Optimized Type for E-Ink Displays
Christopher Kirwan, Mark Stafford
August 25th, anno domini 2008.
Introduction: Thesis Concept
I plan to develop a typeface family that is ideally suited to the display of text on E-Ink displays. I want to develop an enhanced reading experience on e-paper devices. E-paper, specifically the E-Ink corporation’s active matrix display technology, is emerging as the dominant player in the market, being used by virtually every currently produced e-book reader on the market today. As a relatively young technology, there is a lot of work to be done in terms of optimizing content for display on such devices. In terms of text, this falls into two realms: the rendering of type on screen, and the development of typefaces specifically suited for E-Ink screens.
E-ink is physically different from the computer moitors we’re used to—both in terms of how the “pixels” are laid out, and how they handle sub-pixel rendering. This means that typefaces and screen rendering technologies developed for LCD screens won’t necessarily be effective on devices like Sony’s PRS 505 or Amazon’s Kindle. Just as type was adapted for screen using a combination of traditional typographic tweaking (changes in x-height and ascender height, opening of counters etc) and digital solutions (hinting, sub-pixel rendering), I believe type can and must again be transformed, leading to a more pleasurable reading experience.
II. Descriiption of “dirty” prototype(s).
At the moment, my reseach has been limited to testing existing typefaces on an Amazon Kindle, seeing how they measure up against the default display text. I am counting on classes I’m taking this semester for the tools I need for designing type and exploring the physical qualities of digital ink. Until then, prototyping is an impossibility.
Fig 1: default text on the Kindle’s screen. Note how the basic pixel size is clearly not the smallest discrete unit the screen can display. Lack of fine control over microcapsule elements in the display result in a “salt-and-pepper” effect which ultimately reduces contrast.
Figure 2: A comparison between Hoefler Text, Georgia, Minion Pro and the Kindle’s default typeface (PMN Caecilia). The first three typefaces were set in photoshop and exported as a raster image, which was then inserted into a Word document along with the block of plain text. The plain text was originally set in Times New Roman, but got converted to PMN Caecilia upon upload to the device.
Figure 3: The first line of each block, presented here for closer comparison. Note the relative darkness of the fourth (default) line, as opposed to the “image” lines.
III. What were your assumptions? What did you learn from testing your assumptions?
I am assuming that e-paper behaves in a way that allows sub-pixel addressing, and that this can be controlled finely enough to allow for sophisticated hinting; that the same principles of letterform design that make typefaces like Georgia, Verdana, Constantia etc. more readable on a computer screen may also apply to e-paper; that current font rasterizers for e-paper devices are primitive, and don’t take advantage of the screen’s physical properties in the way that ClearType hinting does; that since most devices using e-paper are Linux-based, it should be easy to port LaTeX to them and use its advanced typesetting capabilities to improve onscreen typography.
I have yet to be able to test these assumptions.
IV. Why this design concept? What is so “new” about your concept? What is your
motivation? What is your methodology? What are previous explorations in your own
work that are leading you here?
I am fascinated by typography and type design, especially type that is designed to sastisfy real-world (i.e. physical) constraints. This makes the design of type more than a mere exercise in aesthetics, adding the challenge of pragmatism. I am fascinated with type designs that tackle the challenge of screen readability. Since the reading habits of our society are moving steadily towards the digital, the availability of quality screen typefaces becomes critical.
E-Ink devices, mostly dedicated e-book readers, are making headway as a new platform for reading. A large part of the success of these devices comes from the availability of content, but also from the relative ease of reading their displays. To fully take advantage of these displays, there must be typefaces that are optimized for crisp, clear rendering on e-paper, as well as a rasterizer that takes advantages of the physical properties of these screens.
While there is ample research in the domain of screen readability, it is concentrated almost exclusively on LCD screens. This is only natural since they are the dominant form of computer display. In terms of typefaces, there are ample typefaces that are hinted for digital display, but again, the hinting process (by which the type designer encodes specific instructions on how the letterforms are to be rendered at small sizes) is either a generalized method or one that relies on the physical properties of LCD screens. No significant work has been done in screen rasterization or type design for E-Ink screens.
