It should be possible to allow those publishing stories or comments to control where any associated images appear in the text by making use of wiki type tags.
For example if a story is published with same 3 images, then by placing tags like [[image1]] and [[image2]] in the text, then just before Oscailt displays a story or comment it could search for these tags and replace them with the html img src tag. It would need to build a list of these tags found and a list of images found for the given article.
It could also be a lot of work for nothing. Also once published, the publisher would as normal lose control again.