Showing more than one image, side by side

I am dong an experiment using SuperLab where:
1.In Step 1 , I need to show 20 images,one after another, in the centre of the screen, each being shown for 500 ms followed by a 500ms blank screen.
2.In Step 2, I need to show 60 images (20 X 3), where each event consists of showing 3 images, side by side, where one of the images is from the Step1.
3.The subject has to identify the one he/she has seen earlier.
As of now, I can only do the following: Show a series of images, one at a time.
What I cant figure out:
1.How to show more than one image at one time…is there an easier way to go other than creating a composite image?
2.How to build in - for how long the stimulus should be present and then wait for response, the response triggering the next event.
3.How to randomize presentation of images, given some image files?
Any help/input will be much appreciated.
I do like SUPERLAB till now.

SuperLab does not currently offer a built-in method of limiting the amount of time a stimulus is onscreen. In order to do this, you will need two events, each displayed for 500ms. The first event displays the image, while the second erases the screen and displays nothing (e.g. a text event with no text). If you do want to get a response, then the first event will need a feedback (or two) to end the trial on a response.

The question of how to display more than one image at a time is covered in the help files included with SuperLab, though it does not cover everything you need to know in your situation. When trials are built based on stimulus lists, only one stimulus is retrieved per trial, regardless of the number of events. Therefore, you will probably want to use three stimulus lists–one for each of the three images to appear at the same time. Also, stimulus lists can’t be randomized, so the first item in list 1 will pair with the first item in list 2 and list 3. You can, however, randomize the trials within the block, but the list pairings will not be affected by this.

The difficulty of step 2 comes in how the correct response must be specified. If you are using the mouse as your input device, then this is already taken care of for you. Otherwise, specifying the correct response can be difficult, as it is currently not possible to define a correct response based on trial variables (which you need to use to randomize the position on the screen). If you plan on using something other than the mouse, I can try to figure out another method of specifying the correct response. A simple possibility would be that you could figure out whether a response is correct yourself based on the trial variable value printed in the data file.

I know this is a lot of information spewed out in relatively little detail, but it should help to point you in the right direction. Feel free to ask more questions if I’ve done nothing but confuse you.