Tiny Scratches Belong to Movies: Perception of Digitally Cleaned Films - dataset

Supplementary repository for the ReFilm study This dataset contains mixed-methods data collected to investigate audience perception of digitally restored films. The data is composed of two primary types collected from 17 participants (12 male, 5 female): (1) quantitative eye-tracking data (n=17) and (2) qualitative transcripts from semi-structured interviews (n=14). Demographic data (age, film consumption, audiovisual education) were also recorded. 1. Stimuli: The stimuli used to elicit data were 16 video clips, comprising eight unique sequences from seven classic Czech films (sourced from Národní filmový archiv, Prague). Each sequence was presented in two versions: "Raw": An unrestored HD transfer reflecting the film's state before digital restoration, containing a spectrum of material flaws (e.g., scratches, splices, dust). "Clean": The corresponding sequence after it had undergone professional digital retouching and color grading. 2. Methodology and Techniques (Quantitative Data): Data was collected in a controlled laboratory setting. Apparatus: A stationary Tobii Pro Spectrum 300 eye-tracker (sampling at 300 Hz) was used to record eye movements. Stimuli were displayed on a 23.8-inch Eizo EV2451 monitor (1920x1080px). Participants were seated approximately 60 cm from the monitor. Metrics were exported using the I-VT fixation filter (velocity threshold: 30 °/s; minimum fixation duration: 60 ms; merging of adjacent fixations within 75 ms and 0.5° of visual angle). Each row in ReFilm Metrics_fixations.tsv represents one stimulus interval for one participant. Procedure: After a standard calibration procedure, participants viewed all 16 clips (8 pairs of clips). To mitigate order effects, the clips were presented in two different counterbalanced timelines. The dataset includes the metric files from Tobii Pro Lab, from which the metrics of total fixation count per trial (per participant, per clip version) and fixation duration were extracted. 3. Methodology and Techniques (Qualitative Data): Procedure: Immediately following the eye-tracking session, each participant took part in a semi-structured interview. Interviews were audio-recorded. They began with open-ended questions about the viewing experience and became more specific, probing for awareness of visual flaws or differences. An emergent follow-up question concerning "authenticity" was added for participants who spontaneously mentioned the topic. 4. Ethical Considerations: All participants were volunteers. Upon arrival at the laboratory, they were fully briefed on the experimental procedure and subsequently provided written informed consent prior to their participation in the study. Contents Data files: Per-participant demographics and data availability overview; qualitative codebook with code definitions, examples, and usage counts; eye-tracking metrics exported from Tobii Pro Lab (whole fixations, per participant per stimulus); descriptive statistics and linear mixed-effects model results in CSV format.Analysis script: R script for all quantitative analyses, including generalised and log-normal linear mixed-effects models (lme4, lmerTest), estimated marginal means (emmeans), post-hoc power analysis (simr), descriptive tables, and figures. Interview materials: English translation of the semi-structured interview guide.Figures and tables: All figures and tables from the manuscript in full resolution. Data availability notes Eye-tracking data are available for 17 of 18 recruited participants (P2 data unavailable). Interview data are available for 14 participants; audio recordings for three participants were lost due to technical failure.