Details of description part II: Describing images in practice - Tech Forum 2024

This presentation explores the practical application of image description techniques. Familiar guidelines will be demonstrated in practice, and descriptions will be developed “live”! If you have learned a lot about the theory of image description techniques but want to feel more confident putting them into practice, this is the presentation for you. There will be useful, actionable information for everyone, whether you are working with authors, colleagues, alone, or leveraging AI as a collaborator.

Link to presentation recording and transcript: https://bnctechforum.ca/sessions/details-of-description-part-ii-describing-images-in-practice/

Presented by BookNet Canada on June 25, 2024, with support from the Department of Canadian Heritage.

1. Details of description part II: Describing images in practice Leah Brochu June 25, 2024

2. Overview Image Description Guidelines Image Description Techniques Alt-text Long descriptions Writing and editing Collaborators Humans Robots! (Artificial Intelligence) Resources

3. Image Description Guidelines

4. Overview of Image Description Guidelines Decorative images do not need to be described Context is key Consider your audience Use a clear structure Be (or try to be!) concise Use present tense Be objective (with nuance!) Do not censor Write out text in images

5. Is it decorative or non-decorative? Always determine if an image is decorative or non-decorative Ask yourself: Is it important/meaningful to the narrative or the content? What’s the likelihood of it being discussed by readers? If left undescribed, would it have a negative impact on the reader’s experience?

6. Context is Key! Always take the context into consideration The context will help dictate the level of description needed

7. Consider the Audience! The audience is part of the context – images may need to be described differently depending on whether they are being written for a child, an adult, or a PhD student If the audience is highly knowledgeable, you may need to consult subject matter specialists

8. Write Clearly Structured Descriptions! Broadly: work from the General to the Specific Consider the context – the reason for the picture being there – this will inform the structure

9. Flowers – examples Seven people recline around a table on a marble stage, among broad marble columns. Behind them, there is a statue of an adult, a child, and a dog. The room is in the open air and behind them, hills and mountains are visible. The reclining diners look out toward the floor in front of them, which has people lying on cushions, looking this way and that, as thousands of pink petals blanket them. OR A group of Romans dine at a banquet; some at a table on a raised platform, and some on the floor before them. The diners on the floor are being blanketed in thousands of pink petals that float down from above.

10. Be (or try to be) Concise! Keep alt-text descriptions to just a few sentences at the most If you need to write more, write a long description and include it somewhere in the book (often used for maps, graphs, charts, and other situations where multiple paragraphs, lists, or other structure is needed for clarity) Why be concise? To avoid cognitive overload! Steps to preventing cognitive overload: Explain core concepts first Eliminate unnecessary information Draw attention to the most important elements Use simple and straightforward language

11. Use the Present Simple Tense! Whenever possible, use present simple tense Present continuous: “A few minutes later” Peter Parker is looking up at a wrestling ring where a bald, muscular man in briefs is holding a slightly smaller, but still muscular, man in the air over his head. Peter is is stroking his chin and thinking “Hmmm… this will be a good chance to test my power again!” A large sign by the ring reads “$100 to the man who can stay in the ring three minutes with Crusher Hogan!” A crowd of people is excitedly watching the wrestlers.” Present simple: “A few minutes later” Peter Parker comes upon a wrestling match where a bald, muscular man in briefs holds a slightly smaller, but still muscular, man in the air over his head. Peter strokes his chin and thinks “Hmmm… this will be a good chance to test my power again!” A large sign offers “$100 to the man who can stay in the ring three minutes with Crusher Hogan!” A crowd excitedly watches the wrestlers.”

12. Be Objective! Image descriptions should be objective, so that the writer’s opinion does not influence the reader But be careful: don’t be overly objective in a way that could distract from the intent of the image

13. Objectivity in Humour This image is a comic panel. It shows a playground slide with a spider web constructed at the bottom. Two spiders on the side of the slide, looking at the web. The caption at the bottom of the image reads, "If we pull this off, we'll eat like kings." The scene humorously suggests that the spiders plan to catch something big enough in their web to provide a feast. OR In a black and white cartoon drawing, t wo spiders sit on a near the bottom of a playground slide where they have constructed a large spiderweb. One says to the other: “If we pull this off, we’ll eat like kings.”

14. Objectivity Elsewhere Public Doman Review: The Book of Hallowe’en

15. Objectivity Elsewhere – Descriptions Too objective: This image is a black-and-white photograph titled "In Hallowe'en Time." It shows a rural scene with several stacks of harvested crops, likely cornstalks, arranged in tepee-like formations in a field. In the background, there is a wooden building, possibly a barn, and several leafless trees, suggesting an autumn setting. The scene evokes a rustic, seasonal atmosphere typical of Halloween time in the countryside. (ChatGPT) Reasonably objective: A sepia tone photo titled “In Hallowe’en Time”. Bunches of harvested cornstalks are arranged into tepee-like shapes, set in a field in front of a barn. The drying vegetation and leafless trees in the background make for an eerie scene.

