Reading Speed, Image Quality Ratings, and Comfort Ratings in Augmented Reality

Anand Singh; Minjung Kim; Saeideh Ghahghaei Nezamabadi; Trisha Lian

arxiv: 2604.27203 · v2 · submitted 2026-04-29 · 💻 cs.HC

Reading Speed, Image Quality Ratings, and Comfort Ratings in Augmented Reality

Minjung Kim , Saeideh Ghahghaei Nezamabadi , Trisha Lian , Anand Singh This is my paper

Pith reviewed 2026-05-14 21:16 UTC · model grok-4.3

classification 💻 cs.HC

keywords augmented realityreading speedimage quality ratingscomfort ratingsAR datasetbenchmarkingtext displayheadset evaluation

0 comments

The pith

The Read-AR dataset collects over 11,000 reading speeds and nearly 6,000 quality and comfort ratings to benchmark AR headset text display.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents the Read-AR dataset, gathered from a single controlled setup with more than 11,000 reading speed measurements and almost 6,000 visual quality and comfort ratings across over 80 conditions. This uniform collection method turns the data into a reference standard for comparing different augmented reality headset architectures. A reader would care because text rendering is a primary AR use case, and shared benchmarks can help identify which display designs support faster, clearer, and more comfortable reading. The work focuses on establishing this dataset rather than testing new hardware or algorithms.

Core claim

The authors assembled a large collection of reading speed data along with subjective image quality and comfort ratings under tightly controlled conditions on one experimental AR setup, producing a resource that functions as a common reference for benchmarking the text display quality of varied AR headset architectures.

What carries the argument

The single consistent and controlled experimental setup used across all 80-plus conditions.

Load-bearing premise

The tested conditions and participant responses produce ratings and speeds that apply beyond this specific lab setup to other AR devices and real-world situations.

What would settle it

New measurements of reading speed, quality ratings, or comfort taken with the same text content but on a different AR headset or in uncontrolled daily environments that deviate substantially from the dataset values.

read the original abstract

The rendering and display of text is a key use-case for augmented reality (AR). Here, we present the Read-AR, a dataset of reading in AR, for which we collected over 11,000 reading speeds and almost 6000 visual quality and comfort ratings across over 80 different experiment conditions on the same experiment set-up. The consistent, controlled set-up enables the dataset to function as a reference for benchmarking the quality of different AR headset architectures.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

A sizable new dataset of AR reading speeds and ratings from one fixed setup, but no evidence it transfers as a benchmark for other headset architectures.

read the letter

This paper collects a new dataset called Read-AR with over 11,000 reading speed measurements and nearly 6,000 quality and comfort ratings across 80 conditions, all gathered on the same controlled AR rig. That scale and consistency is the real addition here, and it gives HCI researchers a solid pile of data on how text rendering parameters affect reading in augmented reality under fixed conditions. The authors do a straightforward job of running the same protocol repeatedly, which keeps the collection clean and comparable within their setup. That part is useful for anyone who wants baseline numbers on speed and subjective ratings for AR text display. The soft spot is the benchmarking claim. Everything comes from one single experiment setup with no cross-headset measurements, no tests on different architectures, and no analysis showing how the metrics would shift or normalize elsewhere. The dataset therefore describes one implementation in detail rather than supplying a validated reference that others can apply directly. The abstract also skips participant counts, exact stats, and error handling, so the reliability of the numbers is hard to judge from the summary alone. This work is aimed at AR display researchers who need raw data for their own experiments or comparisons within similar rigs. A reader expecting a ready-made benchmark for arbitrary headsets will find the evidence thin. I would send it to peer review because the data collection itself looks like a practical resource worth checking in full, even if the generalization argument needs tightening or additional experiments.

Referee Report

1 major / 1 minor

Summary. The manuscript presents the Read-AR dataset, which comprises over 11,000 reading speed measurements and nearly 6,000 visual quality and comfort ratings collected across more than 80 experimental conditions using a single fixed augmented reality setup. The authors assert that the consistent, controlled nature of this data collection allows the dataset to serve as a reference benchmark for evaluating the quality of different AR headset architectures.

