AccessComics: An Accessible Digital Comic Book Reader for People with Visual Impairments
Yunjung Lee, Hwayeon Joh, Suhyeon Yoo, Uran Oh · 2021 · Proceedings of the 18th International Web for All Conference (W4A) · doi:10.1145/3430263.3452425
Summary
This paper from Ewha Womans University in Seoul presents AccessComics, a web-based accessible digital comic book reader for people with visual impairments (PVI). Comics are a popular medium available on many digital platforms, yet almost none support screen readers, leaving blind and low vision readers with extremely limited access. The research began with a formative online survey of 68 participants (46 blind, 22 low vision) from South Korea's largest online community for PVI, exploring their prior experiences with comics and their preferences for accessible digital reading formats. The survey found that most participants had read comics and continued to do so despite accessibility barriers — accessing them through audiobooks, digital devices, asking others to read aloud, tactile versions, or animated adaptations. Scene descriptions were the most desired information type (N=57), followed by facial expressions (N=42), character postures (N=39), and scripts/dialogue (N=38). Preferences were split nearly evenly between audiobooks and eBooks, each offering distinct advantages. Based on these findings, AccessComics was designed as a screen reader-compatible web application using HTML, JavaScript, and CSS with Amazon Polly for text-to-speech, assigning distinct synthesized voices to different characters. The system structures comics into introduction pages (character names, appearances, voice previews) and reading pages that present panel-by-panel descriptions including overall scene, character descriptions with facial expressions, and dialogue. Users can customize reading units (panel, strip, or page), speech rate, filtering of specific information types, and automatic versus manual reading.
Key findings
The prototype evaluation with eight blind participants achieved a SUS score of 76.6 (Grade B, "good"), with participants who preferred eBooks scoring higher (82.5) than those preferring audiobooks (66.7). Three of eight participants preferred AccessComics over both audiobooks and eBooks, valuing the character introductions, voice previews, and information filtering capabilities. The most valued feature was the introduction page with character voice previews, which helped users associate distinct voices with characters before reading. Key insights included: synthesized voices were preferred for speed reading since human voice recordings become unintelligible at high speeds; participants wanted to customize which types of visual information they receive (some wanted all details, others found balloon shapes and panel layouts distracting); larger reading units (page mode) were preferred over panel-by-panel to reduce constant swiping; background sounds and sound effects were desired for immersion but must not overlap with dialogue; and the ability to search text and check word spellings was valued from the eBook experience. Participants noted that existing digital comic platforms consist only of images with no descriptions, making them completely inaccessible to blind users.
Relevance
This research opens up a largely unexplored area of digital accessibility — comic books and graphic narratives are a significant cultural medium with millions of readers, yet accessibility efforts have focused almost entirely on prose text, photos, and web content. The formative survey with 68 PVI participants is one of the largest studies of its kind for comic accessibility and provides robust evidence of demand. The design implications are broadly applicable: the need for customizable information density (not everyone wants every visual detail described), the value of character-specific voices for distinguishing speakers in audio formats, the preference for larger reading units to reduce interaction fatigue, and the complementary strengths of audiobook features (background sounds, voice acting, immersion) and eBook features (speed control, search, spelling access). The work demonstrates how computer vision techniques for panel segmentation and character recognition could be combined with text-to-speech to automate accessible comic production at scale, addressing the current bottleneck where manual production takes six months or more per title.
Tags: visual impairment · blindness · low vision · content accessibility · screen readers · audio description · digital media · reading accessibility