16. Do Not Censor! All readers deserve the same experience If you can’t describe a graphic image, ask someone else Maybe even the author! Aino Ackte as Salome, by Norman Morrow , 1910 Gertrude Hoffmann  with the head of  John the Baptist   (1908) – Public Domain

17. Write out Text in Images! Text within images needs to be written out – all readers must be able to receive the same information – this is part of the Web Content Accessibility Guidelines . If the image has legible text that you expected sighted readers to read, it needs to be available to assistive technology as well

18. Text in Images: AI Example ChatGPT: This image is a vintage advertisement for Camel cigarettes. The ad features a doctor in pajamas sitting up in bed, talking on a telephone, with a look of concern on his face. Next to him on the nightstand are a lamp, an ashtray with a cigarette, a clock, and other items. The text at the top left says, "I'll Be Right Over!" Below that, it reads, "24 hours a day your doctor is 'on duty'... guarding health... protecting and prolonging life..." This is followed by a passage emphasizing the dedication of doctors to their patients. The main slogan in the middle of the ad states, "More Doctors Smoke Camels Than Any Other Cigarette!" The text further claims that, according to a recent nationwide survey, more doctors choose Camel cigarettes over other brands. At the bottom right, there is an image of a pack of Camel cigarettes with a smaller inset of a woman holding a cigarette. The ad includes text promoting the smoothness and taste of Camel cigarettes, appealing to both doctors and the general public.

19. Text in Images: Human Example A full-page vintage advertisement for Camel Cigarettes. In a drawing, a doctor answers the phone from bed; on the nightstand, his bedside lamp is on, and there is a watch and doctor’s bag. Above this, text reads: "I'll Be Right Over!" Below that: "24 hours a day your doctor is 'on duty'... guarding health... protecting and prolonging life... Plays... novels... motion pictures have been written about the "man in white" and his devotion to duty. But in his daily routine he lives more drama, and displays more devotion to the oath he has taken, than the most imaginative mind could ever invent. And he asks no special credit. When there's a job to do, he does it. A few words of cheer... a few puffs of a cigarette... and he's back at the job again.... Below the image, large text reads: According to a recent Nationwide survey: More Doctors smoke camels than any other cigarette. Below this, small text reads: Tens of America's leading independent research organizations asked 113,597 doctors this question: "What cigarette do you smoke, Doctor?" The brand named most was Camel! Doctors in every branch of medicine! Doctors in every branch of medicine! Three independent research organizations asked this question of 113,597 doctors: "What cigarette do you smoke, Doctor?" Camel was the brand named most! Yes, doctors, too, smoke for pleasure. Camel's mildness and flavor are just as enjoyable to them as they are to you or any smoker. Smoke Camels and see for yourself. Beside this, there is a drawing of an open pack of Camel cigarettes overlaid on top of another text area. Here there is a drawing of a woman smoking a cigarette and the text: The "T-ZONE" Test Will Tell You The "T-Zone"—T for taste and T for throat—is your own proving ground for any cigarette. For only your taste and your throat can decide which cigarette tastes best to you... and how it affects your throat. Based on the experience of millions of smokers, we believe Camels will suit your "T-Zone" to a "T.“ Finally, at the very bottom, it says: Camels: Costlier Tobaccos.

20. Text in Images: Human Example (Illegible Small Text A full-page vintage ad for Camel cigarettes, emphasizing that doctors choose Camel cigarettes. In a central image a pajama-wearing doctor answers the phone from bed; a doctor’s bag, glasses, watch and lamp are on the nightstand. Text above this reads “I’ll be right over”, and “…24 hours a day your doctor is ‘on duty’… guarding health…protecting and prolonging life…” Smaller unreadable text appears underneath. Below the image, large text reads: According to Nationwide survey: More Doctors smoke Camels than any other cigarette! More unreadable text appears below, beside a drawing of a woman smoking a cigarette, and a pack of Camels. A headline over the woman reads “The ‘T-Zone’ test will tell you”. Finally, at the bottom, it reads: Camels: Costlier Tobaccos.

21. Image Description Techniques

22. Writing Alt-text: Quick-glance Approach Decide if an image is decorative Think about the context and audience Think about what stands out to you Stop looking at the image! Draft a short description based on what your brain recalls as important Look again at the image and context, and add anything else that you decide is important.

23. Quick Glance – Alt-text A large spherical terrarium with English ivy inside …it is in a dark room and a lamp shines light onto it.

24. Writing Alt-text: Keep-on staring approach Decide if an image is decorative Think about the context and audience Think about what stands out to you Draft your description while continually looking back to the image, noticing and noting more and more things, Look again at the image and context, and start editing down your description, cutting out anything else that you decide is important.

