When testing different campaign/angles ect intially, how many images will you test to decide whether the campaign is worth going on with?
There is no set rule mate, you have to feel it out. I would personally test 5 different ads per angle to see if there is any potential. Then work with the winner to see if you can make the campaign a winner.
To echo Mr Green, what you're looking for at the start is to understand what an audience across a specific traffic source is going to respond to relative to the offer you're promoting. Sometimes it's a game of playing change rooms until you find whats going to work across the source. This differs wildly source to source.
Keep your variables flat to begin with, focus on finding the strongest angle then work from there. Keep your head in check whilst you're working through discovery - you're hunting for Bambi not running from a bear. Get a rep on the traffic source when you start to get angle/offer signals.
If you're on a new source and you don't yet understand how to tickle it in all the right areas, I like to swing high and low on bids which tends to give me a more accurate indicator on the strength of the angle. I.e. you might access higher quality inventory on higher bids etc etc. Again this differs between sources.
it might be worthwhile digging around on the forums for opinions on testing strategies as well. But just make sure you take action on it and don't fall into the trap of analysis paralysis.
Good luck man!
Use your golden images. Every top affiliate has those 5-20 images that are approval-sure, great CTR, and work with basically all demos.