Disclosed is a method and system for checking and identifying images and other document components that may not be suitably rendered by an output device or process. The method employs a dual-mode analysis of visual image attributes to determine an aggregate confidence while also checking individual image attributes, to determine whether image content requires adjustment prior to being printed or rendered.