Optimal output settings when generating images for use in PowerPoint

October 15, 2007 5:22 AM

What are the optimal output settings when generating images for use in PowerPoint?

I am producing plots for use in PowerPoint in R. I tried using Windows Metafile output but it didn't work if I examined the file on a Mac. It looks like I am stuck with a raster image format, so I'm using PNG. Various FAQ lists I have found suggest using 1024×768 resolution.

R will produce its own titles so I don't want to mess with PowerPoint's slide templates. They also have very generous margins, but I want to use as much of the projected area as possible for my plots. Clearly I have to stop somewhere or I will be in danger of the edges being cut off, and since they have the slide title and axis labels, I'd rather avoid that.

So what's the minimum margin space I can use without having to worry that somewhere the stuff inside the margins will be clipped? A suggestion measured in pixels would be most helpful.

II previously asked this question on Microsoft's PowerPoint newsgroup, but no one felt able to give me a minimum number. I'm hoping that there will be some suggestions here with the wealth of PowerPoint experience at MetaFilter.

Please don't suggest using another software package instead of PowerPoint. I am well aware of the alternatives.

posted by grouse to computers & internet (8 comments total)

The answer depends highly upon the resolution of the display device. Typically, your optimal output resolution will be equal to the resolution of the display device.
posted by pmbuko at 6:16 AM on October 15, 2007

Unfortunately I won't always know what the resolution of the display device is, as it will be projectors in various places around the world. So I have to guess, and hopefully for something useful everywhere.
posted by grouse at 6:23 AM on October 15, 2007

Just to give you a solid answer, I find that 1600x1200 is the perfect PowerPoint-making resolution. I resize photos to that resolution and try to output graphs and figures to that resolution... This allows me to crop useless borders with room to spare, and it also allows a bit of photo cropping to highlight a feature. As for the "minimum margin space," I often go all the way to the edge, but a good practice would be to save yourself 5% margins (~50 pixels at 1024, ~80 at 1600), just for a bit of buffer.

As fr sizes, of course it depends on the resolution of the display device. Here is my rule of thumb, as an academic:

1) old (4+year) classroom, probably 1024x768
2) department portable projector, probably 1280x1024
3) nice lab projector bought on a grant, probably 1600x1200

For all its downsides, PowerPoint does do a halfway decent job of scaling imagery on the fly. If you find any graniness in your display and you want to readjust the images or graphs before inputting them into PP, you should use gimp/photoshop/irfanview with a cubic interpolator to resize (usually the slowest method, as it recalculates each pixel instead of just cutting out rows and columns). Small text renders poorly when resized by PP, so err on a side of caution (bold helps).

These are my "best practices," but I'd be more than happy to hear others' tips. I'm also a big fan of outputting the PP format to PDF through OpenOffice, for distribution online, and "just in case"
posted by zachxman at 7:13 AM on October 15, 2007 [2 favorites has favorites]

zachxman: that is exactly the kind of answer I'm looking for. If anyone else wants to concur or disagree, that would still be helpful.
posted by grouse at 7:21 AM on October 15, 2007

1) old (4+year) classroom, probably 1024x768
2) department portable projector, probably 1280x1024
3) nice lab projector bought on a grant, probably 1600x1200

Unless your lab is incredibly wealthy, most projectors being purchased will top out at 1280x1024, if you're lucky.

Nearly all the lab PIs I know buy 1024x768 projectors, because they are in the $1000 "sweet-spot" and work just fine with their laptops and presentations.

I'd create your presentations for 1024x768, because if you don't know where you're presenting, assume the lowest common denominator.
posted by Blazecock Pileon at 12:13 PM on October 15, 2007

Blazecock Pileon: don't you think there is still some wisdom in outputting 1600×1200 which can always be scaled down? Some places may even have 800×600 projectors. If some crucial point were missed at that resolution then that probably means I'm designing things too small to be read anyway.
posted by grouse at 12:54 PM on October 15, 2007

Blazecock Pileon: don't you think there is still some wisdom in outputting 1600×1200 which can always be scaled down?

Looking past how well (or poorly) bitmap images scale and positioning of text elements between resolutions, another issue is how large you want your presentations to get.

Higher resolutions scale up file sizes by a square factor, especially if you plan to keep images nice and sharp at that resolution. How do you plan to distribute these files? It might make sense to create a couple versions of a presentation and see if that matters.

Perhaps you won't get much consensus on what projectors people have. The closer to HD-scale resolutions you get in a projector, the price ramps up quickly.

From my own experience in a biology department, I will say the number of 1600x1200 projectors I've come across are few and far between, and the mean is much closer to 1024x768 than 1280x1024. Perhaps I'm just far behind, or the labs want to spend their funds on research...
posted by Blazecock Pileon at 2:39 PM on October 15, 2007

Try ungrouping the pasted windows metafiles. Ignore the warning dialog and notice that this ungrouping is simply a "translation"of sorts. The resulting object will be a collection of still-grouped objects that can be ungrouped by ungrouping a second time. But you will likely not need to take this second step.

This translation might or might not produce good results. But if the source is vector-based, it is always worth giving it a chance before switching to bitmaps.
posted by magullo at 2:57 PM on October 15, 2007