Significance. If the experimental conditions are fully specified and replicable, the large-scale collection of reading speeds and subjective ratings under controlled conditions could provide a useful standardized reference point for future AR text rendering research. However, the claimed benchmarking utility across architectures does not hold without supporting cross-headset data or mapping procedures.

major comments (1)

[Abstract] Abstract: The claim that 'the consistent, controlled set-up enables the dataset to function as a reference for benchmarking the quality of different AR headset architectures' is not supported by the reported work. All 11,000+ speeds and 6,000 ratings derive from a single rig; the manuscript contains no cross-headset measurements, no sensitivity analysis to architectural parameters, and no normalization or mapping procedure that would allow transfer to other headsets.

minor comments (1)

The abstract asserts data collection at scale but provides no details on participant numbers, statistical analysis methods, error handling, or validation procedures; these must be explicitly reported in the methods section for the dataset to be usable as a reference.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful review. We agree that the abstract claim about benchmarking across AR headset architectures is not supported by the single-rig data and will revise it. We respond to the comment below.

read point-by-point responses

Referee: [Abstract] Abstract: The claim that 'the consistent, controlled set-up enables the dataset to function as a reference for benchmarking the quality of different AR headset architectures' is not supported by the reported work. All 11,000+ speeds and 6,000 ratings derive from a single rig; the manuscript contains no cross-headset measurements, no sensitivity analysis to architectural parameters, and no normalization or mapping procedure that would allow transfer to other headsets.

Authors: We agree that the dataset was collected exclusively on a single fixed AR setup with no cross-headset measurements, sensitivity analysis to architectural parameters, or mapping/normalization procedures. The original abstract phrasing overstated the dataset's direct utility for benchmarking different architectures. We will revise the abstract to remove this claim and instead describe the work as providing a large-scale, controlled collection of reading speeds and subjective ratings under replicable conditions that can serve as a reference baseline for future AR text rendering studies. revision: yes

Circularity Check

0 steps flagged

No circularity: purely empirical dataset with no derivations or self-referential claims

full rationale

The manuscript collects reading speeds and subjective ratings across 80 conditions on a single fixed AR rig and presents the resulting dataset. No equations, models, fitted parameters, or derivations appear in the provided text. The claim that the controlled setup allows the data to serve as a benchmark is a descriptive statement about the experimental design, not a prediction or result obtained by reducing one quantity to another by construction. No self-citations, uniqueness theorems, or ansatzes are invoked. The contribution rests on the raw measurements themselves.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The work is empirical data collection; no free parameters, axioms, or invented entities are introduced in the abstract.

pith-pipeline@v0.9.0 · 5376 in / 1077 out tokens · 45893 ms · 2026-05-14T21:16:21.145548+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

The consistent, controlled set-up enables the dataset to function as a reference for benchmarking the quality of different AR headset architectures.

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

5 extracted references · 5 canonical work pages

[1]

(2025).Photoshop 2025(Version 26.8.1)

Adobe Inc. (2025).Photoshop 2025(Version 26.8.1). https://www.adobe.com/products/photoshop.html Atilgan, N., Xiong, Y.-Z., & Legge, G. E. (2020). Reconciling print-size and display-size constraints on reading.Proceedings of the National Academy of Sciences,117(48), 30276–30284. Bailey, I. L., & Lovie, J. E. (1976). New design principles for visual acuity ...

work page 2025
[2]

READING IN AR 20 Dillon, A. (1992). Reading from paper versus screens: A critical review of the empirical literature.Ergonomics,35(10), 1297–1326. Dobres, J., Chahine, N., Reimer, B., Gould, D., Mehler, B., & Coughlin, J. F. (2016). Utilising psychophysical techniques to investigate the effects of age, typeface design, size and display polarity on glance ...

work page 1992
[3]

