The number of connections and/or requests needed to obtain or provide image content in an electronic environment can be reduced by dynamically composing the content when needed to serve a request. In a case where a large number of images are to be transmitted across a network, the images can be determined at the time of the request for transmission and can be composed into a single, master image. Positional information for the images can be stored and sent with the master image to a device requesting the transmission. Using the positional information with dimensional information for the images, the device is able to use an application such as cascading style sheets (CSS) sprites to extract each of the plurality of images, thus obtaining all the images using a single request and single connection. Multiple master images can be used if appropriate.