Without having more detail as to how you are using these, it's hard to be too specific. Are you showing all images in a random order or are you looking to only choose one of the 9 images on each survey randomly? If you are just choosing one randomly but are looking to get an equal representation in your study then you may consider setting a quota question with the least fill setting being used. Create 9 quota items (1 for each image) and set each to the same total quota number and set it to use least fill. Then you can show the correct image based on which quota number that respondent fell into. You could return the image using Unverified Perl, or you could create a list where each image is a list item. You could then create a constructed list and add the list member based on the quota selected. Then everywhere you want the image to be shown just return the listlabel for item one of your constructed list.
