This ticket needs a bit more information:
Are we talking about the feed page, or a specific entity type (e.g. blog)? Is it related to open graph and/or the new structured data format? Or both?
I'm assuming you're talking about longform posts which attempt to extract a preview from images within the text, and I suspect (although I've not looked at the code) will fall back to owner if not found.
Is this what you're talking about here?