After discussion through private mail, here is the solution.
As always, document structure must be as simple as possible for reliability, stability and performance. Here the case is complicated by the fact that the document must be translated to foreign format, exacerbating the need for a simple structure so that the converter understands it.
The starting state is a captioned image created by menu Insert
>Caption
. This has created a frame containing a paragraph, styled Figure with auto-numbering, anchoring the picture frame. The composite frame is itself anchored to a paragraph which resides in a table cell. This double nesting makes tuning the exact position of the frame and the image very difficult and unstable (a slight change in cell text may easily flush the captioned frame on next page or into the bottom margin).
In the document, the caption and the image are intended to appear sequentially in the main text, not as a side illustration. Consequently, as is frequently the case, Insert
>Caption
is not the good solution. It is simpler to enter a Figure paragraph followed by an Inserted Image paragraph whose sole purpose is to be host the image. The paragraph styles are customised as follows:
- built-in Figure: Next Style = Inserted Image, Keep with next paragraph enabled
- custom Inserted Image, alignment = center
The image is anchored As character. Thus positioning can be tuned with paragraph indents and spacing which causes less problems than frames. Caption is manually entered which the use of a field for the Figure number range.
To provide the same “comfort” as Insert
>Caption
, the whole sequence with “Figure xxx: <void caption> Enter <empty “Inserted Image” paragraph> Enter” can be stored in an AutoText entry. All is left is then to paste the image, anchor is As character and type the real caption.
If the caption is to be positioned after the image, style Inserted Image is to be flagged Keep with next paragraph with Next Style set to Figure while Figure is not.