Fig 1.
ACT easy font. Examples of ACT Easy (A) and its use in commercial applications (B). Use of this figure was approved by Isobel Cluskey, Director of Creative, Circle K Procurement and Global Brands.
Fig 2.
Fonts used in Experiment 1. Examples of unfiltered (A) and filtered (B) test sentences for the five fonts are shown. Two red lines mark the equal x-heights across all five fonts.
Fig 3.
Results for one participant in Experiment 1. Reading curves showing reading speed (in words per minute) as a function of print size (in logMAR), for one sample participant. The left and right panels show reading curves under the normal and simulated low vision conditions. Each plot corresponds to one font, with the dots representing the empirically measured reading speed, and the curves showing fitted reading speed.
Table 1.
Means and standard deviations of reading measures.
Fig 4.
Reading results for Experiment 1. Reading curves showing reading speed (in words per minute) as a function of print size (in logMAR), means across all participants. The left and right panels show reading curves under the normal and simulated low vision conditions. Each plot corresponds to one font, with the dots representing the empirically measured reading speed, and the curves showing fitted reading speed. Error bars represent standard errors but are small and hard to distinguish from the data points.
Fig 5.
Summary of the four reading parameters across the five fonts. A. Maximum reading speed in words per minute. B. Critical print size in logMAR. C. Reading acuity in logMAR. D. Preference rating from 1-5. The upper and lower panel show results from the normal and simulated low vision conditions. In each plot, each symbol represents one of the five fonts. Error bars represent standard error across all participants.
Fig 6.
Comparison of model and human reading acuity. The bottom x-axis marks model reading acuity in log units, and the top x-axis marks the corresponding pixels per x-height, with 0 log unit corresponding to 4 pixels per x-height. Each panel represents a model. Each symbol represents a font. Pairs of symbols joined by dashed lines show performance in the normal and simulated low-vision conditions.
Table 2.
Correlation between OCR and human reading acuities across fonts.
Fig 7.
Comparison of acuity for humans and two Gemini models. Comparison of model and human reading acuity under the normal (A) and simulated low vision (B) conditions, for Gemini_15_pro (left) and Gemini_15_flash (right). The bottom x-axis marks model reading acuity in log units, and the top x-axis marks the corresponding pixels per x-height, with 0 log unit corresponding to 4 pixels per x-height. Each symbol represents a font. The hollow symbols represent data from the pilot study, and the solid symbols represent data from the main study. Dashed lines are the regression lines across all fonts. The correlation results are also provided with each plot.