[itimpi]

Conversion from PDF files is always a bit hit-and-miss although the new Calibre conversion engine (due in the not too distant future) is meant to improve things. Unfortunately that means there is no generic answer as to the best way to handle PDF input.

I am afraid that the only real answer is to try and avoid PDF as the source format. Almost any other format is better as a starting point! I would normally be starting with LIT or HTML formats which is probably why i am not seeing the sort of issues you describe. Selecting a different output profile in Calibre is not going to help as the problem is in the source material.

It is possible to load the generated ePub file into an ePub editor such as Sigil to see if the results can be improved. However even if it does you may be reluctant to invest the effort that would require.