Study on text information presentation method under AR environment

Fontworks Inc., Ltd. has collaborated with Hirose, Tanigawa, Narumi Laboratory, Graduate School of Information Engineering, The University of Tokyo, and Hirota Laboratory, Graduate School of Information Science and Engineering, The University of Electro-Communications (Specially Appointed Assistant Professor Sho Sakurai) to create an AR environment (augmented reality). In order to verify the method of effectively presenting character information in an augmented reality (AR) environment and the font characteristics that are considered appropriate in various information presentation situations, "Study on character information presentation method under AR environment" was carried out.

Research overview

1. Classification of purpose of information presentation in AR environment

To select an appropriate font in the AR environment, first Clarify the purpose and circumstances for which text information is presented thought it was necessary.
From a survey of previous research, etc.,
(A) Purpose of presentation
(B) Presentation method
(C) Presentation status
classified as follows:

(A) Classification by purpose of information presentation

For the purpose of presenting character information, The purpose is to get people to pay more attention to the text information than the content (=discovery type) When, The purpose is to have the content understood (= understanding type) classified into

・Discovery type
Information presentation whose main purpose is to draw attention to the information being presented. Examples include user interfaces such as menus and notifications, Advertisement, billboards, and alerts.

・Understanding type
It is the presentation of information whose main purpose is to get the content of the information to be scrutinized. Examples include online articles, manuals, and email Text.

(B) Classification by information presentation method

You can freely move the position of the character information in the AR space. Things that are fixed in space and presented in association with places and things (= spatially fixed presentation) When, Something that follows the user and is always presented in the user's field of view (=user-following presentation) classified into

・Spatial fixed presentation
It is suitable for presenting information that is fixed to a specific place and is associated with things or places. For example, product information annotation, route navigation, and tourist information display.

・User follow-up presentation
It is suitable for presenting user interfaces, etc., by presenting information that follows the movement of the user. For example, menu screens, notifications, etc.

(C) Classification by information presentation status

Wearable AR devices such as HoloLens make it possible to live while constantly viewing information presented in the AR environment. Depending on whether the background for presenting the information changes significantly, User stationary When User walking classified into

・When the user is stationary
A situation in which the user stays in a particular position and does little more than look around. It is assumed that the background of the textual information presented will not change significantly. For example, a situation where you are sitting and reading emails or online articles.

・When the user moves
A situation where the user moves around, such as walking or riding a mobility vehicle. The background of the presented text information changes greatly. For example, you can check your mail while walking, or present it on an in-vehicle head-up display.

Supplement: Character presentation method
In addition, another factor is Classification by character presentation method is given. There are three methods for displaying characters: a method that displays only the characters (Plain), a method that draws a single-color rectangle called a billboard in the background of the characters (Billboard), and a method that displays blurred edges around the characters (Anti- interference), and a method that casts shadows behind characters (Shadow).
Of these, the billboard method hides the actual background by drawing the billboard. Not suitable. In addition, when billboards are used, it is thought that the situation will be similar to presenting characters written on ordinary paper. , which deals with plain text rendering.

2. Font comparison experiment

You can create an environment that presents 8 character information according to the 3 types of classification of A, B, and C. However, we excluded the following two environments.

・(A) Discovery type, (B) Spatial fixed presentation, (C) When user moves
In order to conduct an appropriate experiment, it was not possible to assume that a large space would be required, and that the size of the space and the location where it should be placed would be appropriate.

・(A) comprehension type, (B) spatially fixed presentation, (C) when user moves
We could not assume a situation where the user moves around and checks the spatially fixed presentation of comprehensible text.
Therefore, we investigated the following six environments, excluding two. The first research focused on spatially fixed presentation, and the second research focused on user-following presentation.

当社はデザイン性豊かな多くのフォントを揃えています。その全てのフォントを研究の対象とすることもできますが、研究の背景、実験の規模を考え、第一次研究では18フォントを基本としました。
第一次研究の結果、濃度の高いフォントが選ばれる傾向が見えたため、第二次研究では整理し、10フォントを基本としました。更に、濃度の高い2フォントを研究の対象として追加しました。

Font comparison experiment summary

(1) Classification differences
If the classification of A, B, and C is different, it is thought that the tendency of the optimal font will be different.

(2) Mincho typeface
In the first research, we concluded that the Mincho typeface is superior to the Gothic typeface in presenting discovery-type information when following the user. However, from the results of the second research, it is considered that the Mincho typeface is not very suitable for the AR environment.

(3) Comprehensible type x Fixed space x At rest　

Kiyohara et al. investigated the influence of fonts and display media on comprehension of text content (reference [2]). In this study, it has been shown that Gothic typefaces perform better than Mincho typefaces in tasks that test comprehension of text content.

From this result, it is considered that the Gothic typeface is more advantageous than the Mincho typeface for the presentation of comprehensible information on paper and on a display. It is thought that this characteristic does not make a big difference in the presentation of comprehensible information fixed in space from that on paper. For this reason, Gothic typeface is considered to be more advantageous than Mincho typeface for the presentation of comprehensible information even in spatially fixed AR character presentation.

(4) Comprehensible × User-following × When moving

It is speculated that general square Gothic and round Gothic that eliminate design elements are suitable. The characters swayed considerably during the movement of the experiment.
In order to understand the content in that situation, it is considered that a font with a simpler design is suitable.

(5) Discovery type and understanding type x user tracking x stationary

When excluding the Mincho font data, there seems to be a particularly strong correlation with density, but the appropriate density varies for each font. Fonts with extremely high density, such as RaglanPunch, are thought to be less suitable (although RaglanPunch was not used in the first experiment, similar results are expected).

Experiment details

■Experiment 1-1) (A) Discovery type (B) Spatial fixed presentation (C) User stationary

■Experiment 2-1) (A) Understanding type (B) User tracking presentation (C) When user moves

■Experiment 2-2) (A) Discovery type (B) User tracking presentation (C) When user moves

■Experiment 2-3) (A) Discovery type (B) User tracking presentation (C) When user is stationary

■Experiment 2-4) (A) Understanding type (B) User tracking presentation (C) When user is stationary

Experimental result

Relative scales were derived by Thurston's paired comparison method.
Since "1 combination was performed twice" per subject, it was calculated based on the response rate of the total number of people (rounded to the fourth decimal place). Higher numbers indicate better readability. (This table is from the second research.)

Density is the area of the black part of the character. It is one of the values that indicates the physical characteristics of the font we use, and is the average value for all characters. The maximum value is 1,000,000, which indicates that RaglanPunch is a font with a density of nearly 80%. The data is arranged in order of density.
We also calculated the glyph area, which is the area of the largest inscribed rectangle surrounding a character.

References
[1] Michele Fiorentino, Saverio Debernardis, Antonio E Uva, and Giuseppe Monno. Augmented reality text style readability with see-through head-mounted displays in industrial context, Presence: Teleoperators and Virtual Environments, Vol. 22, No. 2, pp. 171–190, 2013
[2] Kazuaki Kiyohara, Minoru Nakayama, Hiroshige Kimura, Hideo Shimizu, and Yasutaka Shimizu (2003) The influence of text display media and display format on text comprehension, Japanese Journal of Educational Technology, 27(2), 117-126