I’m motivated by a desire to see e-paper technology reach its full potential, and by so doing enrich the reading experience for anyone using it. Ultimately, I would like to see digital books that have all the luxuriousness of a limited edition first printing. Freed from the constraints of paper, books in digital form can have expert typography applied to them at no extra cost to the publisher. CSS demonstrates online the potential for applying style templates to digital text. Digital text on e-readers will need considerably more sophisticated styling than CSS provides, but I think that’s an achievable goal. The result would be a democratization of texts: a book that would be produced physically as a cheap, poorly printed and typeset paperback, could be rendered on a digital device with all the typographic refinements of a first rate printing job. While content is still king, context plays a role in people’s perception of a text. A professional typography job leaves the reader with a text that easier to read, endowed with a sense of elegance and luxury. It is this feeling of luxury that I want to impart to digital texts, which are fast outstripping their dead-tree counterparts.
As for methodology, I’m starting off with the question of type design. I will be studying this discipline under Joshua Darden in the Fall semester. I intend to gain a better understanding of how E-Ink screens work through the Digital Ink collab studio, as well as conversation with experts in the field. Once I know how exactly the sub-pixel addressing mechanisms work in such devices, I will be able to design my typefaces accordingly. During this process, I anticipate I’ll learn quite a bit about type rasterization, or how an outline of a letterform is converted to a discrete grid of pixels for display. I suspect that a type rasterizer which is optimized for e-paper will be just as important to my goals (if not more so) as the development of fonts.
Previous research includes my attempts to create a flash-based e-book reader, together with an accompanying XML schema based loosely on LaTeX syntax. This project explored a very different aspect of the e-book problem, that of interface. I wanted to explore the use of the book as an interface metaphor. Most e-readers on the market mimic books to a certain extent, but at some point the metaphor is broken, often in clumsy and jarring ways. Additionally, content provided by Amazon and others tend to use proprietary formats that restrict the variety of materials viewable on the device. I explored the possibility of breaking the book metaphor “elegantly”, in ways that the final result still acted like a book and drew upon familiar interactions, but also transcended the traditional analog book: making a device be not “almost-a-book” but “more-than-a-book”.
V. Research: Primary and Secondary. Whose shoulders are you standing on (precedents
and influences)? Who are our mentors? Who are you speaking to?
The first major advance in terms of digital typesetting must be attributed to Donald Knuth and TeX. Knuth, a computer scientist famous for his publications dealing with the fundamentals of his discipline, set out to devise a digital typesetting system that would enable him to have complete control over how his books were printed. What was meant to be a small side project ended up consuming his attentions for ten years, resulting in the TeX typesetting system. In conjunction with Tex, Knuth worked on a program called METAFONT, which allowed creation and sophisticated manipulation of digital typefaces. He wrote extensively during this period on digital typography, and contributed important breakthroughs in font rendering, page breaking, text flow and justification, and many other aspects of digital type.
An extension of Knuth’s work, Leslie Lamport’s LaTeX document preparation platform is essentially a layer that runs on top of TeX, allowing for more intuitive creation of documents. TeX, while robust, is considered a “low level” system. Doing complex typography involves long sequences of commands, often grouped into macros. LaTeX is essentially a set of predefined and reusable macros, like a library that a programming language may use. One might compare TeX to a low- or mid-level programming language like C, and LaTeX to a high-level language like Python or Ruby.
Bot TeX and LaTeX are open source software, and are very popular in academic circles for their ability to display complicated mathematical formulas and advanced handling of citations. LaTeX inspired my early project of creating an open-source XML schema for digital texts. Knuth’s research on rasterization and defining font outlines is a helpful basis for the work I will be undertaking.
As I’ve stated, considerable effort has gone into the design on LCD-compatible typefaces. Matthew Carter’s Verdana, Georgia and Tahoma are the first exmples of fonts for screens, and are widely used today. Part of the process of making them successful involved a massive amount of hinting, done by Tom Rickner of Monotype. Microsoft, who commissioned the above faces from Carter, went a step further and developed the ClearType type rendering engine, further improving the crispness of on-screen type. To demonstrate ClearType’s advanced capabilities, the Microsoft Typography team headed by Bill Hill commissioned a family of typefaces that became the default typefaces used in the Vista operating system. The final designs include work by Lucas De Groot, Gary Munch, Jelle Bosma, Steve Matteson, Robin Nicholas, John Hudson and Jeremy Tankard.
Rendering engines for type are mostly proprietary, closed systems. An exception to this is FreeType, an open source font rendering engine used in most Linux systems, as well as in Fontlab Studio. The FreeType project makes available important information on designing and implementing a text rasterizer.
