PDF → PowerPoint — Images: shape or background

An image on a PowerPoint slide can be a shape (one object in the z-order alongside text and graphics) or the slide background (a property of the slide itself, drawn beneath everything and outside the normal selection model). The two roles look identical when rendered. They behave differently the moment the user touches the slide.

The two models, side by side

Image as a shape

A picture-type shape with coordinates, dimensions, z-order, cropping, and effects. It can be moved, resized, deleted, framed, rotated, and edited through the normal Format pane. In the slide XML:

<p:pic>
  <p:nvPicPr>...</p:nvPicPr>
  <p:blipFill>...</p:blipFill>  <!-- reference to /ppt/media/ -->
  <p:spPr>...position, size...</p:spPr>
</p:pic>

Image as a background

Not a shape. A property of the slide, declared at the <p:cSld> level. Drawn beneath everything else, fills the whole slide, cannot be selected by clicking, edited only through the slide-background dialog:

<p:bg>
  <p:bgPr>
    <a:blipFill><a:blip r:embed="rId1"/></a:blipFill>
  </p:bgPr>
</p:bg>

All other shapes draw on top.

Deciding which role an image belongs in

The standard heuristic is coverage area combined with z-order position:

• An image covering more than 70% of the slide and sitting at the bottom of the z-order (drawn first) is a background.
• Anything that covers less than 70%, or sits above other objects, is a shape.

A few cases need finer judgment. Watermarks (semi-transparent images at full slide size) usually want to be backgrounds, but if they sit on top of everything in z-order they are overlays, which PowerPoint handles through a different code path. A large image beneath other images is a background. A large image positioned between blocks of text is a content image, even if it crosses the 70% threshold.

What you gain by getting it right

Background classification is theme-aware. A user who switches themes keeps the custom background while only fonts and placeholder colors change. The background stays out of outline view and out of the editor’s selection model. It can be turned off at print time via “Don’t print background pictures.”

Misclassified as a shape, the image fights the theme: the new theme paints its own background on top, then the original image covers that background, and the result is a slide that doesn’t match anything else in the deck.

Most converters skip the distinction. Every image becomes a <p:pic>. The slide looks correct, but the theme system has no idea any image was ever a background. Better converters apply the coverage heuristic and promote qualifying images to <p:bg>.

Decorative graphics

Frames, borders, color bars, dividers. In a PDF these come through as vector primitives (rectangles, lines) sitting beneath the content at fixed positions, repeating from page to page.

Their right home is the slide master, where elements appear automatically on every slide. A converter that detects the repetition and promotes them produces a deck where editing one decoration changes every slide at once.

Real converters almost never do this. Every decoration gets copied onto every slide as an individual shape. A 30-slide deck with a corner ornament has 30 copies, and changing one does nothing to the other 29.

Format and quality

The image-format rules are the same as on the Word side (see Article 9 of the Word series):

• JPEG → JPEG. Direct copy.
• PNG → PNG. Direct copy.
• JBIG2 → PNG. Decoded, because PowerPoint has no JBIG2 support.
• JPEG 2000 → PNG or JPEG. Decoded and re-encoded, because PowerPoint has no JP2K support.

DPI targets depend on output:

• Projector display: 96–100 DPI.
• Printed slides: 200+ DPI.
• Reference target for 16:9 slides: about 1920×1080, which works out to roughly 144 DPI on a 13.333” slide.

A converter that exposes downsampling settings keeps the output file from ballooning into hundreds of megabytes when the source PDF embeds high-resolution images everywhere.

Transparency

PowerPoint supports per-shape transparency through an alpha parameter. PDF can carry transparency in three forms: images with an alpha channel or SMask, objects with a transparent fill via the /CA graphics state, and color-key or stencil masks.

The translation:

• PNG with alpha — direct, no conversion needed.
• SMask (a separate soft-mask image) — extracted and either applied to the source image or folded into the PNG alpha channel.
• Semi-transparent shapes — <a:alpha val="50000"/> in DrawingML, where the value is ST_PositiveFixedPercentage measured in thousandths of a percent (so 50000 = 50%).

Simple transparency round-trips cleanly. Complex transparency groups rarely survive intact.

Video and audio

PDF supports embedded video through RichMedia annotations, animated GIFs, and audio. PowerPoint also supports video and audio. In principle the converter could extract one and embed it as the other; in practice almost none do. If the source PDF had a video, the output PPT has the still poster frame and nothing else.

Speaker notes

Notes live in /ppt/notesSlides/notesSlide1.xml. There is no PDF analog. PowerPoint can export notes as separate PDF pages (the “Notes pages” option), and a determined converter can detect those pages (they look distinctive: text-only, formatted as notes) and route them back into the notes section of each slide. Few tools attempt it.