Web Image Formats Research

Compare AVIF with 4:4:4 and 4:2:0 chroma subsampling at the same quality levels using the DIV2K validation set. · 100 images · 1400 measurements · AVIF

Study how cropping (reducing the image area while keeping the analysis fragment at full resolution) affects AVIF compression quality. Unlike the resolution-impact study which downscales the entire image, this study crops the image to different sizes around a fixed 200×200 analysis fragment, eliminating the confounding effect of reduced resolution on quality metrics. Quality metrics are measured only on the analysis fragment, ensuring consistent comparison across crop levels. · 100 images · 2475 measurements · AVIF

Study how AVIF encoding speed affects quality and file size. Tests speed settings from 0 (slowest/best) to 10 (fastest) at representative quality levels to find the optimal speed/quality tradeoff. · 51 images · 2244 measurements · AVIF

Compare all four formats (JPEG, WebP, AVIF, JPEG XL) at the same quality levels using the DIV2K validation set. · 86 images · 4558 measurements · AVIF, JPEG, JXL, WEBP

Study how cropping (reducing the image area while keeping the analysis fragment at full resolution) affects JPEG compression quality. Unlike the resolution-impact study which downscales the entire image, this study crops the image to different sizes around a fixed 200×200 analysis fragment, eliminating the confounding effect of reduced resolution on quality metrics. Quality metrics are measured only on the analysis fragment, ensuring consistent comparison across crop levels. · 100 images · 3465 measurements · JPEG

Study how cropping (reducing the image area while keeping the analysis fragment at full resolution) affects Jpeg XL compression quality. Unlike the resolution-impact study which downscales the entire image, this study crops the image to different sizes around a fixed 200×200 analysis fragment, eliminating the confounding effect of reduced resolution on quality metrics. Quality metrics are measured only on the analysis fragment, ensuring consistent comparison across crop levels. · 100 images · 2970 measurements · JXL

Study how JPEG XL encoding effort affects quality and file size. Tests effort settings from 1 (fastest) to 10 (slowest/best) at representative quality levels to find the optimal effort/quality tradeoff. · 100 images · 6000 measurements · JXL

Study how input resolution affects compression efficiency across different image formats (JPEG, WebP, AVIF, JPEG XL). Tests the same images at multiple downscaled resolutions to understand resolution-dependent performance characteristics. Quality settings are chosen to target around 80 SSIMULACRA2 score for each format at the original resolution. · 100 images · 3600 measurements · AVIF, JPEG, JXL, WEBP

Study how cropping (reducing the image area while keeping the analysis fragment at full resolution) affects WebP compression quality. Unlike the resolution-impact study which downscales the entire image, this study crops the image to different sizes around a fixed 200×200 analysis fragment, eliminating the confounding effect of reduced resolution on quality metrics. Quality metrics are measured only on the analysis fragment, ensuring consistent comparison across crop levels. · 100 images · 3960 measurements · WEBP

Study how WebP encoding method affects quality and file size. Tests method settings from 0 (fastest) to 6 (slowest/best) at representative quality levels to find the optimal method/quality tradeoff. · 100 images · 5600 measurements · WEBP