I am currently in discussion with Bill Hill at Microsoft, and am trying to contact Joseph Jacobson at MIT, whose team developed E-Ink and co-founded the E-Ink corporation. I will be attempting to talk with some of the type designers behind the most prominent screen fonts, including Carter, De Groot, and Matteson.
VI. Make a visual map of the domains or fields of study in which your thesis exists.
fig 4: a visual map of the domains of my thesis, exploring sub topics and tangential contexts. Embedded in the visualization is a meta-comment about visual maps included at no extra cost to the reader.
VII. What conclusions have you reached that could be the framework for further
explorations? What are alternative or antithetical prototypes that could further
push your research? What are your next steps?
Shouldn’t this question come at the end of the semester? My research so far hasn’t really led to any conclusions, just questions and directions for exploration. I’ve already said that prototyping is utterly impractical at this stage. I may as well scribble out an alphabet on a page. My next steps are the ones I described in my methodology section. In short: learn how E-Ink displays work, design some fonts accordingly, learn about font rendering. Badabing, badaboom: thesis.
André, Jaques and Herch, Roger D, ed. Raster Imaging and Digital Typography. Cambridge: Cambridge University Press, 1989.
Baines, Phil, and Haslam, Andrew. Type & Typography. New York: Watson Guptil, 2005.
Bartram, Alan. Typeforms: A History. London: The British Library and Oak Knoll Press, 2007.
Bringhurst, Robert. The Elements of Typographic Style. Vancouver: Hartley & Marks, 2005.
Cabarga, Leslie. Learn FONTLAB Fast. Los Angeles: Iconoclassics Publishing, 2007.
—. The Logo, Font & Lettering Bible. Cincinnati: HOW Design Books, 2004.
Cheng, Karen. Designing Type. New Haven: Yale UP, 2005.
Ellison, Andy. The Complete Guide to Digital Type: Creative Use of Typography in the Digital Arts. New York: Harper Collins, 2006.
Gordon, Bob. Making Digital Type Look Good. London: Thames & Hudson, 2001.
Hersch, Roger D., ed. Visual and Technical Aspects of Type. Cambridge: Cambridge University Press, 1993.
Karow, Peter. Digital Typefaces: Descriptions and Formats. Berlin: Springer-Verlag, 1994.
Knuth, Donald E. Digital Typography. Stanford, CA: CSLI Publications, 1999.
—. The METAFONTbook (Computers and Typesetting vol. C). Reading MA: Addison Wesley, 1986.
—. The TeXbook (Computers and Typesetting vol. A). Reading MA: Addison Wesley, 1984.
Lamport, Leslie. LaTeX: A Document Preparation System. Reading MA: Addison Wesley, 1984.
Lawson, Alexander. Anatomy of a Typeface. Boston: David R. Godine, 1990.
Paterson, Donald G. How To Make Type Readable. New York and London: Harper & Brothers, 1940.
Smeijers, Fred. Type Now: A manifesto. London: Hyphen Press, 2003.
Adams, Debra Anne. “A Dialogue of Forms: Letters and Digital Font Design.” MIT Masters Thesis (1978).
Bigelow, Charles. “The Design of a Unicode Font.” Electronic Publishing 6 (1993): 289–305.
Comiskey, Barrett, Albert, J.D., Yoshizawaw, Hidekazu and Jacobson, Joseph. “An Electrophoretic Ink for All-Printed Reflective Electronic Displays.” Nature 394 (1998): 253–255.
Dillon, Andrew et al. “Visual Search and Reading Tasks Using ClearType and Regular Displays: Two Experiments.” CHI proceedings (2006): 503–511.
Gugerty, Leo et al. “The Effects of Subpixel Addressing on Users’ Performance and Preferences During Reading-Related Tasks.” ACM Transactions on Applied Perception 2 (2004): 81–101.
Hersch, Roger D. “Character Generation Under Grid Constraints.” Computer Graphics 4 (1987): 243–252.
Jacobson, Joseph et al. “The Last Book.” IBM Systems Journal 3 (1997): 457–463.
Rueckner, Stan and Uszkalo, Kirsten C. “Binding the Electronic Book: Design Features for Bibliophiles.” Visible Language 41 (2007): 50–69.
Zender, Mike. “Advancing Icon Design for Global Non Verbal Communication: Or What Does the Word Bow Mean?” Visible Language 40 (2006): 178–206.
Zonkger, Douglas E, Wade, Geraldine and Salesin, David H. “Example-Based Hinting of TrueType Fonts.” Proceedings of the 27th annual conference on Computer graphics and interactive techniques (2000): 411–416.