We propose a new methodology for large-scale subjective quality assessment of compressed still images in the high fidelity range. Combining two different assessment protocols, one based on pairwise comparisons, the other on absolute opinions, it is designed to assess this range of qualities not well-covered by previous methodologies. The methodology was applied to create the Cloudinary Image Dataset ’22 (CID22), consisting of 22,153 annotated images (with scores based on 1.4 million opinions), originating from 250 pristine images compressed using JPEG, JPEG 2000, JPEG XL, HEIC, WebP, and AVIF at high fidelity settings. Using this data, we evaluate various image encoders and objective metrics.