007) established a categorisation of causes for the obsolescence of the files: error in reading the bits, error in reading the format, loss of metadata. Based on this categorisation, it can be established that the preservation of images will depend on the deterioration of the media and the obsolescence of formats, hardware and software. Therefore, strategies that can be devised for preservation must take these three aspects into account. Media The fragility of computer media is something that we have all experienced. To ensure the permanence of data and as the first premise, we can say that one should duplicate data, periodically change the media and verify the integrity of the data contained. This strategy begins with the selection of the preservation media. The guide for the selection of media prepared by the National Archives of the United Kingdom [A Brown, 2008] provides some criteria that may be useful for the selection. It is based on the estimation of longevity, capacity, viability, obsolescence, cost and susceptibility. Formats The formats are a central element in the strategy of preservation. The possibility of accessing the bits contained on a particular medium does not ensure that the image can be displayed. The data from which it is formed needs to be interpretable by the software and hardware. This means recognising the format in which the data is structured and technical specifications which give that data meaning. This is probably the largest challenge of digital preservation since the technological evolution hinders the survival of archives, and their transformation and updating is not always easy to carry out. That is why it is important to have strategies that minimise the risk of obsolescence. The most common strategies are: working with standard formats, migrating to new formats and new versions, and the strict management of metadata for its preservation. With regard to the choice of standard formats, we have the criteria laid down in the guidelines for the selection of formats prepared by the National Archives of the United Kingdom [A Brown, 2008] where the following aspects are considered: ubiquity, sustainability, dissemination, documentation, stability, intellectual property (patents), metadata, complexity, interoperability and viability. Migration is one of the most important functions of preservation. It is the transformation of the image data from the original format to a current archive format. In this way we ensure that we are working with an information structure that is sustainable and over whose evolution we can have greater control. JPEG2000 and TIFF are among the preservation formats most accepted by archives. - 9 -
Metadata Metadata is an important element in the choice of standard formats and in the subsequent migration, but it can also become the focus of the strategy, as within this we find interpretive keys to the files and, therefore, the possibility of keeping them current. This issue has been widely considered and there is abundant literature on the subject. PREMIS (Preservation Metadata Implementation Strategies) is a dictionary of metadata which has, as its objective, the preservation of electronic documents in general and, therefore, also digital photography. The technical metadata largely agrees with that of Z39.87, but PREMIS contains other types of metadata geared exclusively to preservation. Hardware and software We discussed strategies aimed at the images, but we must also take into account the technological environment strategies: the hardware and the software. One of the strategies is to preserve the technology. This involves maintaining in working order, the computers, screens, operating systems and programs that form the technological environment in which a digital image was created at particular moment. This ensures that the data that defines the digital object continues to be current. The other strategy is that of emulation, in essence consisting of recreating, using software, a technological environment that has already been superseded. It can be directed at the hardware, the operating system or specific applications. It has the advantage that there is no intervention to the image file and, therefore, it avoids the hazards involved in any transformation. The strategy is gaining prominence and is present in many digital preservation projects which are developing tools for this purpose. 5. Conclusions This guide provides the basic content for understanding the digital image with the aim of facilitating its management within the ambit of the archive. The fundamental idea of the digital image resides in the fact that it contains the information about each pixel of the image in a way that allows its content to be interpreted by different devices: a camera, a screen or a printer. No further intermediaries are required beyond those set out by the nature of the digital: hardware, software, technical specifications, etc. This technological context, which allows the viewing, editing or printing of images, does not appear to have defined boundaries. Any added functionality can be integrated into a digital object. And here, primarily, lies the role of the archive which relies on the informative potential of metadata. This is the major player in the management of digital image collections, because it is involved in almost all functions of the archive and, most notably, in the preservation and recovery of information. - 10 -
Bibliography and Webliography ANSI/NISO Z39.87. Data dictionary technical metadata for digital still images. NISO, 2006. http://djvu.org/docs/z39-87-2006.pdf Arms, Caroline, R.; Carl Fleischhauer. Digital formats: factors for sustainability, functionality, and quality. Washington: Library of Congress, 2005. http://memory.loc.gov/ammem/techdocs/digform/formats_ist05_paper.pdf Exchangeable image file format for digital still cameras: Exif version 2.3. JEITA, 2010. http://www.cipa.jp/english/hyoujunka/kikaku/pdf/dc-008-2010_e.pdf Extensible Metadata Platform (XMP) Specification: Part 1, Data Model, Serialitzation and Core Properties. Adobe, 2012. http://www.adobe.com/content/dam/adobe/en/devnet/xmp/pdfs/xmpspecificationpart1. pdf Extensible Metadata Platform (XMP) Specification: Part 2, Additional Properties. Adobe, 2012. http://www.adobe.com/content/dam/adobe/en/devnet/xmp/pdfs/xmpspecificationpart2. pdf Extensible Metadata Platform (XMP) Specification: Part 3, Storage in Files. Adobe, 2012. http://www.adobe.com/content/dam/adobe/en/devnet/xmp/pdfs/xmpspecificationpart3. pdf Frey, Franziska. Measuring quality of digital masters. A: Guides to quality in visual resource imaging. Council on Library and Information Resources, 2000. http://www.diglib.org/pubs/dlf091/dlf091.htm#visguide4 Iglésias Franch, David. La definición de un mapa de información conceptual para el archivo fotográfico. En: 12es Jornades Imatge i Recerca. Girona: Ajuntament de Girona, 2012. P. 111-120. http://www.girona.cat/sgdap/docs/vu23efriglesias.pdf International Press Telecommunication Council, IPTC PhotoMetadata White Paper 2007, 2010: http://www.iptc.org/std/photometadata/0.0/documentation/iptc- PhotoMetadataWhitePaper2007_11.pdf PRONOM National Archives [UK], 2002. http://www.nationalarchives.gov.uk/pronom (Online register of file formats and of the software used for these formats). - 11 -