25. Keep-on-staring approach: alt-text A large terrarium, around 18 inches in diameter, containing vines of English ivy. The clear glass of the bottle is foggy inside from the humidity within. The bottle stands in a darkened room; a small spiky plant stands beside it, and floral curtains are visible in the background. The curtains are drawn and the room is dark; the terrarium is lit by a single lamp.

26. Writing Alt-text: Chat-GPT Collaboration Approach Decide if an image is decorative Think about the context and audience Think about what stands out to you Copy and paste, or upload, the image to an AI, and instruct it to describe the image. Use this as your starting point – review, edit, correct, and refine what the AI returns.

27. Long descriptions: Graphs Review the context of the graph – it may be fully described in the text, so take a look! Not all graphs will require a long description. But if it does: Provide the title of the graph and its main purpose D escribe the layout and type of the graph, and the information on the x and y axes. Provide an over view of the info shown on the graph. If more detail is needed for understanding, create a long description This can break down all of the info presented by the graph, or can be provided as a data table, if appropriate

28. Long Descriptions: Simple Graph Example Graph displaying observed and predicted Australian Post Output Value over time – Wikimedia Commons Alt-text: Dot graph displaying the observed and predicted values of the Australian Outpost from 1976 to 2002. The years are on the x axis, and total output value in millions of dollars is on the y, ranging from 400-4000. Observed and predicted values are nearly identical, and the value increases steadily each year, from about 500,000,000 in 1976 to 3,900,000,000 in 2002, at a rate of around 100,000,000 per year.

29. Long Descriptions: Complex Graph Example (alt-text) Causes of Mortality in the Army from Wellcome Collection Alt-text: Diagram of the causes of mortality in the army in the East, from April 1854 to March 1855. Data is shown as a pie chart, with wedges of different sizes and heights, and multiple colours per wedge. Follow link below image for long description.

30. Long Descriptions: Complex Graph Example (Long Description) Causes of Mortality in the Army from Wellcome Collection Long Description: Text reads: The Areas of the blue, red and black wedges are each measured from the centre as the common vertex. The blue wedges measured from the centre of the circle represent area for area the deaths from Preventible or Mitigable Zymotic diseases, the red wedges measured from the centre : the deaths from wounds; & the black wedges measured from the centre the deaths from all other causes. The black line across the red triangle in Nov 184 marks the boundary of the deaths from all other causes during the month. Each wedge represents a month: April 1854: A very small wedge, just black. May 1854: Wedge a tiny amount bigger than April; 85% black, 15% blue (approximate) June 1854: Wedge slightly smaller than May; 30% black, 70% blue. Etc.!

31. Long descriptions: Maps Review the context of the map – it may be fully discussed in the text, so take a look! Not all maps will require a long description. But if it does: Include the Name/Title of the map and include a description of the legend. Describing maps can quickly become overwhelming – make sure to focus on its purpose – don’t include every detail! Describe it in a structured way, such as: Quadrants (Top left, top right, etc.) Compass directions: This is similar to discussing the quadrants of an image, but uses compass directs instead (e.g., North, Northeast, West, Southwest, etc.) Clock: divide the image by the hours of a clockface Check out Diagram Centre for examples!

32. Long descriptions: Diagrams Review the context of the diagram – it may be fully discussed in the text, so take a look! Not all diagrams will require a long description. But if it does: Describe the diagram systematically Use lists, headings, sub-headings – any structure you can use that makes sense is going to be helpful!

33. Long descriptions: Diagram Example

34. Long descriptions: Diagram Example - ChatGPT Alt-text: A diagram showing steps and information about thorough handwashing. See link below for long description. The image shows a series of steps for hand hygiene. The steps depicted include: Apply soap. Rub palms together. Rub the back of each hand with the palm of the other hand with fingers interlaced. Rub palm to palm with fingers interlaced. Rub the backs of fingers to opposing palms with fingers interlocked. Rub each thumb clasped in the opposite hand using a rotational movement. Rub tips of fingers in opposite palm in a circular motion. Rub each wrist with the opposite hand. There is also an image of a clock indicating that the process should take a certain amount of time, a thumbs-up symbol for the correct method, and a no-symbol for incorrect methods, specifically showing not to skip any steps or wash inadequately. The illustration source is mentioned as www.hegasy.de.

35. Collaborating

36. AI as Collaborator Be mindful. “There’s a fundamental mismatch between this technology and environmental sustainability,” (Adrian deVries , researcher). “I think we still don’t appreciate the energy needs of this technology,” Altman said at a public appearance in Davos. He didn’t see how these needs could be met, he went on, “without a breakthrough.” He added, “We need fusion or we need, like, radically cheaper solar plus storage, or something, at massive scale—like, a scale that no one is really planning for.” (Sam Altman, the C.E.O. of OpenAI)

37. Interactive Time! Form Sheet

38. Any questions? [email protected]

39. Resources Guide to Long Descriptions APLN: Image Descriptions DAISY Webinars AccessiblePublishing.ca