I have a
Why is this happening and how can I fix it?
When a PDF is generated artwork can be adjusted, expanded, and/or flattened to accommodate the PDF job options.
In many cases, raster images get divided into sections based upon how they interact with other objects on page. This can also happen with vector content if there are transparency interactions within the objects. For example, if you lay one vector rectangle partially over another and then change the blending mode of the top rectangle, you’ve generated some transparency interaction. Then when saving that art as a PDF/X-1a file for press, the transparency will be flattened and the objects will be divided to maintain their appearance.
When saving a native file in Illustrator there’s an option to Create PDF Compatible file. What this does is actually generate two file formats in a single file — the native .ai data and then the .pdf data.
When you save a PDF from Illustrator and check the Preserve Illustrator Editing Capabilities you do the same two formats in one file thing but in reverse – you save a PDF and you embed the .ai file as well.
Photoshop does the same thing with the Preserve Photoshop Editing Capabilities option when saving as a PDF. If that is checked, two formats are embedded in the PDF — .pdf and .psd. If that is not checked the .psd data is dumped and only the .pdf data is saved.
If a PDF is saved without the “Preserve Editing Capabilities” selected, the native file format information is not contained in the PDF and there’s no way to pull that information out of the PDF.
Once objects art split due to flattening, combining them may be possible, but really it all depends upon the artwork itself.
In many cases, for raster images, you can piece things back together in Photoshop. Or just open the PDF in Photoshop and crop to the image you need.
For vector content, reconstruction greatly depends upon how the objects were divided and what other objects they interact with. There’s really no blanket tutorial or answer on how to combine objects which have been split due to flattening. There are simply too many variables to be definitive.