Data in an image data is stored as a string of bits and is not interpreted by SQL Server. Any interpretation of the data in an image column must be made by the application. For example, an application could store data in an image column using a BMP, TIFF, GIF, or JPEG format. The application that reads the data from the image column must recognize the format of the data and display it correctly. All an image column does is provide a location to store the stream of bits that make up the image data value.
Yet the best practice is to store the image some where on the disk and save the pointer to it in the database.