Here's a simple structured content definition with an image and a html editor :

<form description="Article">

	<image id="image" max-width="300" max-height="300" label="Image" required="false">
		<alt-field>
			<field id="alt-text" required="false" label="Légende photo">
				<datatype base="string"/>
			</field>
		</alt-field>
	</image>
		
	<htmleditor id="content" label="Contenu" required="true">
		<width>500</width>
		<height>300</height>
		<file-configuration>complete_configuration.xml</file-configuration>
		<cut-paste/>
		<link-support/>
		<image-support/>
		<alignment/>
	</htmleditor>

</form>

The first time we save the content (if an image was submited), a new additional data is created for the image.

If the content is edited again and if a new version is created (the html editor content must be modified) without changing the image, it will create a new additional data pointing to the same image.
And every time the content is modified, a new additional data is created

So if you have a content with one image, you'll have one element in it's data for the first version, two for the second version, and so on.