Huey, E. B. (1898). Preliminary experiments in the physiology and psychology of reading. The American Journal of Psychology,9(4), 575–586. Hughes, L., & Wilkins, A. (2000). Typography in children’s reading schemes may be suboptimal: Evidence from measures of reading rate.Journal of Research in Reading,23(3), 314–324. International Organization for Standar...

work page arXiv 2000
[4]

A., Mansfield, J

O’Brien, B. A., Mansfield, J. S., & Legge, G. E. (2000). The effect of contrast on reading speed in dyslexia.Vision research,40(14), 1921–1935. Pelli, D. G., & The EasyEyes Team. (2025, July 14).EasyEyes(Version 2025-07-14). https://easyeyes.app READING IN AR 23 Pelli, D. G., Tillman, K. A., Freeman, J., Su, M., Berger, T. D., & Majaj, N. J. (2007). Crowd...

work page 2000
[5]

F., Matteson, S., Gould, D., Chahine, N., & Levantovsky, V

Reimer, B., Mehler, B., Dobres, J., Coughlin, J. F., Matteson, S., Gould, D., Chahine, N., & Levantovsky, V. (2014). Assessing the impact of typeface design in a text-rich automotive user interface.Ergonomics,57(11), 1643–1658. Şimşek, B., Direkci, B., Koparan, B., Canbulat, M., Gülmez, M., & Nalçacıgil, E. (2025). Examining the effect of augmented realit...

work page doi:10.2352/ei.2023.35.8.iqsp-310 2014

[1] [1]

(2025).Photoshop 2025(Version 26.8.1)

Adobe Inc. (2025).Photoshop 2025(Version 26.8.1). https://www.adobe.com/products/photoshop.html Atilgan, N., Xiong, Y.-Z., & Legge, G. E. (2020). Reconciling print-size and display-size constraints on reading.Proceedings of the National Academy of Sciences,117(48), 30276–30284. Bailey, I. L., & Lovie, J. E. (1976). New design principles for visual acuity ...

work page 2025

[2] [2]

READING IN AR 20 Dillon, A. (1992). Reading from paper versus screens: A critical review of the empirical literature.Ergonomics,35(10), 1297–1326. Dobres, J., Chahine, N., Reimer, B., Gould, D., Mehler, B., & Coughlin, J. F. (2016). Utilising psychophysical techniques to investigate the effects of age, typeface design, size and display polarity on glance ...

work page 1992

[3] [3]

Huey, E. B. (1898). Preliminary experiments in the physiology and psychology of reading. The American Journal of Psychology,9(4), 575–586. Hughes, L., & Wilkins, A. (2000). Typography in children’s reading schemes may be suboptimal: Evidence from measures of reading rate.Journal of Research in Reading,23(3), 314–324. International Organization for Standar...

work page arXiv 2000

[4] [4]

A., Mansfield, J

O’Brien, B. A., Mansfield, J. S., & Legge, G. E. (2000). The effect of contrast on reading speed in dyslexia.Vision research,40(14), 1921–1935. Pelli, D. G., & The EasyEyes Team. (2025, July 14).EasyEyes(Version 2025-07-14). https://easyeyes.app READING IN AR 23 Pelli, D. G., Tillman, K. A., Freeman, J., Su, M., Berger, T. D., & Majaj, N. J. (2007). Crowd...

work page 2000

[5] [5]

F., Matteson, S., Gould, D., Chahine, N., & Levantovsky, V

Reimer, B., Mehler, B., Dobres, J., Coughlin, J. F., Matteson, S., Gould, D., Chahine, N., & Levantovsky, V. (2014). Assessing the impact of typeface design in a text-rich automotive user interface.Ergonomics,57(11), 1643–1658. Şimşek, B., Direkci, B., Koparan, B., Canbulat, M., Gülmez, M., & Nalçacıgil, E. (2025). Examining the effect of augmented realit...

work page doi:10.2352/ei.2023.35.8.iqsp-310 2014