A look at OpenOffice Impress's PPTX import filter

(Update: Feb 8, 2013 to add LibreOffice 4.0.)

Additional, more "unit-test-like" tests have been posted.

I've made the comment (mostly on Slashdot) that the PPTX import filter in OpenOffice is "barely alpha quality at best." It is improving but it's still got a very long way to go. At this point, for a lot of PPTX presentations it's not even good enough to follow along with and see what was on the original slides, let alone actually present from it or use to share the same file back and forth.

The following presentation is my evidence. In Spring 2009 I taught undergraduate compilers; these slides are what I used during the first day of class. You can download the original presentation as PPTX or PPT. (The latter was created by saving the presentation from within PowerPoint 2007 as the older format. Note that this does import just fine into Impress, and thus converting to PPT creates a file that's fine for distributing to people who need to read it. However, round-tripping the file basically can't be done, as PowerPoint appears to rasterize any of the graphics that can't be represented in the older format, resulting in something that you basically can't edit.)

I took screenshots of two representative slides. You can download the presentation yourself and take a look to see that I'm not cheating; the ones with graphics really are all that bad.

Slide 2Slide 13
LibreOffice Impress 4.0

(dramatic improvement
since I started logging,
but still a long way to go)
Impress 3.2
Impress 3.1
Impress 3.0

Note that even the "easy" slide isn't correct: the latest version of Impress doesn't color the background. (As of some version after 3.2 and by 4.0, the only complaint I have about the easy slide is the aliased text.) Before that, the date showed up in the bottom left corner of each slide. (Yes, those dates show up while actually presenting.) Before that, there were a bunch of bounding boxes. (I forget if they show up during the presentation mode, but I think they did.) Meanwhile the harder slide (which uses some of the fancy new 3Dish effects introduced in PowerPoint 2007) isn't even remotly usable; this is still true in 4.0

(Sorry for the awful production quality of this page. I didn't even really attempt to get anything that is real HTML, I just used height and width attributes of the img tag to get the thumbnails above which wastes bandwidth, the file names are horribly inconsistent, the screen shots are of different sizes, etc.)