Сегментація та колірна адаптація растрових зображень засобами алгоритму sky replacement у adobe photoshop

Abstract

This paper analyzes the effectiveness of AI-based Sky Replacement algorithms within the context of prepress preparation. The study examines the mechanisms of semantic segmentation and the technical parameters of "Edge Lighting" and "Color Adjustment" functions, which ensure the harmonious integration of foreground objects. Through experimental testing on complex subjects, such as fine foliage and glass structures, both the time-saving advantages and the current limitations of the technology were identified, spec ifically regarding reflections and perspective logic. The results demonstrate that while manual refinement remains necessary for mirrored surfaces, the automation of masking processes reduces total processing time from hours to minutes.